Index Page
Reference Frames

Table of Contents


   Reference Frames
      Abstract
         Purpose
         Intended Audience

   Using Frames
      Frame Functions in CSPICE
         Frame Transformation Functions
         Frame Information Functions
      Frames Supported in SPICE
      ICRF vs J2000
      Kernels Needed For Computing Frame Transformations

   Creating a Frame Kernel
      Frame Classes
      Specifying a New Frame
      Guidelines for Frame Specification
         Selecting a Name
         Selecting a Frame ID
         Selecting the Class
         Selecting the Center
         Selecting a Class ID
         Frame IDs Reserved for Public Use
         Why have a Frame ID and a Class ID?
      Putting the Pieces Together
      Connecting an Object to its Body-fixed Frame
      The rest of the frame information

   Inertial Frames

   PCK Frames

   CK Frames
      SCLK and SPK ID codes

   TK Frames
      Defining a TK Frame Using a Matrix
      Defining a TK Frame Using Euler Angles
      Defining a TK Frame Using a SPICE-style Quaternion
      Gaining Flexibility via TK Frames

   Dynamic Frames

   Parameterized Dynamic Frame Families
         Notation
      Required Keywords for Parameterized Dynamic Frames
      Conditional Keywords for Parameterized Dynamic Frames
         Rotation State
         Freeze Epoch

   Two-Vector Frames
      Defining a Two-Vector Frame in a Frame Kernel
         Kernel Availability
         Specifying the Base Frame
         Specifying the Frame Family
         Specifying the Rotation state or Freeze Epoch
         Specifying the Angular Separation Tolerance
         Frame Axis Labels
      Vector Specifications
         Observer-Target Position Vectors
         Target Near point Vectors
         Observer-Target Velocity Vectors
         Constant Vectors

   Mean Equator and Equinox of Date Frames
      Defining a Mean Equator and Equinox of Date Frame in a Frame Kernel
         Specifying the Base Frame
         Specifying the Frame Family
         Specifying the Precession Model
         Specifying a Rotation State or Freeze Epoch

   True Equator and Equinox of Date Frames
      Defining a True Equator and Equinox of Date Frame in a Frame Kernel
         Specifying the Base Frame
         Specifying the Frame Family
         Specifying the Precession Model
         Specifying the Nutation Model
         Specifying a Rotation State or Freeze Epoch

   Mean Ecliptic and Equinox of Date Frames
      Defining a Mean Ecliptic and Equinox of Date Frame in a Frame Kernel
         Specifying the Base Frame
         Specifying the Frame Family
         Specifying the Precession Model
         Specifying the Mean Obliquity Model
         Specifying a Rotation State or Freeze Epoch

   Euler Frames
      Defining an Euler Frame in a Frame Kernel
         Specifying the Base Frame
         Specifying the Frame Family
         Specifying the Epoch
         Specifying the Euler Angles

   Product Frames
      Defining a Product Frame in a Frame Kernel
         Specifying the Base Frame
         Specifying the Frame Family
         Specifying the Factors

   Dynamic Frame Implementation Considerations
      Introduction
      Simulated Recursion
         The Need for Recursion in the CSPICE Frame Subsystem
         Implementation of Limited Simulated Recursion
         Limits on Recursion in Frame Definitions
      Frame Derivative Accuracy
      Degenerate Geometry
      Efficiency Concerns

   Switch Frames
      Specifying Switch Frames
         The Base Frame List
         Time Intervals Associated with Base Frames
         Binary Search
      Switch Frame Connections
      Switch Frame Buffering

   Appendix. ``Built in'' Inertial Reference Frames
         Complete List of ``Built in'' Inertial Reference Frames
         Inertial Reference Frame References
         Low Level Inertial Reference Frame Functions

   Appendix. ``Built in'' PCK-Based IAU Body-Fixed Reference Frames

   Appendix. High Precision Earth Fixed Frames

   Appendix. Frame Identifiers Reserved for Earth Fixed Frames

   Appendix. Frame Definition Examples
      Inertial Frames
      PCK Frames
      CK Frames
      TK frames
         TK frame --- Alias
         TK frame --- Topographic
         TK frame --- Instrument
      Examples of Two-Vector Parameterized Dynamic Frames
         Geocentric Solar Ecliptic (GSE) Frame
         Geocentric Solar Magnetospheric (GSM) Frame
         Mercury Solar Equatorial (MSEQ) Frame
         Example: Nadir Frame for Mars Orbiting Spacecraft
         Example: Roll-Celestial Spacecraft Frame
      Examples of Mean Equator and Equinox of Date Frames
         Earth Mean Equator and Equinox of Date Frames
      Examples of True Equator and Equinox of Date Frames
      Example of a Mean Ecliptic and Equinox of Date Frame
      Example of an Euler Frame
      Examples of Product Frames
         IAU_EARTH Frame, Augmented with Nutation Model
         Dog-Leg Frame for Saturn
      Examples of Switch Frames
         Switch Frame Using Reconstructed and Predict CKs
         Switch Frame Using CK and Dynamic Frames
         Predicted Attitude Profile for Observation Planning




Top

Reference Frames





Last revised on 2021 DEC 31 by N. J. Bachman



Top

Abstract




The frames subsystem specifies the relationships of various kinds of reference frames supported by SPICE. This facilitates ``behind-the-scenes'' transformations between these frames.



Top

Purpose



This document describes how reference frames are treated within SPICE. The document includes a general discussion of reference frames, detailed information about various types of frames supported within SPICE, and instructions on defining additional reference frames to assist in a user's computations.



Top

Intended Audience



This document addresses the needs of several groups of users. Users looking for a basic discussion of reference frames and a list of the frames supported by the SPICE system should read the chapter ``Using Frames.'' Users desiring to customize their environment by adding new frames should read the chapter ``Creating a Frame Kernel.''

This document assumes you have some familiarity with SPICE concepts and terminology. If you are new to the SPICE system, or just a bit rusty with it, you should consider reviewing ``An Overview of the SPICE System'' and ``An Introduction to SPICE.''



Top

Using Frames







Top

Frame Functions in CSPICE




The SPICE frame subsystem facilitates ``behind-the-scenes'' frame transformations. This allows you to concentrate on questions more closely related to the problem you are trying to solve instead of the details of on how to get position or state vectors in the frame of interest.



Top

Frame Transformation Functions



Several user-level CSPICE functions require that the user supply the name of a reference frame as one of the inputs to the function. The most important of these is the function spkezr_c. This function returns the state (Cartesian position and velocity) of one object relative to another in a user specified reference frame. The choice of reference frame often makes a big difference in the usefulness of a returned state. If the state is given relative to the reference frame of interest to the user, computations involving that state can be much simpler than if the state is returned relative to some other reference frame.

The two user-level interface functions that deal solely with frame transformations are sxform_c and pxform_c. sxform_c supports transformations of Cartesian state vectors (6 components) between reference frames while pxform_c supports transformations of Cartesian position vectors (3 components). pxform_c may be used when only position information is needed, or when the derivatives required for a state transformation are unavailable, for example when one frame is defined by a C-kernel that lacks angular velocity data.

The calling sequences for these functions are

      sxform_c ( from, to, et, xform  );
      pxform_c ( from, to, et, rotate );
The output of sxform_c, `xform', is a 6 by 6 matrix used to transform state vectors relative to a reference frame, the name of which is specified by the `from' input argument, to states relative to another reference frame, the name of which is specified by the `to' input argument, at the epoch `et' (specified in seconds past J2000).

The output of pxform_c, `rotate', is a 3 by 3 transformation matrix equivalent to the upper left 3x3 block of `xform'. This matrix transforms position as opposed to state vectors.



Top

Frame Information Functions



The CSPICE frame subsystem contains a set of functions that enable applications to retrieve information about frames known to CSPICE, whether they are built-in or specified by means of frame kernels:

frmnam_c

Convert frame ID code to frame name.
namfrm_c

Convert frame name to frame ID code.
frinfo_c

Return frame specification parameters: frame center, frame class, and frame class ID.
cidfrm_c

Map body ID code to the default frame centered on the specified body. Both frame name and ID are returned.
cnmfrm_c

Map body name to the default frame centered on the specified body. Both frame name and ID are returned.
ccifrm_c

Map frame class and class ID to frame specification parameters: frame ID code, frame name and frame center are returned.
This function provides a way to identify frames referenced in CK and PCK files: in these files, segment descriptors contain class IDs of frames.
The calling sequences for the frame information functions are

   frmnam_c ( frcode, lenout, frname  )
   namfrm_c ( frname, &frcode )
   frinfo_c ( frcode, &cent,  &class,  &clssid, &found )
   cidfrm_c ( cent,   lenout, &frcode, frname,  &found )
   cnmfrm_c ( cname,  lenout, &frcode, frname,  &found )
   ccifrm_c ( class,  clssid, lenout,  &frcode, frname,
              &cent,  &found )
See the section ``Specifying a New Frame'' below for more information on frame specification parameters.



Top

Frames Supported in SPICE




In both cases -- with the functions requiring specification of a reference frame as one of the inputs (for example spkezr_c), and with the functions computing transformation between two reference frames (sxform_c and pxform_c) -- you specify the frame or frames of interest using a character string that contains the name of the reference frame.

In CSPICE function interfaces, frames are typically designated by C strings. In text kernel files, frame names are designated by strings delimited by single quotes, as in FORTRAN. Examples below showing single-quoted frame names exhibit the names as they appear in text kernels; these same names are double-quoted when referred to as literal strings in C source code.

A number of names are automatically recognized by the frame subsystem because the definitions for these frames are ``built into'' CSPICE software. Among these frames are:

    -- inertial frames such as Earth mean equator and equinox of J2000 frame ('J2000'), Mean ecliptic and equinox of J2000 ('ECLIPJ2000'), Galactic System II frame ('GALACTIC'), Mars Mean Equator and IAU vector of J2000 frame ('MARSIAU'), etc. For the complete list of ``built in'' inertial reference frames refer to the appendix ``built in Inertial Reference Frames'' of this document.

    The ICRF is a special case. See the section titled ``ICRF vs J2000'' below.

    -- body-fixed frames based on IAU rotation models provided in text PCK files, such as Earth body-fixed rotating frame ('IAU_EARTH') and Mars body-fixed rotating frame ('IAU_MARS'), and body-fixed frames based on high precision Earth rotation models provided in binary PCK files such as 'ITRF93'. For the complete lists of ``built in'' body-fixed reference frames refer to the appendixes ``built in PCK-Based Reference Frames'' and High Precision Earth Fixed Frames'' of this document.

For all other frames the names are not ``built into'' SPICE. Instead, these names, as well as the parameters specifying the frames, are provided via keywords included in a text kernel file. Text kernel frame definitions cannot overwrite definitions ``built into'' SPICE. The ``built-in'' frames are always accessed first, making text kernel frames with the same names or IDs invisible to the Toolkit.

The types of frames defined in text kernels include:

    -- body-fixed frames based on text or binary PCK data for bodies whose rotational data is not yet included in the IAU rotational constants reports

    -- CK-based frames, i.e. frames for which orientation is provided in CK files

    -- Fixed offset frames, i.e. frames for which orientation is constant with respect to another frame and is specified as part of the frame definition stored in a text kernel. Fixed offset frames are also called TK frames.

    -- Dynamic frames, i.e. frames for which orientation is based on dynamic directions computed based on SPICE kernel data (SPKs, CK, PCKs), on mathematical models implemented in CSPICE functions, or on formulas defined in frame kernels.

    -- Switch frames, i.e. frames that choose at run time other frames with which to align their orientation. Switch frames ``switch'' the base frames they align with as a function of time, using a prioritized list of base frames and optional, associated time bounds; this list is provided as part of the switch frame definition stored in a text kernel.

You can find the names of these frames by examining the text kernel file that contains the frame definitions. Normally definitions of all frames specific for a given mission are stored in that mission's Frames Kernel (FK) file but they can also be provided in the Instrument Kernels (IK) or any other text kernels. In order to make frame definitions from the text kernels available to SPICE, these kernels need to be loaded via a call to furnsh_c. For example, to load an FK named ``myframe.tf'', call furnsh_c as follows:

      furnsh_c ( "myframe.tf" );


Top

ICRF vs J2000




The International Celestial Reference System (ICRS) defines coordinate axes that are closely aligned with those of the J2000 (aka EME2000) reference frame. The International Celestial Reference Frame (ICRF) and later versions of it (ICRF1, etc.) are realizations of the ICRS. For brevity, we'll simply refer to ``the ICRF'' below.

The rotational offset between the J2000 frame and the ICRS has magnitude of under 0.1 arcseconds.

Certain JPL data products are referenced to the ICRF or later versions of it. These include, but are not limited to,

    -- DE4xx series planetary ephemerides

    -- Satellite ephemerides compatible with DE4xx planetary ephemerides

    -- Small body ephemerides compatible with DE4xx planetary ephemerides

    -- Orientation of the terrestrial frame ITRF93

    -- Orientation of the lunar principal axes frame

Rotation models provided by the IAU are referenced to the ICRF.

Modern spacecraft ephemerides and attitude data, other than those for Earth orbiters, are likely referenced to the ICRF. Users should consult documentation or data providers to verify this for data sets of interest.

SPK and binary PCK files produced by NAIF from the data sources listed above are referenced to the same version of the ICRF as the corresponding data sources. For historical and backward compatibility reasons, these data products are labeled as being referenced to the J2000 frame.

No transformation is required to convert state vectors or orientation data from the J2000 frame to the ICRF (or later version), if the vectors or orientation data are computed using SPICE kernels created from the data sources listed above. For example:

    -- A call to spkezr_c with the input frame name J2000 will return a state vector referenced to the ICRF, if the SPK data are from a JPL planetary ephemeris SPK, or from any other SPK in which data are referenced to the ICRF and labeled as referenced to the J2000 frame.

    -- A call to pxform_c with the input ``from'' frame name J2000 and input ``to'' frame name ITRF93 will return a 3x3 matrix that transforms position vectors from the ICRF to the ITRF93 terrestrial frame, if the Earth orientation data are provided by a NAIF high-precision, binary Earth PCK.

    -- A call to pxform_c with the input ``from'' frame name J2000 and input ``to'' frame name IAU_MARS will return a 3x3 matrix that transforms position vectors from the ICRF to the Mars body-fixed, body-centered IAU_MARS frame, if the orientation data are provided by a NAIF generic text PCK.

    -- A call to pxform_c with the input ``from'' frame name J2000 and an input ``to'' CK frame name will return a 3x3 matrix that transforms position vectors from the ICRF to the specified CK frame, if the CK data used by this call are referenced to the ICRF and labeled as referenced to the J2000 frame.

SPICE kernel creators intending to support use of data referenced to the ICRF, as shown above, should write the data without first converting it to the J2000 frame. Segments of such SPK, CK, or binary PCK files should indicate the frame is J2000. It is strongly recommended that kernel creators add comments to the files to explain the actual characteristics of the data.

SPICE users who export kernel data to non-SPICE file formats may need to transform the data, depending on the frame to which the SPICE data are actually referenced (as opposed to the frame to which the kernel indicates the data are referenced), and depending on the desired output frame.



Top

Kernels Needed For Computing Frame Transformations




In many cases data needed to compute transformation of one frame relative to another is stored in SPICE kernels: PCK, CK, FK, and even SPK. The appropriate kernels must be loaded for the SPICE system to compute a frame transformation from a non-inertial frame to any other frame.

The ``built in'' inertial frames are the only frames the transformations between which can be computed without loading any SPICE kernels.

Since the body-fixed frames are tied to the rotation of planets, satellites, asteroids, etc, the information about how the orientation of these frames is changing with respect to inertial frames is stored in SPICE PCK files. It is important to note that although the names of these frames are ``built in'' their relationship to inertial frames is not. This information must be ``loaded'' into the SPICE system from a PCK file. Without loading this information you cannot compute the transformation to or from a body-fixed frame.

As the name suggests, the orientation of CK-based frames is computed using data provided in CK files and cannot be computed without loading these. In addition to the CKs, an SCLK kernel establishing time correlation for the on-board clock that is used to tag data in the CKs must be loaded to support time conversion between that clock and ephemeris time.

Because the fixed offset frame definitions stored in text kernels provide all information needed to determine their orientation relative to the frame with respect to which they are defined, only the text kernel containing the definition need be loaded.

Depending on the particular family to which a dynamic frame belongs, no additional data may be needed in order to compute its orientation, or one or more types of SPICE kernels, including SPKs, PCKs, CKs, and SCLK, may have to be loaded.

Data required to compute orientation of switch frames may be any required to compute orientation of PCK, CK, or TK frames. Data for dynamic and switch base frames are not required because the orientation of a switch frame relative to base frames of those types is the identity. In practice, data sufficient to connect the orientation of a switch frame's base frames to other frames of interest are required by most applications.



Top

Creating a Frame Kernel





To create a frame kernel you will need to understand the SPICE text kernel file format described in detail in the Kernel Required Reading document, kernel.req. When making a new frame kernel, make sure that the first line of the file contains the proper SPICE file identification word for the FK files -- ``KPL/FK'' -- left-justified, on a line by itself.

You will also need to understand the concept of a frame class.



Top

Frame Classes




The method by which a frame is related to some other frame is a function of the ``class'' of the frame. You describe the class of a frame with an integer called the frame's ``class number.'' The reference frame classes are enumerated below.

    1. Inertial frames. These frames do not rotate with respect to the star background. They are the frames in which Newton's laws of motion apply. The class number associated with inertial frames is 1.

    2. PCK (body-fixed) frames. PCK frames are reference frames whose orientation with respect to inertial frames is supplied through either binary or text PCK files. To determine a transformation to or from a PCK frame, you must load a PCK file that describes the orientation of the frame with respect to one of the inertial frames ``built into'' SPICE. The class number associated with PCK frames is 2.

    3. CK frames. CK frames are reference frames whose orientation with respect to some other reference frame is supplied via a SPICE C-kernel. The other reference frame may be any of the four classes of frames described here. C-kernels use spacecraft clock ``ticks'' as their basic time unit. Consequently you need to load a spacecraft clock kernel appropriate for the C-kernel to determine the transformation from or to a C-kernel frame. In addition you will need to load a PCK, CK, or TK frame kernel if the ``other'' frame belongs to one of these classes. The class number associated with CK frames is 3.

    4. Fixed offset frames. These frames are also called Text Kernel (TK) frames because they have a constant orientation with respect to some other reference frame and this orientation is included in the frame definition provided in a SPICE text kernel. They may be defined relative to a frame of any of the other classes of reference frames. The class number associated with TK frames is 4.

    5. Dynamic frames. These are time-dependent reference frames defined via parameters or formulas specified in a frame kernel. The class number associated with dynamic frames is 5.

    6. Switch frames. These are time-dependent frames that choose at run time other frames with which to align their orientation. Switch frames ``switch'' the base frames they align with as a function of time, using a prioritized list of base frames and optional, associated time bounds; this list is provided as part of the switch frame definition stored in a text kernel. The class number associated with switch frames is 6.



Top

Specifying a New Frame




In addition to the data/model needed to specify the orientation of a frame with respect to some other reference frame, you must tell the SPICE system how to find the data or model. This specification requires five pieces of information:

    1. the name of the frame,

    2. the ID code for the frame,

    3. the class number of the frame,

    4. the SPK ID code or name for the frame center,

    5. the internal ID code used by the class (CLASS_ID) to refer to the frame.

The rules for selecting these items are given in the next section, but for the moment let's assume that the rules have been obeyed and we have arrived at the following values.

   Frame Name    :    'WALDO'
   Frame ID code :    1234567   (A number guaranteed to be suitable
                                 for private use)
   Frame Class   :          3   (C-kernel)
   Frame Center  :     -10001   (Waldo Spacecraft ID code)
   Frame Class_id:  -10000001   (ID code in C-kernel for Waldo)
The frame kernel that specifies this frame is given below:

   \begindata
 
      FRAME_WALDO            =  1234567
      FRAME_1234567_NAME     = 'WALDO'
      FRAME_1234567_CLASS    =  3
      FRAME_1234567_CENTER   = -10001
      FRAME_1234567_CLASS_ID = -10000001
 
   \begintext
Note that single quotes are used to delimit strings in SPICE text kernels.



Top

Guidelines for Frame Specification






Top

Selecting a Name



The name chosen for a frame must not exceed 26 characters taken from the set including uppercase letters, numbers, underscore, and plus and minus signs. It should have some mnemonic value so that users can recognize what the name means. Finally, it should not be the name of one of the ``built in'' frames listed above or the name of any other frame you wish to specify. If you try to use a ``built in'' name, the frame subsystem will ignore your frame specification. In the example given above, we chose the name 'WALDO' for the name of our reference frame. If ``Waldo'' would be a lander and would need to specify a local level frame at its landing site, we could have named that frame 'WALDO_LOCAL_LEVEL'. A good name for a frame associated with the camera flown on ``Waldo'' would be 'WALDO_CAMERA'.



Top

Selecting a Frame ID



What you choose for a frame ID depends upon the class of the frame.

If the class is CK, you may use the same ID as you use for the CLASS_ID. In the previous example, we selected the Frame ID to be 1234567. (Since our example frame above is of class 3, a CK frame, we would normally use the same number for the frame ID as we used for the class ID. However, in this example, we have chosen a different value to illustrate the connection between the frame ID and the variables needed to define the frame.)

For TK frames, the frame and class IDs must be identical. For TK frames associated with an instrument, the instrument ID is used for both frame ID and class ID. For topocentric TK frames at tracking station sites, both frame ID and class ID are created by ``combining'' the ID of the body on which the station is located with the station number (for example frame and class ID 1399012 is used for ``DSS-12'', with the formula used to arrive at this ID being 1000000 + ``Earth ID''*1000 + ``station ID''.) For local level and surface fixed TK frames at a landing site, both frame ID and class ID are based on the ID of the lander (for example frame and class ID of -222999 would be the natural choice for the lander with ID -222.)

If the frame is a PCK frame or a dynamic frame and you are working without consultation with NAIF, select an integer in the range from 1400000 to 2000000.



Top

Selecting the Class



This is usually the easiest part of specifying a frame. Presumably you know how the orientation of the frame with respect to some other frame will be computed. Simply choose the appropriate class number. In the example above, the class number is 3 because we are defining a CK-based frame.



Top

Selecting the Center



A frame is used to specify the orientation of some object. The frame consists of a set of coordinate axes relative to some point -- the origin of the reference frame. When viewed from some other frame the axes rotate about the origin. The origin about which the rotation takes place is the center of the frame. For body-fixed frames this is the center of the body to which they are fixed. For C-kernel frames the center is often the spacecraft whose orientation is provided by the C-kernel. Simply find the SPK ID code or name for the object to which the frame is attached and use that as the value for the center. In our example, the SPK ID code for the ``Waldo'' spacecraft is -10001.

Note that this center ID is used to look up the position of the frame origin when SPICE computes frame orientation adjusted for light time. Therefore, only centers for which supporting SPK data are expected to be available should be picked. It is usually an issue only for TK and CK frames associated with instruments because the positions of instruments are rarely available in SPKs. To get around the need to provide the instrument positions, it is appropriate to specify the ID of the spacecraft on which an instrument is mounted as the center of a TK or CK frame associated with it.



Top

Selecting a Class ID



A frame's ``CLASS_ID'' is an integer used internally by CSPICE software. It is the integer code used by the CSPICE reference frame subsystem to look up reference frame information.

If your frame is a PCK class frame the CLASS_ID is the ID code for the body for which rotation constants are provided in the text PCK file or the ID associated with the orientation data provided in the binary PCK file.

If your frame is a CK class frame, the CLASS_ID is the ID code used in the C-kernel to describe the orientation of the spacecraft.

If the frame is a TK frame, the class ID must match the frame ID.

If the frame is a dynamic frame, the class ID must match the frame ID.

If the frame is a switch frame, it is recommended that the class ID match the frame ID.



Top

Frame IDs Reserved for Public Use



The range 1400000 to 2000000 has been set aside by NAIF as ranges of Frame IDs that can be used freely by SPICE users without fear of conflict with ``officially recognized'' frames. However, if you and a colleague plan to create several such frames, you will need to coordinate your work to ensure that your definitions are not in conflict with one another.



Top

Why have a Frame ID and a Class ID?



When the CSPICE software receives a request to compute a frame transformation, it first translates the name of the frame to the corresponding frame ID. There is a one to one correspondence between frame names and frame IDs. Once the frame ID is in hand, the class of the frame can be located and an appropriate subsystem identified for carrying out the initial computations needed to construct a frame transformation matrix. However, the frame subsystem evolved to unify several distinct reference frame systems. In each of these systems, reference frames are identified by integer codes. Unfortunately, since these subsystems evolved independently, the numeric codes used to identify the reference systems overlapped from one system to the next. Moreover, to support backward compatibility, NAIF was not free to change the numeric codes used by the various systems or the meaning of the frame codes that were already present in existing data products.

To support existing data products and allow extension of the SPICE system, NAIF needed to associate the old ID code with the new frame ID. The CLASS_ID fills this role. When the frame is identified, the ID code suitable for the frame class is located and passed onto the frame's class so that the initial portion of the frame transformation can be carried out.



Top

Putting the Pieces Together




Once you've determined the name, ID code, center, class and class ID of your frame, you create the frame specification by filling in the frame template below. This should be stored in a text kernel called a Frame Specification Kernel or Frames Kernel (FK).

   FRAME_<name>             = <ID code>
   FRAME_<ID code>_NAME     = '<name>'
   FRAME_<ID code>_CLASS    = <class>
   FRAME_<ID code>_CLASS_ID = <classid>
   FRAME_<ID code>_CENTER   = <center>
The example we used for the frame 'WALDO' illustrates this.

   \begindata
 
      FRAME_WALDO            =  1234567
      FRAME_1234567_NAME     = 'WALDO'
      FRAME_1234567_CLASS    =  3
      FRAME_1234567_CENTER   = -10001
      FRAME_1234567_CLASS_ID = -10000001
 
   \begintext
Once you've completed the frame specification you tell the SPICE system about the frame by ``loading'' the frame kernel that contains it. As with all text kernels, you load it via the routine furnsh_c. For example if the frame kernel containing your frame specification is contained in the file ``myframe.tf'' you load the kernel via the call

      furnsh_c ( "myframe.tf" );


Top

Connecting an Object to its Body-fixed Frame




Every extended object has both a position and orientation in space. The SPICE ephemeris subsystem (SPK) allows you to specify the location of such an object. The frame subsystem allows you to name the body-fixed frame that describes the orientation of the object, and to retrieve the orientation of the frame relative to some other frame as a function of time. Given the name or SPK ID code associated with an object we can locate its position through the SPK subsystem. Unfortunately, the body-fixed frame of the object cannot always be determined from the object's name or ID code. For example, we have already mentioned that there are two ``built in'' reference frames that describe the orientation of the Earth: 'IAU_EARTH' and 'ITRF93'. For other objects, such as the asteroid Simbad, there is no ``built in'' frame associated with the object. The body-fixed frame of Simbad must be defined through a text kernel. In both cases, the connection between the object and its body-fixed frame needs to be supplied via a kernel pool variable. There are two ways to do this.

   OBJECT_<name or spk_id>_FRAME =  '<frame name>'
or

   OBJECT_<name or spk_id>_FRAME =  <frame ID code>
You may use the ID codes for either the object, the frame or both. As example, four of the following assignments could serve to connect the Earth with the 'ITRF93' frame.

   OBJECT_399_FRAME   =  13000
   OBJECT_399_FRAME   = 'ITRF93'
   OBJECT_EARTH_FRAME =  13000
   OBJECT_EARTH_FRAME = 'ITRF93'
Note: if you use the name of either the object or frame, you must use upper case letters.

Of these four means of specifying an object's body-fixed frame the second (OBJECT_399_FRAME = 'ITRF93') is the most robust.

For the sun, the planets and their satellites the frame subsystem maintains a default connection between the object and its body-fixed frame ``built into'' SPICE. The complete list of ``built in'' body-fixed frames is provided in the ``built in PCK-Based IAU Body-Fixed Reference Frames'' appendix of this document.



Top

The rest of the frame information




The information supplied in the frame specification tells the SPICE system where to look for a particular frame model. However, the specification alone doesn't tell the SPICE system how to actually transform from the specified frame to some other frame of interest. To do this you need to supply other information. How this information is supplied depends upon the class of the frame.



Top

Inertial Frames





Inertial frames are ``built into'' the SPICE system via the routine chgirf_. Only the frames defined in that routine are available as inertial (class 1) frames. It is not possible to override these definitions.

It is possible to create aliases for built-in inertial frames. For example you might define EME2000 as another name for the J2000 frame.

See the appendix containing frame definition examples for information on how to create a frame alias using a TK frame.



Top

PCK Frames





If you specify a PCK frame, you will need to load either a text or binary PCK file for the body with which the frame is associated. The construction of PC kernels is discussed in the SPICE document PCK Required Reading (pck.req.)



Top

CK Frames





If a frame is defined as a CK frame, you will need both a C-kernel for the structure identified by the FRAME_..._CLASS_ID variable and an SCLK kernel for converting ephemeris time to the ``ticks'' used to represent time in the C-kernel. Both the C-kernel(s) and SCLK kernel must be loaded prior to attempting to use the CK frame.



Top

SCLK and SPK ID codes




For many C-kernels, the spacecraft clock and spacecraft ID codes can be determined by performing an integer division of the C-kernel ID code by 1000. However, under some circumstances this numerical correspondence between C-kernel ID code and the associated SCLK or spacecraft ID may break down. When the numerical relationship fails you need to tell the SPICE system the ID code of the SCLK or spacecraft via two kernel pool variables.

   CK_<ck_ID code>_SCLK = <ID code of SCLK>
   CK_<ck_ID code>_SPK  = <SPK ID code>
These variables are normally placed in either the SCLK kernel or in the frame specification kernel (FK).

To illustrate how you would create a C-kernel frame, we shall suppose that we have a C-kernel for structure -100001 aboard the fictional spacecraft ``Waldo'' which has ID code -1001. Moreover we shall assume that the clock ID appropriate for this structure is -1002. Below is a frame specification together with the CK_..._SCLK and CK_..._SPK variable definitions for the 'WALDO' frame.

   \begindata
 
      FRAME_WALDO            = -100001
      FRAME_-100001_NAME     = 'WALDO'
      FRAME_-100001_CLASS    = 3
      FRAME_-100001_CLASS_ID = -100001
      FRAME_-100001_CENTER   = -1001
 
      CK_-100001_SCLK        = -1002
      CK_-100001_SPK         = -1001
 
   \begintext


Top

TK Frames





The relationship between a constant offset Text Kernel (TK) frame and the frame it is offset from is given via a text kernel that can be loaded via the kernel pool routine furnsh_c. The first five kernel pool variables required for TK frame specification are the same as for any other frame defined via a text kernel:

   FRAME_<name>             = <ID code>
   FRAME_<ID code>_NAME     = '<name>'
   FRAME_<ID code>_CLASS    = 4
   FRAME_<ID code>_CLASS_ID = <ID code>
   FRAME_<ID code>_CENTER   = <center>
You need to supply information that indicates the frame, RELATIVE, from which the TK frame is offset. It is done using this kernel pool variable:

   TKFRAME_<frame>_RELATIVE = '<name of relative frame>'
where `frame' is the ID code or name you used in the frame specification.

Because the rotation from the TK frame to the RELATIVE frame is fixed (time invariant) it can be specified in the FK along with the frame specification information described above. This rotation data can be provided in any of three ways:

    1. as a 3 by 3 matrix, M, that converts vectors from the TK frame to the RELATIVE frame by left multiplication

               V_relative = M * V_tkframe
    2. as a set of 3 Euler angles and axes that can be used to produce M

    3. as a SPICE-style quaternion representing M.

You let the frame subsystem know which method you've chosen for representing the rotation via the kernel pool variable

   TKFRAME_<frame>_SPEC.
To use a matrix to define the rotation, use the assignment:

   TKFRAME_<frame>_SPEC = 'MATRIX'
To define the rotation via three Euler angles, use the assignment:

   TKFRAME_<frame>_SPEC = 'ANGLES'
To define the rotation via a SPICE-style quaternion, use the assignment:

   TKFRAME_<frame>_SPEC = 'QUATERNION'
Depending upon the value of the `SPEC' variable, you need to supply one of the following sets of kernel pool variables.



Top

Defining a TK Frame Using a Matrix




If you've chosen to define the rotation using a matrix, supply the matrix using the kernel pool variable assignment below:

   TKFRAME_<frame>_MATRIX = ( matrix_value[0][0],
                              matrix_value[1][0],
                              matrix_value[2][0],
                              matrix_value[0][1],
                              matrix_value[1][1],
                              matrix_value[2][1],
                              matrix_value[0][2],
                              matrix_value[1][2],
                              matrix_value[2][2]  )
For example, if the matrix defining your TK frame is

   0.4   -0.6   0.0
   0.6    0.4   0.0
   0.0    0.0   1.0
and the ID code you've selected for the frame is 1234567, then you would supply the following information in a text kernel.

   TKFRAME_1234567_SPEC   = 'MATRIX'
 
   TKFRAME_1234567_MATRIX = (  0.4
                               0.6
                               0.0
                              -0.6
                               0.4
                               0.0
                               0.0
                               0.0
                               1.0 )


Top

Defining a TK Frame Using Euler Angles




If you've chosen to define a TK frame as a sequence of three Euler angle rotations about specified coordinate axes, you need to supply the following pieces of information:

    1. The values of the three Euler angles;

    2. The axes about which the Euler rotations are performed;

    3. The units associated with the three Euler angles. The recognized units are: 'DEGREES', 'RADIANS', 'ARCSECONDS', 'ARCMINUTES' 'HOURANGLE', 'MINUTEANGLE', 'SECONDANGLE'.

This information is supplied to the SPICE system using the kernel pool variables shown below.

   TKFRAME_<frame>_ANGLES = ( angle_1, angle_2, angle_3 )
   TKFRAME_<frame>_AXES   = ( axis_1,  axis_2,  axis_3  )
   TKFRAME_<frame>_UNITS  = 'units_of_angles'
The units must be from the list given above. The axes must be chosen from the set of integers 1,2,3 where 1 stands for the x-axis, 2 for the y-axis, and 3 for the z-axis. If M is the matrix that converts vectors relative to the TK frame to the RELATIVE frame by left multiplication, then the angles and axes must satisfy the following relationship:

   M = [angle_1]      [angle_2]      [angle_3]
                axis_1         axis_2         axis_3
 
where the symbol

   [ A ]
        i
stands for a rotation by the angle A about the i'th axis.

   +-                     -+
   |   1       0      0    |
   |   0     cos A   sin A |   =  [ A ]
   |   0    -sin A   cos A |           1
   +-                     -+
 
   +-                     -+
   |  cos A    0    -sin A |
   |   0       1      0    |   =  [ A ]
   |  sin A    0     cos A |           2
   +-                     -+
 
   +-                     -+
   |  cos A   sin A   0    |
   | -sin A   cos A   0    |   =  [ A ]
   |   0       0      1    |           3
   +-                     -+
This method of definition is particularly well suited for defining topocentric frames on the surface of the Earth. For example, suppose you have an SPK (ephemeris) file that specifies the location of some surface point on the Earth, and that the SPK ID code of this point is 399100. Moreover suppose you have the geodetic co-latitude (COLAT) and longitude (LONG) measured in degrees for this point. (Note that the co-latitude is the complement of latitude: latitude + co-latitude = 90 degrees.)

Given this information we can easily define a topocentric reference frame at the point such that the x-axis points north along the local meridian, the y-axis points west along the local latitude and the z-axis points up from the reference spheroid.

The transformation from Earth body-fixed frame to topocentric frame is given by

   BF2TP = [180] [COLAT] [LONG]
                3       2      3
Consequently the transformation from the topocentric frame to the body-fixed frame is given by

   M = TP2BF = [-LONG] [-COLAT] [180]
                      3        2     3
Let 1234567 be the ID code for the topocentric frame; let the name of this frame be 'MYTOPO'; and define this relative to the IAU frame for the Earth (one of the ``built in'' frames). The topocentric frame at the ephemeris point 399100 is specified as shown below:

   \begindata
 
      FRAME_MYTOPO             = 1234567
      FRAME_1234567_NAME       = 'MYTOPO'
      FRAME_1234567_CLASS      = 4
      FRAME_1234567_CLASS_ID   = 1234567
      FRAME_1234567_CENTER     = 399100
 
      TKFRAME_1234567_SPEC     = 'ANGLES'
      TKFRAME_1234567_RELATIVE = 'IAU_EARTH'
      TKFRAME_1234567_ANGLES   = ( <-long>, <-colat>, 180 )
      TKFRAME_1234567_AXES     = (       3,        2,   3 )
      TKFRAME_1234567_UNITS    = 'DEGREES'
 
   \begintext
As we'll see a bit later, we can make a more flexible definition for this topocentric frame.



Top

Defining a TK Frame Using a SPICE-style Quaternion




If you've chosen to define a TK frame using a SPICE-style quaternion, supply the quaternion using the kernel pool variable assignment below:

   TKFRAME_<frame>_Q = ( q_0, q_1, q_2, q_3 )
where component zero is the so-called ``real'' component of the quaternion (the ``cosine'' component of the quaternion). The last 3 components (components 1 through 3) are the ``axis'' components of the quaternion -- the i, j, and k components respectively of the quaternion. The quaternion must be a unit quaternion.

        2        2        2        2
   (q_0)  + (q_1)  + (q_2)  + (q_3)  = 1
A more detailed discussion of quaternions is available in the reference document ``Rotations Required Reading'' (rotation.req), and in a ``Quaternions White Paper'' available from NAIF.



Top

Gaining Flexibility via TK Frames




The use of non-inertial frames gives you an easy means of creating ephemerides for points on the surface of a body such as the Earth, Moon or Mars. The ephemeris is simply the body-fixed location of the object relative to a body-fixed frame for the same object. However, the model used to relate the body-fixed frame to other reference frames may not be fixed. Indeed, for the Earth there are several different methods with varying degrees of accuracy that give the orientation of the Earth with respect to inertial space. Each of these different realizations may have a different frame ID code. This ability to ``plug in'' different orientations is one of the strengths of the SPICE system. However, if you create an ephemeris relative to one of these specific models, you won't be able to use it unless you've loaded the correct model. To make the ephemeris usable regardless of the orientation model you happen to have at your disposal, you should define the body-fixed ephemeris relative to a TK frame. Then define the TK frame so that rotation from the TK frame to the PCK frame is the identity matrix. For example, you can define a lunar body-fixed frame as shown below.

   \begindata
 
      FRAME_MOONFIXED          = 3010000
      FRAME_3010000_NAME       = 'MOONFIXED'
      FRAME_3010000_CLASS      = 4
      FRAME_3010000_CLASS_ID   = 3010000
      FRAME_3010000_CENTER     = 301
 
      TKFRAME_3010000_SPEC     = 'MATRIX'
      TKFRAME_3010000_RELATIVE = '<name of base frame>'
      TKFRAME_3010000_MATRIX   = ( 1,
                                   0,
                                   0,
                                   0,
                                   1,
                                   0,
                                   0,
                                   0,
                                   1 )
 
   \begintext
By editing this definition you can make the MOONFIXED frame be the IAU MOON frame or some other model if one is available. Or you can create several such definitions and, at run-time, load the file that best fits your current environment.

Using this indirect method of defining the various frames for which more than one orientation model may be available, you can avoid limiting how various kernels can be used.



Top

Dynamic Frames





In CSPICE documentation, the term ``dynamic frame'' designates a time-dependent reference frame defined via a frame kernel.

A ``parameterized dynamic frame'' is a dynamic frame defined by a formula implemented in CSPICE code and having user-selectable parameters set via a frame kernel. The formula defining a dynamic frame may rely on data from other SPICE kernels, for example state vectors provided by SPK files or rotation matrices from C-kernels or PCK files.

An example of a parameterized dynamic frame is a nadir-pointing reference frame for a spacecraft orbiting a planet, where the spacecraft's nadir direction and velocity vector define the frame. Using a frame kernel, a CSPICE user may specify the planet and spacecraft, the relationship between the nadir and velocity vectors and the frame's axes, and a small set of additional parameters required to define the frame.

Currently parameterized dynamic frames are the only type of dynamic frame supported by CSPICE. Other types of dynamic frames, such as frames defined by complete formulas (as opposed to parameters) provided in frame kernels, may be implemented in future versions of CSPICE.

Below we'll discuss the various types of supported dynamic frames, how to create frame kernels that define dynamic frames, and dynamic frame implementation considerations. The appendix ``Frame Definition Examples'' contains frame definition templates for a variety of popular dynamic frames.



Top

Parameterized Dynamic Frame Families





The ``family'' to which a parameterized dynamic frame belongs indicates the underlying mathematical formula by which the frame is defined. Currently there are six parameterized dynamic frame families:

    -- Two-vector frames: a reference frame is defined by two vectors. The first vector is parallel to one axis of the frame; the component of the second vector orthogonal to the first is parallel to another axis of the frame, and the cross product of the two vectors is parallel to the remaining axis.

    -- Mean equator and equinox of date frames: these use mathematical precession models to define orientation of a body's equatorial plane and location of the frame's x-axis. Currently these frames are supported only for the earth.

    -- True equator and equinox of date frames: these use mathematical precession and nutation models to define orientation of a body's equatorial plane and location of the frame's x-axis. Currently these frames are supported only for the earth.

    -- Mean ecliptic and equinox of date frames: these use mathematical precession and mean obliquity models to define orientation of a body's orbital plane and location of the frame's x-axis. Currently these frames are supported only for the earth.

    -- Euler frames: polynomial coefficients, a reference epoch, and an axis sequence are used to specify time-dependent Euler angles giving the orientation of the frame relative to a second, specified frame as a function of time.

    -- Product frames: these define the orientation of a frame relative to a base frame as the product of a specified sequence of frame transformations. All of the factor transformations must be computable by CSPICE.



Top

Notation



A lower case letter `x' is used to designate the cross product operator, as in

   C = A x B
Double vertical bars bracketing the name of a vector indicate the norm of the vector:

   ||A||
Throughout this discussion we'll use text enclosed in angle brackets to indicate values to be filled in by the creator of a frame kernel. Examples are:

      Token                 Replacement Value
   -------------            -----------------------------------------
   <vec_ID>                 'PRI' or 'SEC' [See discussion of
                            two-vector frames below.]
   <frame_name>             SPICE frame name, .e.g. 'J2000'
   <frame_ID>               Integer frame ID code
   <observer_ID>            NAIF integer ID for the observing body
   <aberration correction>  String indicating aberration correction,
                            e.g.:  'NONE', 'LT', 'XLT', 'LT+S'


Top

Required Keywords for Parameterized Dynamic Frames




All parameterized dynamic frame kernel definitions contain the assignments shown here:

   FRAME_<frame_name>                  =  <frame_ID>
   FRAME_<frame_ID>_NAME               =  <frame_name>
   FRAME_<frame_ID>_CLASS              =  5
   FRAME_<frame_ID>_CLASS_ID           =  <frame_ID>
   FRAME_<frame_ID>_CENTER             =  <center_ID>
 
   FRAME_<frame_ID>_RELATIVE           =  <base_frame_name>
   FRAME_<frame_ID>_DEF_STYLE          =  'PARAMETERIZED'
   FRAME_<frame_ID>_FAMILY             =  <frame_family>
These first five of the assignments are common to all CSPICE frame definitions; the class code 5 indicates that the frame is dynamic. See the section ``Guidelines for Frame Specification'' in the chapter ``Creating a Frame Kernel'' above for a detailed discussion of these assignments.

The sixth assignment (for keyword FRAME_<frame_ID>_RELATIVE) is the ``base frame'' specification; this indicates the frame the transformation defined by the frame kernel ``maps to'': starting with an epoch ET and a state vector S specified relative to the defined frame

   <frame name>
the frame definition determines the 6x6 state transformation matrix XFORM such that the product

   XFORM * S
yields the equivalent state specified relative to the base frame at ET.

The seventh assignment (for keyword FRAME_<frame_ID>_DEF_STYLE) is used to simplify future implementation of other dynamic frame definition styles. Only the value

   'PARAMETERIZED'
is currently supported.

The last assignment indicates the frame family. The possible values are

   'TWO-VECTOR'
   'MEAN_EQUATOR_AND_EQUINOX_OF_DATE'
   'TRUE_EQUATOR_AND_EQUINOX_OF_DATE'
   'MEAN_ECLIPTIC_AND_EQUINOX_OF_DATE'
   'EULER'
Additional, required frame kernel assignments are a function of the family to which a dynamic frame belongs. These are discussed below.



Top

Conditional Keywords for Parameterized Dynamic Frames






Top

Rotation State



A parameterized dynamic frame definition can specify a frame's ``rotation state'' as ``rotating'' or ``inertial.'' Rotating frames are nominally time-dependent, although it is possible for them to be constant (an Euler frame with all Euler angles constant is an example).

When a parameterized dynamic frame is specified as ``inertial,'' the derivative with respect to time of the transformation between the frame and any inertial frame, for example the J2000 frame, is zero. The rotation between the frame and any inertial frame is still treated as time-dependent. For such a frame F, the call

   sxform_c ( "F", "J2000", t, xform );
yields a 6x6 state transformation matrix `xform' having the structure

   +-----+-----+
   | R(t)|  0  |
   +-----+-----+
   |  0  | R(t)|
   +-----+-----+
where R(t) is the 3x3 rotation matrix that transforms vectors from frame F to the J2000 frame at time `t'. By contrast, when the rotation state of F is ``rotating,'' `xform' has the structure

   +-----+-----+
   | R(t)|  0  |
   +-----+-----+
   |dR/dt| R(t)|
   +-----+-----+
So, when the rotation state of frame F is ``inertial,'' velocities are transformed from frame F to J2000 by left-multiplication by R(t); the time derivative of the rotation from F to J2000 is simply ignored.

Normally the inertial rotation state makes sense only for slowly rotating frames such as the earth mean equator and equinox of date frame.

A parameterized dynamic frame's rotation state is specified via the assignment

   FRAME_<frame_ID>_ROTATION_STATE     =  <state>
where

   <state>
is one of

   'ROTATING'
   'INERTIAL'
For frames belonging to the parameterized dynamic frame families

   'MEAN_EQUATOR_AND_EQUINOX_OF_DATE'
   'TRUE_EQUATOR_AND_EQUINOX_OF_DATE'
   'MEAN_ECLIPTIC_AND_EQUINOX_OF_DATE'
either the rotation state must be specified, or the frame must be frozen (see ``Frozen Frames'' below).

For two-vector and Euler frames, the rotation state specification is optional; these frames are considered to be rotating by default.

When the rotation state of a parameterized frame is specified, the frame cannot be frozen; these options are mutually exclusive.



Top

Freeze Epoch



A parameterized dynamic frame definition can specify a frame as ``frozen'' at a particular epoch. The rotation between a frozen frame and its base frame is constant; the derivative with respect to time of this rotation is zero.

A frozen frame whose base frame is time-varying is still time-varying: it is the relationship between the frozen frame and the base frame that is time-independent.

A frame is declared frozen by specifying a ``freeze epoch.'' This is done via the assignment:

   FRAME_<frame_ID>_FREEZE_EPOCH       =  <time_spec>
where

   <time_spec>
is a TDB calendar date whose format conforms to the SPICE text kernel date format specification. These dates

    -- are unquoted

    -- start with the character

            @
    -- contain no embedded blanks

An example of a template for these calendar strings is

   @YYYY-MON-DD/HR:MN.SEC.###
Literal examples include

   @7-MAR-2005
   @March-7-2005-3:10:39.221
   @2005-MAR-07/3:10:39.221
Note that unlike time strings supported by the CSPICE function str2et_c, time system tokens such as

   UTC
   TDT
   TDB
are not supported; times are always assumed to be TDB.

For frames belonging to the parameterized dynamic frame families

   'MEAN_EQUATOR_AND_EQUINOX_OF_DATE'
   'TRUE_EQUATOR_AND_EQUINOX_OF_DATE'
   'MEAN_ECLIPTIC_AND_EQUINOX_OF_DATE'
either the frame must be frozen or the rotation state must be specified, (see ``Rotation State'' above).

For two-vector and Euler frames, the freeze epoch specification is optional; these frames are considered to be time-varying relative to their base frames by default.

When a parameterized frame is frozen, the rotation state of the frame cannot be specified; these options are mutually exclusive.



Top

Two-Vector Frames





Two-vector frames use two user-specified, non-parallel vectors to define the mutually orthogonal axes of a right-handed reference frame.

In a two-vector frame definition, one defining vector is parallel to a specified axis of the reference frame; this vector is called the ``primary vector.'' The other vector, called the ``secondary vector,'' defines another axis: the component of the secondary vector orthogonal to the primary vector is parallel to a specified axis of the reference frame. The secondary vector itself need not be, and typically is not, aligned with an axis of the defined frame.

Below, we'll call the primary and secondary defining vectors PRI and SEC, and we'll name the axes of the right-handed frame X, Y, and Z. The unit +Z vector is the cross product of the unit +X and +Y vector.

In a two-vector frame definition, the vectors PRI and SEC are specified geometrically; for example, PRI could be the position of the earth relative to a spacecraft, and SEC could be defined by the right ascension and declination of a given star in a specified reference frame.

In a frame kernel, the vectors PRI and SEC are associated with two members of the set of unit vectors

   { X, -X, Y, -Y, Z, -Z }
An example: in this case PRI is associated with -Z and SEC is associated with +X. SEC itself is not parallel to the X axis, but the component of SEC orthogonal to PRI points in the +X direction.

The diagram below shows the relationship between PRI, SEC, X, Y, and Z:

 
      Component of SEC orthogonal to PRI
                      |
                      |      ^
                      v      |
                    <-----+--+
                     \    |  |
                      \   +--+
                       \     |
                    SEC \    |  +Z  = - PRI / ||PRI||
                         \   |
                          \  |
                           \ +--+
                            \|  |
   +X = Y x Z  <---------+---+--+
                        /   /|
                       +---/ |
                          /| /
                         / |/|
                        /  + |  -Z  =   PRI / ||PRI||
                       /     |
                      /      |
                     v       v  PRI
 
            Z x SEC
     +Y = -----------
          ||Z x SEC||
 
      =   Z x X
By defining PRI and SEC we can create a concrete frame definition. Continuing the above example, we can define a nadir-pointing frame for the Mars Global Surveyor (MGS) spacecraft as follows:

   PRI  =  Vector from MGS to nearest point on Mars reference
           ellipsoid
 
   Z    =  -PRI / ||PRI||
 
   SEC  =  Inertially referenced velocity of MGS relative to Mars
 
   Y    =  Z x SEC / ||Z x SEC||
 
   X    =  Y x Z
For this nadir-pointing frame, -Z is the nadir direction, X points roughly in the direction of the inertially referenced spacecraft velocity, and Y is aligned with the orbital angular velocity vector.

By converting the above definition into the frame kernel ``keyword=value'' format, we can make the definition usable by the CSPICE system. Above, for brevity, we've glossed over a few aspects of the vector definitions. Below we'll discuss in detail all of the elements of two-vector frame specifications.



Top

Defining a Two-Vector Frame in a Frame Kernel






Top

Kernel Availability



In the following discussion, for brevity, we will use the term ``computable'' to describe frames whose definitions are known to CSPICE and for which kernels have been loaded sufficient to enable computation of the transformations between these frames and their base frames.

We'll also call a frame transformation between frames F1 and F2 ``computable'' if both frames F1 and F2 are computable and kernels have been loaded sufficient to enable computation of the transformation between F1 and F2. For example, the transformation between the J2000 and IAU_TITAN frames is computable once a PCK containing rotational elements for TITAN has been loaded.



Top

Specifying the Base Frame



When a two-vector frame F is defined with a base frame F_BASE, and when the necessary kernels are loaded, the transformation between F and F_BASE (in both directions) becomes computable by the CSPICE frame subsystem. In addition, for any frame F2 such that the transformation between F2 to F_BASE is computable, the transformation from F2 to F (in both directions) becomes computable.

For a two-vector frame, the base frame may be any frame F_BASE such that the transformation between F_BASE and the J2000 reference frame is computable at the time the two-vector frame definition is referenced.

Normally for two-vector frames the base frame should be set to 'J2000'; this choice yields optimal run-time efficiency. The assignment is made as follows.

   FRAME_<frame_ID>_RELATIVE           =  'J2000'
Base frame specifications are part of the two-vector frame definition because the base frame can be used to control how CSPICE chains together two-vector frames with other frames. However, from a mathematical point of view, two-vector frames are fully defined without reference to a base frame. For example, suppose the two-vector frame F1 is defined by the earth-moon position vector and the earth-sun position vector, and the base frame for F1 is IAU_EARTH. Suppose that the two-vector frame F2 is defined by the same vectors and that the base frame of F2 is J2000. Then, ignoring small round-off errors, the transformation between F1 and F2 is the identity transformation.

Base frames should not be confused with other frames occurring in two-vector frame definitions: constant vectors and velocity vectors have associated frames which are also specified by keyword assignments. See the discussion below under the heading ``Constant Vectors'' and ``Velocity Vectors'' for details.



Top

Specifying the Frame Family



Definitions of two-vector frames include the frame family specification:

   FRAME_<frame_ID>_FAMILY             =  'TWO-VECTOR'
Further assignments (discussed below) define the primary and secondary vectors and relate these vectors to the frame's axes.



Top

Specifying the Rotation state or Freeze Epoch



These specifications are optional for two-vector frames. See the section above titled ``Conditional Keywords for Parameterized Dynamic Frames'' for details.



Top

Specifying the Angular Separation Tolerance



This specification applies only to two-vector frames and is optional. To diagnose near-degenerate geometry, specifically cases where the defining vectors have angular separation too close to zero or pi radians, users can specify a limit on these angular separations. This is done via the keyword assignment

   FRAME_<frame_ID>_ANGLE_SEP_TOL      = <tolerance>
where <tolerance> is the separation limit in radians. If the angular separation of the defining vectors differs from zero or pi radians by less than the specified tolerance, an error will be signaled at run time.

When a two-vector frame definition omits specification of an angular separation tolerance, CSPICE uses a default value of one milliradian.



Top

Frame Axis Labels



The primary defining vector is associated with a frame axis via the assignment

   FRAME_<frame_ID>_PRI_AXIS           = <label>
Here

   <label>
may be any of

   { 'X',  '-X',  'Y',  '-Y',  'Z',  '-Z' }
Blanks and case in the label are not significant. Unsigned axis designations are treated as positive; optionally '+' signs may be used to prefix positive axis designations. The primary vector is aligned with the indicated axis and has the sense indicated by the implied or explicit sign.

The secondary defining vector is associated with a frame axis via the assignment

   FRAME_<frame_ID>_SEC_AXIS           = <label>
where the axis labels are as above. The assignment means that the component of the secondary vector orthogonal to the primary vector is aligned with the indicated axis and has the sense indicated by the implied or explicit sign.



Top

Vector Specifications




The vectors used to define a two-vector frame are specified by geometric means. Each defining vector may be any of:

    -- The position of one ephemeris object relative to another

    -- The vector from an observer to the nearest point on an extended body to the observer

    -- The velocity of one ephemeris object relative to another in a specified reference frame

    -- A constant vector in a specified reference frame

The frames (explicit or implicit) associated with the two defining vectors need not match each other or the base frame. CSPICE will map the defining vectors to a common frame before performing vector arithmetic to derive the axes of the defined frame.

All keywords comprising the primary vector definition start with the prefix

   FRAME_<frame_ID>_PRI_
All keywords for the second defining vector are prefixed by

   FRAME_<frame_ID>_SEC_
Here <frame_ID> is the integer ID code for the frame being defined.

Both the primary and secondary vectors are specified using the sets of keywords described below.



Top

Observer-Target Position Vectors



An observer-target position vector is simply the position of one ephemeris object relative to another. These vectors are defined by an observer, a target, an aberration correction, a reference frame, and an epoch. In the frame kernel, there is no need to specify the reference frame or epoch: the CSPICE frame subsystem will determine which frame to use, and the epoch is supplied by the calling application at run time.

The observer and target are specified by name or ID code. The aberration correction may be any value accepted by spkezr_c.

The frame kernel assignments used to define an observer-target position vector are:

   FRAME_<frame_ID>_<vec_ID>_VECTOR_DEF = 'OBSERVER_TARGET_POSITION'
   FRAME_<frame_ID>_<vec_ID>_OBSERVER   = <observer name or ID code>
   FRAME_<frame_ID>_<vec_ID>_TARGET     = <target name or ID code>
   FRAME_<frame_ID>_<vec_ID>_ABCORR     = <aberration correction>
where <vec_ID> may be either PRI or SEC, and <frame_ID> is the ID code of the frame established by the generic assignments described above.

In order for a two-vector frame using a position vector as part of its definition to be computable, kernel data must be loaded that enable computation of the specified position vector with respect to the J2000 frame.

For an example of a two-vector frame definition using an observer-target position vector, see the subsection titled ``Geocentric Solar Ecliptic (GSE) Frame'' in the appendix ``Frame Definition Examples.''



Top

Target Near point Vectors



Target near point vectors point from an observer to the closest point on an extended target body to the observer.

Target near point vectors are defined by an observer, a target, an aberration correction, a frame, and an epoch. As with position vectors, the frame and epoch are not specified in the frame kernel.

The observer and target are specified by name or ID code. Aberration corrections may be any supported by the CSPICE function subpt_c. Light time corrections are applied both to the observer- target center vector and to the rotation of the target body. The stellar aberration correction, if specified, is applied to the observer-target center vector.

The frame kernel assignments used to define a target near point position vector are:

   FRAME_<frame_ID>_<vec_ID>_VECTOR_DEF = 'TARGET_NEAR_POINT'
   FRAME_<frame_ID>_<vec_ID>_OBSERVER   = <observer name or ID code>
   FRAME_<frame_ID>_<vec_ID>_TARGET     = <target name or ID code>
   FRAME_<frame_ID>_<vec_ID>_ABCORR     = <aberration correction>
In order for a two-vector frame using a target near point vector as part of its definition to be computable, kernel data must be loaded that enable computation of the target near point vector with respect to the J2000 frame.

For an example of a two-vector frame definition using a target near point vector, see the subsection titled ``Nadir Frame for Mars Orbiting Spacecraft'' in the appendix ``Frame Definition Examples.''



Top

Observer-Target Velocity Vectors



An observer-target velocity vector is the velocity portion of the state of one ephemeris object relative to another. These vectors are defined by an observer, a target, an aberration correction, a reference frame, and an epoch. Of these, only the epoch is not specified in the frame kernel. Unlike observer-target position vectors, velocity vectors require a user-supplied frame specification. The specified frame (we'll call this the ``velocity frame'') will be used to look up the velocity vector from the CSPICE SPK subsystem.

When the velocity frame is non-inertial and aberration corrections are used, the epoch at which the velocity frame is evaluated will be adjusted by the one-way light time between the observer and the frame's center---just as is done by spkezr_c (see the header of that function for details).

The reason the velocity frame specification is crucial is that, (unlike rotations) state transformations between non-inertial frames don't preserve geometric properties of velocity vectors. Example: compare the specific angular momentum vector of a geosynchronous satellite (obtained by taking the cross product of the satellite's geocentric position and velocity vectors) in both the J2000 frame and in the earth body-fixed frame. In the latter frame, the specific angular momentum is zero. A valid two-vector frame could be defined using the satellite's position and velocity in the J2000 frame, while using the position and velocity in the earth body-fixed frame gives rise to a degenerate case for which the two-vector frame is undefined.

The observer and target defining the velocity vector are specified by name or ID code. The aberration correction may be any value accepted by spkezr_c. The velocity frame may be any computable by CSPICE, including a dynamic frame, as long as the transformation between the velocity frame and the J2000 frame doesn't require multiple levels of simulated recursion (see the discussion of recursion in the chapter ``Dynamic Frame Implementation Considerations'' below for details).

The frame kernel assignments used to define an observer-target velocity vector are:

   FRAME_<frame_ID>_<vec_ID>_VECTOR_DEF = 'OBSERVER_TARGET_VELOCITY'
   FRAME_<frame_ID>_<vec_ID>_OBSERVER   = <observer name or ID code>
   FRAME_<frame_ID>_<vec_ID>_TARGET     = <target name or ID code>
   FRAME_<frame_ID>_<vec_ID>_FRAME      = <frame_name>
   FRAME_<frame_ID>_<vec_ID>_ABCORR     = <aberration correction>
In order for a two-vector frame using a velocity vector as part of its definition to be computable, kernel data must be loaded that enable computation of the velocity vector with respect to both the velocity frame and the J2000 frame.

For an example of a two-vector frame definition using an observer-target velocity vector, see the subsection titled ``Geocentric Solar Ecliptic (GSE) Frame'' in the appendix ``Frame Definition Examples.''



Top

Constant Vectors



Constant vectors are defined by specifying a reference frame and a vector expressed relative to that frame. Optionally, aberration corrections may be specified.

The coordinates of a constant vector may be specified in any of the rectangular, latitudinal, or RA/DEC (right ascension and declination) systems. If the coordinates are angular, the associated angular units must be specified; any angular units supported by the CSPICE function convrt_c may be used.

All constant vectors require the frame kernel assignments

   FRAME_<frame_ID>_<vec_ID>_VECTOR_DEF = 'CONSTANT'
   FRAME_<frame_ID>_<vec_ID>_SPEC       = <coordinate_system>
   FRAME_<frame_ID>_<vec_ID>_FRAME      = <frame_name>
where <coordinate_system> is one of

   'RECTANGULAR'
   'LATITUDINAL'
   'RA/DEC'
and the frame is any computable by CSPICE, including a dynamic frame, as long as the transformation between the constant vector's frame and the J2000 frame doesn't require multiple levels of simulated recursion (see the discussion of recursion in the chapter ``Dynamic Frame Implementation Considerations'' below for details).

When the coordinate system is rectangular, the vector is specified by the frame kernel assignment

   FRAME_<frame_ID>_<vec_ID>_SPEC   = 'RECTANGULAR'
   FRAME_<frame_ID>_<vec_ID>_VECTOR = ( <X component>,
                                        <Y component>,
                                        <Z component>  )
When the coordinate system is latitudinal, the vector is specified by the frame kernel assignments

   FRAME_<frame_ID>_<vec_ID>_SPEC      = 'LATITUDINAL'
   FRAME_<frame_ID>_<vec_ID>_UNITS     = <angular_units>
   FRAME_<frame_ID>_<vec_ID>_LONGITUDE = <longitude>
   FRAME_<frame_ID>_<vec_ID>_LATITUDE  = <latitude>
where <angular_units> designates one of the units supported by the CSPICE function convrt_c. The set of supported units includes

   'RADIANS'
   'DEGREES'
   'ARCSECONDS'
When the coordinate system is RA/DEC, the vector is specified by the frame kernel assignments

   FRAME_<frame_ID>_<vec_ID>_SPEC      = 'RA/DEC'
   FRAME_<frame_ID>_<vec_ID>_UNITS     = <angular_units>
   FRAME_<frame_ID>_<vec_ID>_RA        = <RA>
   FRAME_<frame_ID>_<vec_ID>_DEC       = <DEC>
where <angular_units> are as described above.

Aberration corrections are optional for constant vectors. The set of available corrections is unique to this application: either light time correction or stellar aberration correction may be applied, but both cannot be applied together.

Light time corrections adjust the orientation of the constant vector's frame for the one-way light time between the center of the frame and a specified observer. The application to the frame of light time correction is identical to that performed by the CSPICE function spkezr_c when it is asked to compute a light-time corrected state relative to a non-inertial reference frame. Supported light time corrections are any of those supported by spkezr_c that don't include stellar aberration correction.

The user may also correct the constant vector for stellar aberration; this correction is a function of the constant vector and the velocity of an observer relative to the solar system barycenter. A typical application would be to correct an inertially referenced star direction vector for the stellar aberration induced by motion of an observing spacecraft. The supported stellar aberration corrections are

   'S'      {correct for stellar aberration, reception case}
   'XS'     {correct for stellar aberration, transmission case}
In the application above, one would correct the apparent observer-star direction by selecting the 'S' option. See the discussion in the header of the CSPICE function spkezr_c for a description of the ``reception'' and ``transmission'' aberration correction cases.

When aberration corrections are desired, the observer and the correction are specified by the frame kernel assignments

   FRAME_<frame_ID>_<vec_ID>_OBSERVER  = <observer name or ID code>
   FRAME_<frame_ID>_<vec_ID>_ABCORR    = <aberration correction>
In order for a two-vector frame using a constant vector as part of its definition to be computable, kernel data must be loaded that enable computation of the specified vector with respect to both the constant vector's frame and the J2000 frame.

For examples of two-vector frame definitions using constant vectors, see the subsections titled ``Geocentric Solar Magnetospheric (GSM) Frame'' and ``Mercury Solar Equatorial (MSEQ) Frame'' in the appendix ``Frame Definition Examples.''



Top

Mean Equator and Equinox of Date Frames





Mean Equator and Equinox of Date Frames are defined for a solar system body (for example, a planet) using mathematical models of the orientation of the body's mean equatorial and orbital planes. The term ``mean equator'' indicates that orientation of the equatorial plane is modeled accounting for precession only. The ``mean equinox'' is the intersection of the body's mean orbital plane with the mean equatorial plane. The X-axis of such a frame is aligned with the cross product of the north-pointing vectors normal to the body's mean equator and mean orbital plane of date. The Z-axis is aligned with the first of these normal vectors. The Y axis is the cross product of the Z and X axes. The resulting reference frame is time-varying; the term ``of date'' means this frame is evaluated at a specified epoch.

The mathematical model for a mean equator and equinox of date frame is typically called a ``precession model''; CSPICE adopts this usage.

The CSPICE frame subsystem supports mean equator and equinox of date frames via precession models built into CSPICE. In principle, for any body, a frame kernel definition for a mean equator and equinox of date frame identifies which precession model to use for that body. Currently CSPICE supports only one precession model: the 1976 IAU precession model for the earth.



Top

Defining a Mean Equator and Equinox of Date Frame in a Frame Kernel






Top

Specifying the Base Frame



The base frame of a mean equator and equinox of date frame is a function of the precession model. For the 1976 IAU earth precession model the base frame is J2000. This association is made via the assignment:

   FRAME_<frame_ID>_RELATIVE           =  'J2000'


Top

Specifying the Frame Family



A mean equator and equinox of date frame is identified by frame family specification:

   FRAME_<frame_ID>_FAMILY = 'MEAN_EQUATOR_AND_EQUINOX_OF_DATE'


Top

Specifying the Precession Model



The 1976 IAU precession model is ``selected'' via the assignment:

   FRAME_<frame_ID>_PREC_MODEL   = 'EARTH_IAU_1976'


Top

Specifying a Rotation State or Freeze Epoch



Although mean equator and equinox of date frames are, strictly speaking, non-inertial, their time variation may be very slow. In some cases it may be desirable to treat them as inertial (specifically, non-rotating), perhaps in order to simplify computations or to ensure compatibility with computations from another source.

Users can instruct the CSPICE frame subsystem to treat a mean equator and equinox of date frame as either inertial or rotating by making a ``rotation state'' assignment. Users can also direct the frame subsystem to treat a mean equator and equinox of date frame as though it were ``frozen'' at a specified epoch. See the section above titled ``Conditional Keywords for Parameterized Dynamic Frames'' for instructions on how to make these assignments.

Definitions of mean equator and equinox of date frames require either, but not both, the rotation state or a freeze epoch to be specified.

For examples of Mean Equator and Equinox of Date frame definitions, see the subsection titled ``Earth Mean Equator and Equinox of Date Frames'' in the appendix ``Frame Definition Examples.''



Top

True Equator and Equinox of Date Frames





True Equator and Equinox of Date Frames may be viewed as a refinement of mean equator and equinox of date frames. The term ``true equator'' indicates that orientation of a body's equatorial plane is modeled accounting for precession and nutation. The ``true equinox'' is the intersection of the body's mean orbital plane with the true equatorial plane. The X-axis of such a frame is aligned with the cross product of the north-pointing vectors normal to the body's true equator and mean orbital plane of date. The Z-axis is aligned with the first of these normal vectors. The Y axis is the cross product of the Z and X axes. The term ``of date'' means that these axes are evaluated at a specified epoch.



Top

Defining a True Equator and Equinox of Date Frame in a Frame Kernel




True Equator and Equinox of date frame definitions are nearly identical to those for mean of date frames (see above): the only differences are the frame family specification and the addition of an assignment identifying the nutation model.



Top

Specifying the Base Frame



The base frame of a true equator and equinox of date frame is a function of the precession model. For the 1976 IAU earth precession model the base frame is J2000. This association is made via the assignment:

   FRAME_<frame_ID>_RELATIVE           =  'J2000'


Top

Specifying the Frame Family



A true equator and equinox of date frame is identified by frame family specification:

   FRAME_<frame_ID>_FAMILY = 'TRUE_EQUATOR_AND_EQUINOX_OF_DATE'


Top

Specifying the Precession Model



Currently CSPICE supports only one precession model: the 1976 IAU precession model for the earth.

The 1976 IAU precession model is ``selected'' via the assignment:

   FRAME_<frame_ID>_PREC_MODEL   = 'EARTH_IAU_1976'


Top

Specifying the Nutation Model



The choice of nutation model is specified by the assignment:

   FRAME_<frame_ID>_NUT_MODEL     = <nutation_model>
Currently the only available nutation model is the 1980 IAU nutation model for the earth. An assignment specifying this model has the form:

   FRAME_<frame_ID>_NUT_MODEL     = 'EARTH_IAU_1980'


Top

Specifying a Rotation State or Freeze Epoch



Although true equator and equinox of date frames are, strictly speaking, non-inertial, their time variation may be very slow. In some cases it may be desirable to treat them as inertial (specifically, non-rotating), perhaps in order to simplify computations or to ensure compatibility with computations from another source.

Users can instruct the CSPICE frame subsystem to treat a true equator and equinox of date frame as either inertial or rotating by making a ``rotation state'' assignment. Users can also direct the frame subsystem to treat a true equator and equinox of date frame as though it were ``frozen'' at a specified epoch. See the section above titled ``Conditional Keywords for Parameterized Dynamic Frames'' for instructions on how to make these assignments.

Definitions of true equator and equinox of date frames require either, but not both, the rotation state or a freeze epoch to be specified.

For examples of True Equator and Equinox of Date frame definitions, see the subsection titled ``Earth True Equator and Equinox of Date Frames'' in the appendix ``Frame Definition Examples.''



Top

Mean Ecliptic and Equinox of Date Frames





Mean Ecliptic and Equinox of Date Frames are closely related to mean equator and equinox of date frames: for a given body, the former is obtained by rotating the latter about the X-axis by the mean obliquity of date.

The term ``mean equator'' indicates that orientation of a body's equatorial plane is modeled accounting for precession. The ``mean equinox'' is the intersection of the body's mean orbital plane with the mean equatorial plane. The X-axis of such a frame is aligned with the cross product of the north-pointing vectors normal to the body's mean equator and mean orbital plane of date. The Z-axis is aligned with the second of these normal vectors. The Y axis is the cross product of the Z and X axes. The term ``of date'' means that these axes are evaluated at a specified epoch.



Top

Defining a Mean Ecliptic and Equinox of Date Frame in a Frame Kernel




Mean Ecliptic and Equinox of date frame definitions are nearly identical to those for mean of date frames (see above): the only differences are the frame family specification and the addition of an assignment identifying the mean obliquity model.



Top

Specifying the Base Frame



The base frame of a mean ecliptic and equinox of date frame is a function of the precession model. For the 1976 IAU earth precession model the base frame is J2000. This association is made via the assignment:

   FRAME_<frame_ID>_RELATIVE           =  'J2000'


Top

Specifying the Frame Family



A mean ecliptic and equinox of date frame is identified by frame family specification:

   FRAME_<frame_ID>_FAMILY = 'MEAN_ECLIPTIC_AND_EQUINOX_OF_DATE'


Top

Specifying the Precession Model



Currently CSPICE supports only one precession model: the 1976 IAU precession model for the earth.

The 1976 IAU precession model is ``selected'' via the assignment:

   FRAME_<frame_ID>_PREC_MODEL   = 'EARTH_IAU_1976'


Top

Specifying the Mean Obliquity Model



The choice of mean obliquity model is specified by the assignment:

   FRAME_<frame_ID>_OBLIQ_MODEL     = <obliquity_model>
Currently the only available mean obliquity model is the 1980 IAU obliquity model for the earth. An assignment specifying this model has the form:

   FRAME_<frame_ID>_OBLIQ_MODEL     = 'EARTH_IAU_1980'


Top

Specifying a Rotation State or Freeze Epoch



Although mean ecliptic and equinox of date frames are, strictly speaking, non-inertial, their time variation may be very slow. In some cases it may be desirable to treat them as inertial (specifically, non-rotating), perhaps in order to simplify computations or to ensure compatibility with computations from another source.

Users can instruct the CSPICE frame subsystem to treat a mean ecliptic and equinox of date frame as either inertial or rotating by making a ``rotation state'' assignment. Users can also direct the frame subsystem to treat a mean ecliptic and equinox of date frame as though it were ``frozen'' at a specified epoch. See the section above titled ``Conditional Keywords for Parameterized Dynamic Frames'' for instructions on how to make these assignments.

Definitions of mean ecliptic and equinox of date frames require either, but not both, the rotation state or a freeze epoch to be specified.

For examples of Mean Ecliptic and Equinox of Date frame definitions, see the subsection titled ``Earth Mean Ecliptic and Equinox of Date Frames'' in the appendix ``Frame Definition Examples.''



Top

Euler Frames





An Euler frame is defined by a sequence of rotation axes and corresponding time-dependent Euler angles. Each angle is defined by a set of polynomial coefficients. A reference epoch must be provided in the frame definition; the independent variable of each polynomial represents ephemeris seconds past the J2000 TDB epoch.

The rotation defined by the Euler angles maps position vectors via left multiplication from the defined Euler reference frame to the base frame:

   V           = r(t) * V
    base_frame           Euler_frame
This rotation can be considered to be a time-dependent matrix

   r(t)
where r(t) represents the composition of the rotations defined by the input angle-axis pairs. Naming the axis indices and angles of the Euler angle sequence

   axindx_i, angle_i,  i = 1, 2, 3
r(t) is

   r(t) = [ angle_1(t) ]      [ angle_2(t) ]      [ angle_3(t) ]
                     axindx_1            axindx_2            axindx_3
The axis indices axindx_i, for i = 1, 2, 3, are in the set { 1, 2, 3 }; axindx_2 cannot equal axindx_1 or axindx_3. For example, we could have

   axindx_1 = 3
   axindx_2 = 1
   axindx_3 = 3
Here the notation

   [ A ]
        j
stands for a frame rotation by the angle A radians about the jth axis of a right-handed frame, where we assign the axes {X, Y, Z} the indices {1, 2, 3} respectively:

   +-                     -+
   |   1       0      0    |
   |   0     cos A   sin A |   =  [ A ]
   |   0    -sin A   cos A |           1
   +-                     -+
 
   +-                     -+
   |  cos A    0    -sin A |
   |   0       1      0    |   =  [ A ]
   |  sin A    0     cos A |           2
   +-                     -+
 
   +-                     -+
   |  cos A   sin A   0    |
   | -sin A   cos A   0    |   =  [ A ]
   |   0       0      1    |           3
   +-                     -+
The base frame can be constructed from the Euler frame via a sequence of Euler angle rotations as follows:

    1. Rotate the axes of the Euler frame by angle_3 about the axis indexed by axindx_3.

    2. Rotate the axes of the frame resulting from the first rotation by angle_2 about the axis indexed by axindx_2.

    3. Rotate the axes of the frame resulting from the second rotation by angle_1 about the axis indexed by axindx_1.

The resulting set of axes are those of the base frame.

The rotation angles are defined as follows: letting t0 represent the reference epoch, and letting

   c   ,  i = 1, 2, 3;   j = 0, ... , ni
    i,j
be the polynomial coefficients for the ith angle, we have

                                                        n1
   angle_1(t) = c   + c   * (t-t0) + ... + c    * (t-t0)
                 1,0   1,1                  1,n1
 
                                                        n2
   angle_2(t) = c   + c   * (t-t0) + ... + c    * (t-t0)
                 2,0   2,1                  2,n2
 
                                                        n3
   angle_3(t) = c   + c   * (t-t0) + ... + c    * (t-t0)
                 3,0   3,1                  3,n3
See the Rotation Required Reading, rotation.req, or the header of the CSPICE function eul2m_c for details concerning definition of rotations via Euler angles. Note however that the referenced document and source code use a different convention for labeling Euler angles and their rotation axes: here the elements of the rotation sequence are numbered left to right; in those documents the order is that in which rotations are performed, namely right to left.



Top

Defining an Euler Frame in a Frame Kernel






Top

Specifying the Base Frame



The base frame of an Euler frame is specified via the assignment:

   FRAME_<frame_ID>_RELATIVE           =  '<frame_name>'


Top

Specifying the Frame Family



An Euler frame is identified by frame family specification:

   FRAME_<frame_ID>_FAMILY = 'EULER'


Top

Specifying the Epoch



The zero epoch for the independent variable of the polynomials is defined using the SPICE text kernel calendar ephemeris time syntax. A sample template is shown below:

   FRAME_<frame_ID>_EPOCH           =  @YYYY-MON-DD/HR:MN.SEC.###
A concrete example is:

   FRAME_<frame_ID>_EPOCH           =  @2000-JAN-1/12:00:00.000
The calendar time string is assumed to represent a TDB epoch.

See the discussion in the section ``Freeze Epoch'' above or the Kernel Required Reading, kernel.req, for further information.



Top

Specifying the Euler Angles



Euler angles are specified by an axis sequence, a set of polynomial coefficients, and associated units. The axes are specified by an assignment of the form:

   FRAME_<frame_ID>_AXES            =  ( <index of axis 1>
                                         <index of axis 2>
                                         <index of axis 3> )
The axis indices must be taken from the set

   { 1, 2, 3 }
and the middle value must differ from its neighbors. The first integer listed is the axis index for angle 1, the second for angle 2, and the last for angle 3, where the role of the angles is as shown in the equation for r(t) above.

Let n1, n2, and n3 represent the maximum degrees of the polynomials for angles 1, 2, and 3 respectively. Then the polynomial coefficients are defined by the assignments

   FRAME_<frame_ID>_ANGLE_1_COEFFS = ( <order 0 coefficient>
                                       <order 1 coefficient>
                                              ...
                                       <order n1 coefficient>  )
 
   FRAME_<frame_ID>_ANGLE_2_COEFFS = ( <order 0 coefficient>
                                       <order 1 coefficient>
                                              ...
                                       <order n2 coefficient>  )
 
   FRAME_<frame_ID>_ANGLE_3_COEFFS = ( <order 0 coefficient>
                                       <order 1 coefficient>
                                              ...
                                       <order n3 coefficient>  )
Angular units are specified by the frame kernel assignment

   FRAME_<frame_ID>_UNITS     = <angular_units>
where <angular_units> designates one of the units supported by the CSPICE function convrt_c. The set of supported units includes

   'RADIANS'
   'DEGREES'
   'ARCSECONDS'
For an example of an Euler frame definition, see the subsection titled ``Euler Frames'' in the appendix ``Frame Definition Examples.''



Top

Product Frames





Product frames may be thought of as a generalization of TK frames. The orientation of a product frame relative to a specified base frame is defined by a product of one or more frame transformations, where each factor may be any transformation computable by the CSPICE frame subsystem.

Using the notation

     B
   T
     A
to indicate the transformation from frame A to frame B, and letting the names

   PRODUCT
   BASE
denote a product frame and a ``base'' frame relative to which the orientation of the product frame is defined, the transformation from the base frame to the product frame is defined by a product of one or more frame transformation ``factors'' consisting of transformations from a given ``from'' frame to a given ``to'' frame:

     PRODUCT      TO_1       TO_2            TO_N-1      TO_N
   T          =  T        * T       * ... * T         * T
     BASE         FROM_1     FROM_2          FROM_N-1    FROM_N
If the vector

   v
     BASE
is expressed relative to the base frame, then applying a product frame transformation to the vector expresses the vector relative to the product frame:

                PRODUCT
   v        =  T        * v
    PRODUCT     BASE       BASE
In implementation of the equation above, the factor transformations on the right hand side of the product frame's definition are applied in right-to-left order.

The ``from'' and ``to'' frames of a product frame definition may be completely arbitrary. The only restriction on these frames is that the transformation from each ``from'' frame to its corresponding ``to'' frame must be computable by CSPICE at the time the product frame is used.

Note that because product frames are parameterized dynamic frames, limits on recursion depth for dynamic frames imply that while the factors may be dynamic frames, they may not be dynamic frames that require a level of recursion in order to evaluate their orientation.



Top

Defining a Product Frame in a Frame Kernel






Top

Specifying the Base Frame



The base frame of a product frame is specified via the assignment:

   FRAME_<frame_ID>_RELATIVE =  '<frame_name>'


Top

Specifying the Frame Family



A product frame is identified by frame family specification:

   FRAME_<frame_ID>_FAMILY = 'PRODUCT'


Top

Specifying the Factors



The factor transformations are specified by the kernel variable assignments

   FRAME_<frame_ID>_FROM_FRAMES = ( <from_frame 1> ... <from_frame N> )
   FRAME_<frame_ID>_TO_FRAMES   = ( <to_frame 1>   ... <to_frame N>   )
The ``from'' and ``to'' frames must be specified by name.

The Ith elements of the respective right-hand-side vectors of ``from'' and ``to'' frame names define the Ith factor transformation. The order of the factors in the kernel variables is the same as the order of the factors in the transformation product. When a vector is transformed from the base frame to the product frame, the transformations defined by the factors are applied in right-to-left order: the factor defined by the frames indexed by ``N'' is applied first.



Top

Dynamic Frame Implementation Considerations







Top

Introduction




This chapter discusses issues affecting implementation of dynamic frames:

    -- Simulated Recursion

    -- Frame Derivative Accuracy

    -- Degenerate Geometry

    -- Efficiency

By necessity, this chapter presents some aspects of the implementation of the CSPICE parameterized dynamic frame subsystem. The implementation described here is not considered part of the CSPICE API specification. Although unlikely, this implementation could be changed in a future version of the CSPICE Toolkit.



Top

Simulated Recursion






Top

The Need for Recursion in the CSPICE Frame Subsystem



In the following discussion, we'll use the graph notation below to indicate that function A calls function B:

   A -> B
A function R_0 is ``recursive'' if it calls itself

   R_0 -> R_0
or if some sequence of calls initiated in the function R_0 results in a call to R_0:

   R_0 -> R_1-> ... -> R_0
ANSI standard Fortran 77 doesn't permit recursive calls. However, the implementation of two-vector frames requires sequences of calls that at face value are recursive. For example, to look up a state vector in the GSE frame (see the appendix ``Frame Definition Examples''), the function SPKEZ must initiate the sequence of calls (ellipses indicate omitted portions of the call graph)

   SPKEZ -> ... -> FRMGET -> ... -> SPKEZ -> ... -> FRMGET
Both SPKEZ and FRMGET are called recursively in this graph.

This issue affects not only SPICELIB but CSPICE and Icy as well because these products rely on the SPICELIB (Fortran) implementation of the frame subsystem.



Top

Implementation of Limited Simulated Recursion



SPICELIB solves the recursion problem by providing renamed duplicates of routines that must be called recursively. For example, the invalid call graph

   SPKEZ -> ... -> FRMGET -> ... -> SPKEZ -> ... -> FRMGET
is implemented in (valid) ANSI standard Fortran 77 using the call graph

   SPKEZ -> ... -> FRMGET -> ... -> ZZSPKEZ0 -> ... -> ZZFRMGT0
To a limited extent, two levels of simulated recursion are supported in the frame subsystem, so call graphs of the form

   SPKEZ -> ... -> FRMGET    -> ... -> ZZSPKEZ0    -> ... -> ZZFRMGT0
         -> ... -> ZZSPKEZ1  -> ... -> ZZFRMGT1
are possible.

For brevity, when we refer to recursion in the following discussion, we'll omit the qualifier ``simulated.''



Top

Limits on Recursion in Frame Definitions



We say a reference frame is ``evaluated'' when the transformation from the frame to its base frame is computed for some epoch. A parameterized dynamic frame normally is evaluated each time it is referenced in a function call. For example, the calls

   spkezr_c ( moon, et, "GSE", "NONE", "EARTH", state, &lt );
 
   sxform_c ( "GSE", "J2000", et, xform );
both cause the GSE parameterized dynamic frame to be evaluated at ET.

When the definition of a parameterized dynamic frame F1 refers to a second frame F2 as

    -- the base frame

    -- the frame relative to which a constant vector is specified

    -- the frame relative to which a velocity vector is specified

the referenced frame F2 may be dynamic, but F2 must not make reference to any dynamic frame. If deeper recursion is required to evaluate the referenced frame F2, an error will occur at run time.

If F2 is not dynamic but its evaluation requires evaluation of a dynamic frame F3, the same restrictions apply to F3.

When a dynamic frame is used as a base frame in either an SPK or CK segment, evaluation of data from that segment may result in a call to the dynamic frame subsystem. That call may result in lookup of another segment whose base frame is dynamic, and so on: the original kernel lookup could easily exhaust the dynamic frame subsystem's ability to handle recursive calls.

Clearly use of dynamic frames in SPK and CK files requires caution. However, there are some ``reasonable'' applications that call for dynamic base frames in kernels, for example: representing ephemerides of earth orbiters expressed relative to the earth true equator and equinox of date frame.



Top

Frame Derivative Accuracy




Transformation of state vectors between frames F1 and F2 via a time-dependent rotation R(t) requires the derivative with respect to time of R(t): d(R(t))/dt. The accuracy of the velocity portion of a transformed state is limited by the accuracy of d(R(t))/dt. When either frame F1 or F2 is dynamic, loss of accuracy in d(R(t))/dt can occur for a number of reasons, including but not limited to:

    -- R(t) depends on CK data. Often angular rates in C-kernels have low accuracy. (This issue applies to non-dynamic frames as well.)

    -- R(t) is defined via a two-vector frame using position vectors, and the velocities associated with those vectors have low accuracy. This can happen for SPK data types for which position and velocity are represented independently, for example SPK types 3 or 9.

    -- R(t) is defined via a two-vector frame using aberration-corrected position vectors. Even if the geometric velocities of the vectors are accurate, the aberration-corrected velocities associated with those vectors will probably have low accuracy due to accuracy limitations of the aberration corrections applied to velocity vectors by the SPK subsystem.

    -- R(t) is defined via a two-vector frame using a velocity vector. The acceleration associated with the velocity vector is required to compute d(R(t))/dt, and this acceleration must be computed numerically. The results are likely to have at best single precision validity.



Top

Degenerate Geometry




Two-vector frame definitions can suffer from singularities: the defining vectors may, in some cases, become extremely close to parallel. In such cases the frame evaluation may generate meaningless results.

Because two-vector frame definitions may be perfectly valid for some epochs and yield degenerate geometry for others, testing can easily fail to reveal problems with these definitions. Careful frame design is the best defense.

As a backup measure, setting the angular separation tolerance in two-vector frame definitions can enable the frame subsystem to diagnose at run time degenerate or near-degenerate geometry. See the section ``Specifying the Angular Separation Tolerance'' above for details.



Top

Efficiency Concerns




In many cases, when recursion is required by a frame evaluation, that evaluation requires a relatively large amount of computation. For example, when an SPK call results in a two-vector frame evaluation, several additional SPK calls may be required to support the original call. The original call may be many times slower than a call requiring only non-dynamic frame evaluation.

To minimize the performance degradation imposed by recursion, avoid unnecessary references to dynamic frames in frame definitions. When possible, use J2000 or another inertial frame as the base frame, or as the frame relative to which constant or velocity vectors are defined. When it is not possible to use an inertial frame, prefer non-dynamic, non-inertial frames to dynamic frames.



Top

Switch Frames





Switch frames choose at run time other frames, called ``base frames,'' with which to align their orientation. Switch frames ``switch'' the base frames they align with as a function of time, using a prioritized list of base frames and optional, associated time bounds; this list is provided as part of the switch frame definition stored in a text kernel.

Switch frames extend the flexibility of the SPICE frame subsystem by allowing a user-defined frame to rely on different data sources at different times. For example:

    -- A switch frame representing orientation of a spacecraft could have a top-priority CK frame for reconstructed data, and a lower-priority CK frame for predicted data, where the two CK frames have different frame class IDs (also called ``instrument IDs'' in some SPICE documentation) and different associated spacecraft clocks.

    -- A switch frame representing orientation of a spacecraft could have a top-priority CK frame for reconstructed data, and a lower-priority dynamic frame for times not covered by the CK data.

    -- A switch frame representing predicted orientation of a spacecraft could have a sequence of base frames of one or more classes covering time intervals which overlap only at their endpoints.

    -- A switch frame representing orientation of a comet's nucleus could have a top-priority CK frame covering a time interval over which accurate orientation data are available, and a lower-priority PCK frame for a time interval of greater extent.

Base frames of a switch frame may have associated time intervals limiting their applicability. Given a request time, a base frame will be considered as a data source only if the request time falls within the time interval associated with that base frame. If time intervals are used in a switch frame definition, they must be provided for all base frames of that switch frame.

If base frames of a switch frame don't have associated time intervals, the base frames are applicable for all request times.

A switch frame selects a base frame as follows: given a request time, the switch frame subsystem attempts to obtain orientation data from the highest priority, applicable base frame. If that base frame is a CK frame and data are unavailable, the next applicable frame in the base frame list is used, and so on. If an applicable base frame is not a CK frame and requested data are unavailable, an error is signaled. The orientation and optional angular velocity of the switch frame are those of the selected base frame.



Top

Specifying Switch Frames




As with other frame classes, switch frames require a frame name, frame ID code, frame class ID code, and center. It is recommended that the frame class ID match the frame ID. The frame class is 6. The initial part of the frame specification has this form:

   FRAME_<name>             = <ID code>
   FRAME_<ID code>_NAME     = '<name>'
   FRAME_<ID code>_CLASS    = 6
   FRAME_<ID code>_CLASS_ID = <frame class ID>
   FRAME_<ID code>_CENTER   = <center ID code or name>


Top

The Base Frame List



The next part of the specification is the prioritized base frame list:

   FRAME_<ID code>_ALIGNED_WITH = (
       < lowest priority base frame ID or name >
       < next-lowest priority base frame ID or name >
          ...
       < highest priority base frame ID or name >  )
All base frames of a given switch frame must be specified by name, or all must be specified by frame ID code.

All base frames of a switch frame must have specifications available at the time the switch frame is used. This applies even to CK base frames. Note that CK frame ID codes and frame class ID codes are not required to match; the latter is the ID stored in CK files. It is the frame ID code that's required in the base frame list; this is provided by a CK frame specification.

A base frame may occur multiple times in the base frame list. This can be useful for base frame lists that have associated time intervals.



Top

Time Intervals Associated with Base Frames



Optional time intervals associated with base frames are specified by two kernel variables, respectively containing start and stop times:

   FRAME_<ID code>_START = (
   < start time for lowest priority base frame >
   < start time for next-lowest priority base frame >
      ...
   < start time for highest priority base frame>  )
 
   FRAME_<ID code>_STOP = (
   < stop time for lowest priority base frame >
   < stop time for next-lowest priority base frame >
      ...
   < stop time for highest priority base frame>  )
If time intervals are provided for a switch frame, the count of start times must match the count of stop times, and each must match the count of entries in the base frame list.

Start and stop times may be specified by single-quoted time strings, double precision numbers, or as times using the text kernel ``@ format.'' For example:

   '2021-12-31T12:00:00'
   694224069.183907
   6.94224069183907E+08
   @2021-DEC-31/12:01:09.183907
Times provided as single-quoted strings must be accepted by the CSPICE function str2et_c. A leapseconds kernel must be loaded in order to use such time strings.

Numeric values are interpreted as seconds past J2000 TDB. Times in text kernel time format are interpreted as TDB calendar dates. Use of times in either of these formats does not require a leapseconds kernel.

See the Kernel Required Reading kernel.req for details concerning the text kernel time format and accepted formats of double precision values.

Each list of times for a given switch frame must be specified by values of the same type: string or numeric. Times in text kernel format are actually considered to be numeric values. The types used for a switch frame's list of start times and for its list of stop times need not match, but use of consistent types is recommended for readability.



Top

Binary Search



To improve efficiency of base frame selection for a given switch frame and request time, the switch frame subsystem may perform a binary search of the base frames' time intervals to find the highest priority frame providing data for the request time.

A switch frame is eligible for binary search if:

    -- Its base frames have associated time intervals

    -- Any pair of time intervals is disjoint or has only singleton overlap. For example, the stop time of one interval may coincide with the start time of another.

    -- The intervals are listed in increasing time order. This requires the intervals for later times to be associated with higher priority base frames than those of intervals for earlier times.

If the base frame identified by a binary search is a CK frame, and if it does not provide data for the request time, it is still possible that data are available in the special case where the request time matches the start time of the selected base, and if that time is also the stop time of preceding, next-lower-priority interval. In this case the preceding interval will be checked for data availability.



Top

Switch Frame Connections




Connections between switch frames and other frames are made by the CSPICE functions frmget_c and rotget_c. If either of these functions is called with the frame ID of a switch frame as input, and if the selected base frame is

    -- a built-in inertial frame, then the output frame ID will be that of J2000, and the returned transformation will be that from the inertial base frame to J2000.

    -- a PCK frame, then the output frame ID will be that of J2000, and the returned transformation will be that from the PCK base frame to J2000.

    -- a CK frame, then the output frame ID will be that of the CK frame's base frame, and the returned transformation will be that from the CK frame to its base frame.

    -- a TK frame, then the output frame ID will be that of the TK frame's base frame, and the returned transformation will be that from the TK frame to its base frame.

    -- a dynamic frame, then the output frame ID will be that of the dynamic base frame itself, and the returned transformation will be the identity matrix of the appropriate dimension.

    -- a switch frame, then the output frame ID will be that of the switch base frame itself, and the returned transformation will be the identity matrix of the appropriate dimension.



Top

Switch Frame Buffering




The switch frame subsystem buffers switch frame specifications in a form suitable for efficient use. Expensive operations such as kernel pool lookups, frame name conversions, and time string conversions are performed only when the buffer contents must be modified.

It is possible for a user application to use many more switch frames than can be buffered concurrently, but changing the buffer contents with high frequency is inefficient.

The buffer limits shown below are hard-coded. They may be increased in future versions of CSPICE.

    -- Maximum count of switch frames loaded concurrently: 1013

    -- Maximum count of base frames of all concurrently loaded switch frames: 15000



Top

Appendix. ``Built in'' Inertial Reference Frames







Top

Complete List of ``Built in'' Inertial Reference Frames



SPICE software includes the definitions of several inertial reference frames. The numeric IDs and names of the inertial frames defined in SPICE software are:

      ID  Name        Description
   -----  --------    -------------------------------------------
       1  J2000       Earth mean equator, dynamical equinox of J2000.
                      The root reference frame for SPICE.
 
       2  B1950       Earth mean equator, dynamical equinox of B1950.
                      The B1950 reference frame is obtained by
                      precessing the J2000 frame backwards from
                      Julian year 2000 to Besselian year 1950, using
                      the 1976 IAU precession model.
 
                      The rotation from B1950 to J2000 is
 
                      [ -z ]  [ theta ]  [ -zeta ]
                             3          2          3
                      The values for z, theta, and zeta are computed
                      from the formulas given in table 5 of [5].
 
                      z     =  1153.04066200330"
                      theta =  1002.26108439117"
                      zeta  =  1152.84248596724"
 
       3  FK4         Fundamental Catalog (4). The FK4 reference
                      frame is derived from the B1950 frame by
                      applying the equinox offset determined by
                      Fricke.
 
                      [ 0.525" ]
                                3
 
       4  DE-118      JPL Developmental Ephemeris (118). The DE-118
                      reference frame is nearly identical to the FK4
                      frame. It is also derived from the B1950 frame.
                      Only the offset is different
 
                      [ 0.53155" ]
                                  3
 
                      In [2], Standish uses two separate rotations,
 
                      [ 0.00073" ]  P [ 0.5316" ]
                                   3              3
 
                      (where P is the precession matrix used above to
                      define the B1950 frame). The major effect of the
                      second rotation is to correct for truncating the
                      magnitude of the first rotation. At his
                      suggestion, we will use the untruncated value,
                      and stick to a single rotation.
 
 
                      Most of the other DE historical reference frames
                      are defined relative to either the DE-118 or
                      B1950 frame. The values below are taken
                      from [4].
 
                      DE number  Offset from DE-118  Offset from B1950
                      ---------  ------------------  -----------------
                             96            +0.1209"           +0.4107"
                            102            +0.3956"           +0.1359"
                            108            +0.0541"           +0.4775"
                            111            -0.0564"           +0.5880"
                            114            -0.0213"           +0.5529"
                            122            +0.0000"           +0.5316"
                            125            -0.0438"           +0.5754"
                            130            +0.0069"           +0.5247"
 
       5  DE-96       JPL Developmental Ephemeris ( 96).
 
       6  DE-102      JPL Developmental Ephemeris (102).
 
       7  DE-108      JPL Developmental Ephemeris (108).
 
       8  DE-111      JPL Developmental Ephemeris (111).
 
       9  DE-114      JPL Developmental Ephemeris (114).
 
      10  DE-122      JPL Developmental Ephemeris (122).
 
      11  DE-125      JPL Developmental Ephemeris (125).
 
      12  DE-130      JPL Developmental Ephemeris (130).
 
      13  GALACTIC    Galactic System II. The Galactic System II
                      reference frame is defined by the following
                      rotations:
                           o          o            o
                      [ 327  ]  [ 62.6  ]  [ 282.25  ]
                              3          1            3
 
                      In the absence of better information, we
                      assume the rotations are relative to the
                      FK4 frame.
 
      14  DE-200      JPL Developmental Ephemeris (200).
 
      15  DE-202      JPL Developmental Ephemeris (202).
 
      16  MARSIAU     Mars Mean Equator and IAU vector of
                      J2000. The IAU-vector at Mars is the point
                      on the mean equator of Mars where the equator
                      ascends through the earth mean equator.
                      This vector is the cross product of Earth
                      mean north with Mars mean north.
 
      17  ECLIPJ2000  Ecliptic coordinates based upon the
                      J2000 frame.
 
                      The value for the obliquity of the
                      ecliptic at J2000 is taken from page 114
                      of [7] equation 3.222-1. This agrees with the
                      expression given in [5].
 
      18  ECLIPB1950  Ecliptic coordinates based upon the B1950
                      frame.
 
                      The value for the obliquity of the ecliptic at
                      B1950 is taken from page 171 of [7].
 
      19  DE-140      JPL Developmental Ephemeris. (140)
                      The DE-140 frame is the DE-400 frame rotated:
 
           0.9999256765384668  0.0111817701197967  0.0048589521583895
          -0.0111817701797229  0.9999374816848701 -0.0000271545195858
          -0.0048589520204830 -0.0000271791849815  0.9999881948535965
 
                      The DE-400 frame is treated as equivalent to
                      the J2000 frame.
 
      20  DE-142      JPL Developmental Ephemeris. (142)
                      The DE-142 frame is the DE-402 frame rotated:
 
           0.9999256765402605  0.0111817697320531  0.0048589526815484
          -0.0111817697907755  0.9999374816892126 -0.0000271547693170
          -0.0048589525464121 -0.0000271789392288  0.9999881948510477
 
                      The DE-402 frame is treated as equivalent to
                      the J2000 frame.
 
      21  DE-143      JPL Developmental Ephemeris. (143)
                      The DE-143 frame is the DE-403 frame rotated:
 
           0.9999256765435852  0.0111817743077255  0.0048589414674762
          -0.0111817743300355  0.9999374816382505 -0.0000271622115251
          -0.0048589414161348 -0.0000271713942366  0.9999881949053349
 
                      The DE-403 frame is treated as equivalent to
                      the J2000 frame.


Top

Inertial Reference Frame References



   [1] Jay Lieske, ``Precession Matrix Based on IAU (1976)
       System of Astronomical Constants,'' Astron. Astrophys.
       73, 282-284 (1979).
 
   [2] E.M. Standish, Jr., ``Orientation of the JPL Ephemerides,
       DE 200/LE 200, to the Dynamical Equinox of J2000,''
       Astron. Astrophys. 114, 297-302 (1982).
 
   [3] E.M. Standish, Jr., ``Conversion of Ephemeris Coordinates
       from the B1950 System to the J2000 System,'' JPL IOM
       314.6-581, 24 June 1985.
 
   [4] E.M. Standish, Jr., ``The Equinox Offsets of the JPL
       Ephemeris,'' JPL IOM 314.6-929, 26 February 1988.
 
   [5] Jay Lieske, ``Expressions for the Precession  Quantities
       Based upon the IAU (1976) System of Astronomical
       Constants'' Astron. Astrophys. 58, 1-16 (1977).
 
   [6] Laura Bass and Robert Cesarone "Mars Observer Planetary
       Constants and Models" JPL D-3444 November 1990.
 
   [7] "Explanatory Supplement to the Astronomical Almanac"
        edited by P. Kenneth Seidelmann. University Science
        Books, 20 Edgehill Road, Mill Valley, CA 94941 (1992)


Top

Low Level Inertial Reference Frame Functions



You may obtain the rotation between any two ``built in'' inertial frames using the CSPICE function irfrot_c and supplying the IDs for the frames of interest. The module header for irfrot_c, and this document, always contain the definitive list of recognized frames.

This example shows how to rotate a position vector from FK4 coordinates to J2000 coordinates (the ID for the FK4 frame is 3, the ID for the J2000 frame is 1);

      SpiceInt from = 3;
      SpiceInt to   = 1;
 
      irfrot_ ( &from, &to, rot );
      mxv_c   ( rot  , old, new );
(`rot' is a 3-by-3 matrix, `old' and `new' are 3-vectors; subroutine mxv_c multiplies a matrix and a vector to produce a vector.)

Two additional subroutines can be used to convert a frame name to ID and vice versa. This example shows how to find the index of the DE-125 frame:

      irfnum_ ( "DE-125", frid, strlen("DE-125") );
This example shows how to find the name corresponding to ID 11:

      SpiceInt frid = 11;
 
      irfnam_ ( &fride, frname, strlen(frname) );


Top

Appendix. ``Built in'' PCK-Based IAU Body-Fixed Reference Frames





SPICE software includes the definitions of body-fixed frames for all natural bodies -- planets, satellites, and some asteroids -- listed in International Astronomical Union (IAU) reports on cartographic constants. These frames are fixed to and do not move with respect to ``surface'' features of a natural object, but they do move with respect to inertial frames as the object rotates. The complete list of body-fixed frames ``built into'' SPICE is given below. Each name is constructed by adding the prefix ``IAU_'' to the name of the body. The prefix ``IAU_'' indicates that the orientation of this frame is typically determined from the IAU model for the body in question. The constants associated with this model are stored in one or more text PCK files, which, therefore, must be loaded in order for orientation of these frames to be computed.

   IAU_52_EUROPA
   IAU_ADRASTEA
   IAU_AMALTHEA
   IAU_ANANKE
   IAU_ARIEL
   IAU_ARROKOTH
   IAU_ATLAS
   IAU_BELINDA
   IAU_BENNU
   IAU_BIANCA
   IAU_BORRELLY
   IAU_CALLIRRHOE
   IAU_CALLISTO
   IAU_CALYPSO
   IAU_CARME
   IAU_CERES
   IAU_CHALDENE
   IAU_CHARON
   IAU_CORDELIA
   IAU_CRESSIDA
   IAU_DAVIDA
   IAU_DEIMOS
   IAU_DESDEMONA
   IAU_DESPINA
   IAU_DIDYMOS
   IAU_DIMORPHOS
   IAU_DIONE
   IAU_DONALDJOHANSON
   IAU_EARTH
   IAU_ELARA
   IAU_ENCELADUS
   IAU_EPIMETHEUS
   IAU_ERINOME
   IAU_EROS
   IAU_EUROPA
   IAU_EURYBATES
   IAU_GALATEA
   IAU_GANYMEDE
   IAU_GASPRA
   IAU_HARPALYKE
   IAU_HELENE
   IAU_HIMALIA
   IAU_HYDRA
   IAU_HYPERION
   IAU_IAPETUS
   IAU_IDA
   IAU_IO
   IAU_IOCASTE
   IAU_ISONOE
   IAU_ITOKAWA
   IAU_JANUS
   IAU_JULIET
   IAU_JUPITER
   IAU_KALYKE
   IAU_LARISSA
   IAU_LEDA
   IAU_LEUCUS
   IAU_LUTETIA
   IAU_LYSITHEA
   IAU_MAGACLITE
   IAU_MARS
   IAU_MENOETIUS
   IAU_MERCURY
   IAU_METIS
   IAU_MIMAS
   IAU_MIRANDA
   IAU_MOON
   IAU_NAIAD
   IAU_NEPTUNE
   IAU_NEREID
   IAU_NIX
   IAU_OBERON
   IAU_OPHELIA
   IAU_ORUS
   IAU_PALLAS
   IAU_PAN
   IAU_PANDORA
   IAU_PASIPHAE
   IAU_PATROCLUS
   IAU_PHOBOS
   IAU_PHOEBE
   IAU_PLUTO
   IAU_POLYMELE
   IAU_PORTIA
   IAU_PRAXIDIKE
   IAU_PROMETHEUS
   IAU_PROTEUS
   IAU_PUCK
   IAU_QUETA
   IAU_RHEA
   IAU_ROSALIND
   IAU_RYUGU
   IAU_SATURN
   IAU_SINOPE
   IAU_STEINS
   IAU_SUN
   IAU_TAYGETE
   IAU_TELESTO
   IAU_TEMPEL_1
   IAU_TETHYS
   IAU_THALASSA
   IAU_THEBE
   IAU_THEMISTO
   IAU_TITAN
   IAU_TITANIA
   IAU_TRITON
   IAU_UMBRIEL
   IAU_URANUS
   IAU_VENUS
   IAU_VESTA


Top

Appendix. High Precision Earth Fixed Frames





In addition to the text PCK based IAU body-fixed frame for Earth, 'IAU_EARTH', these two body-fixed frames for Earth are also ``built into'' the SPICE system:

   ITRF93
   EARTH_FIXED
'ITRF93' is a frame ``fixed'' to the Earth's crust. It provides a high precision model for the orientation of the Earth with respect to J2000. In SPICE this is also a PCK type frame but its orientation is provided in a binary PCK file.

'EARTH_FIXED' is a ``generic frame'' that gives the orientation of the Earth with respect to some other frame (usually 'IAU_EARTH' or 'ITRF93') via a constant rotational offset. Such frames are called Text Kernel (TK) frames. See the subsection `` Gaining Flexibility via TK Frames'' for a discussion of the use of TK frames.



Top

Appendix. Frame Identifiers Reserved for Earth Fixed Frames





NAIF has set aside a range of frame ID codes for Earth fixed frames to be added in the future when/if additional high precision Earth orientation model become available and are implemented in SPICE. This reserved range is from 13000 to 13999. The ID assigned to 'ITRF93', which is only currently implemented frame of this kind, is 13000. All of these frames are PCK based frames. They model the orientation of the Earth with respect to an inertial reference frame such as the J2000 frame. Since the primary customer of these frames is NASA's Deep Space Network (DSN), we shall refer to any frame with ID code in this reserved range as a DSN Earth Fixed frame or simply DSN frame.

The class ID to associate with any DSN frame is the frame ID minus 10000. For example, the class ID associated with frame 13003 is 3003. It is this class ID that should be placed in the PCK file that implements the relationship between the DSN frame and the corresponding inertial frame.

The center of any DSN frame is the center of mass of the Earth, which has SPK ID code 399.

These frames are partially ``built in''. Given a frame ID in the range from 13001 to 13999, the frame subsystem ``knows'' that the frame is a PCK frame, the center of the frame is 399 and the class ID of the frame is the frame ID - 10000. This knowledge cannot be overridden. However, the frame subsystem does not ``know'' the relationship between the names of these frames and their ID codes. The relationship must be specified via the appropriate kernel pool frame definition.

      FRAME_<name>              = <DSN Frame-ID>
      FRAME_<DSN Frame-ID>_NAME = '<name>'
      OBJECT_EARTH_FRAME        = <DSN Frame-ID>
Note that this specification leaves out the items below

   FRAME_<DSN Frame-ID>_CENTER   = 399
   FRAME_<DSN Frame-ID>_CLASS    = 2
   FRAME_<DSN Frame-ID>_CLASS_ID = <DSN Frame-ID  - 10000>
You may supply these values if you like, but they have no effect on the frame subsystem's recognition and interpretation of the frame with the specified frame ID.



Top

Appendix. Frame Definition Examples





Below are examples that you can modify to create frame specifications for similar situations.



Top

Inertial Frames




Inertial (class 1) frames cannot be defined in frame kernels; in particular, built-in definitions of class 1 frames cannot be overridden.

Aliases for inertial frames can be defined; see the section below on creating aliases using TK frames.



Top

PCK Frames




This definition shows how you create a frame definition for the asteroid Eros. Note we also define which frame is associated with the asteroid Eros.

   \begindata
 
      FRAME_EROS_FIXED       =  2000433
      FRAME_2000433_NAME     = 'EROS_FIXED'
      FRAME_2000433_CLASS    =  2
      FRAME_2000433_CLASS_ID =  2000433
      FRAME_2000433_CENTER   =  2000433
 
      OBJECT_2000433_FRAME   = 'EROS_FIXED'
 
   \begintext


Top

CK Frames




This definition shows how you create a frame definition for the MGS spacecraft. Note this frame definition includes the appropriate SCLK definition as well as which frame to attach to the MGS spacecraft.

   \begindata
 
      FRAME_MGS_SPACECRAFT   = -94000
      FRAME_-94000_NAME      = 'MGS_SPACECRAFT'
      FRAME_-94000_CLASS     =  3
      FRAME_-94000_CLASS_ID  = -94000
      FRAME_-94000_CENTER    = -94
 
      CK_-94000_SCLK         = -94
      CK_-94000_SPK          = -94
 
      OBJECT_-94_FRAME       = 'MGS_SPACECRAFT'
 
   \begintext


Top

TK frames




Examples of different uses of TK frames are shown below.



Top

TK frame --- Alias



This example shows how you can make up an alias for a frame using a TK frame. Note we make the reference frame to associate with Mars the MARS_FIXED frame.

   \begindata
 
      FRAME_MARS_FIXED       =  1400499
      FRAME_1400499_NAME     = 'MARS_FIXED'
      FRAME_1400499_CLASS    =  4
      FRAME_1400499_CLASS_ID =  1400499
      FRAME_1400499_CENTER   =  499
 
      OBJECT_499_FRAME       = 'MARS_FIXED'
 
   \begintext
 
   To make this point to another frame just replace
   'IAU_MARS' below with the name of that frame.
 
   \begindata
 
      TKFRAME_1400499_RELATIVE = 'IAU_MARS'
      TKFRAME_1400499_SPEC     = 'MATRIX'
      TKFRAME_1400499_MATRIX   = ( 1   0   0
                                   0   1   0
                                   0   0   1 )
   \begintext


Top

TK frame --- Topographic



This example shows how you could create a topographic frame for the DSN Station DSS-17.

   \begindata
 
      FRAME_DSS-17_TOPO      =  1399017
      FRAME_1399017_NAME     = 'DSS-17_TOPO'
      FRAME_1399017_CLASS    =  4
      FRAME_1399017_CLASS_ID =  1399017
      FRAME_1399017_CENTER   =   399017
 
      OBJECT_399017_FRAME    = 'DSS-17_TOPO'
 
   \begintext
 
   Note that the geodetic longitude and co-latitude of the DSS-17
   tracking station are: 243.126496675 and 54.657822839 respectively.
 
   \begindata
 
      TKFRAME_DSS-17_TOPO_RELATIVE = 'EARTH_FIXED'
      TKFRAME_DSS-17_TOPO_SPEC     = 'ANGLES'
      TKFRAME_DSS-17_TOPO_UNITS    = 'DEGREES'
      TKFRAME_DSS-17_TOPO_AXES     = ( 3, 2, 3 )
      TKFRAME_DSS-17_TOPO_ANGLES   = ( -243.126496675,
                                        -54.657822839,
                                        180.0 )
   \begintext
 
   Recall that the frame `EARTH_FIXED' is a TK frame. As a result
   its relationship to other frames must be specified via
   a kernel pool variable. We make that specification here.
 
   If the ITRF93 PCK kernel is not available we can simply rename the
   "RELATIVE" frame to be IAU_EARTH and still have the topocentric
   frame well defined.
 
   \begindata
 
      TKFRAME_EARTH_FIXED_RELATIVE = 'ITRF93'
      TKFRAME_EARTH_FIXED_SPEC     = 'MATRIX'
      TKFRAME_EARTH_FIXED_MATRIX   = ( 1   0   0
                                       0   1   0
                                       0   0   1 )
 
   \begintext


Top

TK frame --- Instrument



This example shows how you could create a TK frame for the Medium Resolution Imager (MRI) instrument on-board the Deep Impact Flyby (DIF) spacecraft.

The rotation from the DIF spacecraft frame to the MRI instrument frame determined from the in-flight calibration data can be represented by the following rotation angles:

    mri
   M    = |0.129539306414|  * |-45.006884881185|  * |0.004898709285|
    sc                    Z                     Y                   X
The frame definition contains the opposite of these rotation angles -- with the angle order reversed and the angle signs changed to the opposite ones -- because the angles specified in it define the transformation from the MRI frame to the spacecraft frame.

   \begindata
 
      FRAME_DIF_MRI             = -140200
      FRAME_-140200_NAME        = 'DIF_MRI'
      FRAME_-140200_CLASS       = 4
      FRAME_-140200_CLASS_ID    = -140200
      FRAME_-140200_CENTER      = -140
      TKFRAME_-140200_SPEC      = 'ANGLES'
      TKFRAME_-140200_RELATIVE  = 'DIF_SPACECRAFT'
      TKFRAME_-140200_ANGLES    = ( -0.004898709285,
                                    45.006884881185,
                                    -0.129539306414 )
      TKFRAME_-140200_AXES      = ( 1,    2,   3   )
      TKFRAME_-140200_UNITS     = 'DEGREES'
 
   \begintext


Top

Examples of Two-Vector Parameterized Dynamic Frames






Top

Geocentric Solar Ecliptic (GSE) Frame



Definition of the Geocentric Solar Ecliptic frame:

    All vectors are geometric: no aberration corrections are used.

    The position of the sun relative to the earth is the primary vector: the X axis points from the earth to the sun.

    The inertially referenced velocity of the sun relative to the earth is the secondary vector: the Y axis is the component of this velocity vector orthogonal to the X axis.

    The Z axis is X cross Y, completing the right-handed reference frame.

The GSE frame can be defined using the following assignments, where <frame_ID> must be replaced by an integer ID code.

   FRAME_GSE                       =  <frame_ID>
   FRAME_<frame_ID>_NAME           = 'GSE'
   FRAME_<frame_ID>_CLASS          =  5
   FRAME_<frame_ID>_CLASS_ID       =  <frame_ID>
   FRAME_<frame_ID>_CENTER         =  399
   FRAME_<frame_ID>_RELATIVE       = 'J2000'
   FRAME_<frame_ID>_DEF_STYLE      = 'PARAMETERIZED'
   FRAME_<frame_ID>_FAMILY         = 'TWO-VECTOR'
   FRAME_<frame_ID>_PRI_AXIS       = 'X'
   FRAME_<frame_ID>_PRI_VECTOR_DEF = 'OBSERVER_TARGET_POSITION'
   FRAME_<frame_ID>_PRI_OBSERVER   = 'EARTH'
   FRAME_<frame_ID>_PRI_TARGET     = 'SUN'
   FRAME_<frame_ID>_PRI_ABCORR     = 'NONE'
   FRAME_<frame_ID>_SEC_AXIS       = 'Y'
   FRAME_<frame_ID>_SEC_VECTOR_DEF = 'OBSERVER_TARGET_VELOCITY'
   FRAME_<frame_ID>_SEC_OBSERVER   = 'EARTH'
   FRAME_<frame_ID>_SEC_TARGET     = 'SUN'
   FRAME_<frame_ID>_SEC_ABCORR     = 'NONE'
   FRAME_<frame_ID>_SEC_FRAME      = 'J2000'


Top

Geocentric Solar Magnetospheric (GSM) Frame



Definition of the Geocentric Solar Magnetospheric frame:

    All vectors are geometric: no aberration corrections are used.

    The position of the sun relative to the earth is the primary vector: the X axis points from the earth to the sun.

    The earth's geomagnetic centered north dipole vector is secondary: the Z axis is the component of this vector orthogonal to the X axis. For the purpose of this definition, we treat the dipole vector as constant in the IAU_EARTH body-fixed frame. Note that in an earth-fixed reference frame, the north geomagnetic centered dipole is actually time-varying; the values shown here may be unsuitable for your application.

    The Y axis direction is the cross product of the Z-axis and the X-axis.

The GSM frame can be defined using the following assignments, where <frame_ID> must be replaced by an integer ID code.

   FRAME_GSM                       =  <frame_ID>
   FRAME_<frame_ID>_NAME           = 'GSM'
   FRAME_<frame_ID>_CLASS          =  5
   FRAME_<frame_ID>_CLASS_ID       =  <frame_ID>
   FRAME_<frame_ID>_CENTER         =  399
   FRAME_<frame_ID>_RELATIVE       = 'J2000'
   FRAME_<frame_ID>_DEF_STYLE      = 'PARAMETERIZED'
   FRAME_<frame_ID>_FAMILY         = 'TWO-VECTOR'
   FRAME_<frame_ID>_PRI_AXIS       = 'X'
   FRAME_<frame_ID>_PRI_VECTOR_DEF = 'OBSERVER_TARGET_POSITION'
   FRAME_<frame_ID>_PRI_OBSERVER   = 'EARTH'
   FRAME_<frame_ID>_PRI_TARGET     = 'SUN'
   FRAME_<frame_ID>_PRI_ABCORR     = 'NONE'
   FRAME_<frame_ID>_SEC_AXIS       = 'Z'
   FRAME_<frame_ID>_SEC_VECTOR_DEF = 'CONSTANT'
   FRAME_<frame_ID>_SEC_FRAME      = 'IAU_EARTH'
   FRAME_<frame_ID>_SEC_SPEC       = 'LATITUDINAL'
   FRAME_<frame_ID>_SEC_UNITS      = 'DEGREES'
   FRAME_<frame_ID>_SEC_LONGITUDE  =  288.43
   FRAME_<frame_ID>_SEC_LATITUDE   =   79.54


Top

Mercury Solar Equatorial (MSEQ) Frame



Definition of the Mercury Solar Equatorial Frame:

    All vectors are geometric: no aberration corrections are used.

    The sun's north spin axis direction is primary: the Z axis of the MSEQ frame is aligned with this spin axis.

    The position of the Sun relative to Mercury is secondary: the Y axis is aligned with the component of this position orthogonal to the Z axis.

    The X axis direction is the cross product of the Y axis and Z axis.

All vectors are geometric: no aberration corrections are used.

The MSEQ frame can be defined using the following assignments, where <frame_ID> must be replaced by an integer ID code.

   FRAME_MSEQ                      =  <frame_ID>
   FRAME_<frame_ID>_NAME           = 'MSEQ'
   FRAME_<frame_ID>_CLASS          =  5
   FRAME_<frame_ID>_CLASS_ID       =  <frame_ID>
   FRAME_<frame_ID>_CENTER         =  199
   FRAME_<frame_ID>_RELATIVE       = 'J2000'
   FRAME_<frame_ID>_DEF_STYLE      = 'PARAMETERIZED'
   FRAME_<frame_ID>_FAMILY         = 'TWO-VECTOR'
   FRAME_<frame_ID>_PRI_AXIS       = 'Z'
   FRAME_<frame_ID>_PRI_VECTOR_DEF = 'CONSTANT'
   FRAME_<frame_ID>_PRI_FRAME      = 'IAU_SUN'
   FRAME_<frame_ID>_PRI_SPEC       = 'RECTANGULAR'
   FRAME_<frame_ID>_PRI_VECTOR     =  ( 0, 0, 1 )
   FRAME_<frame_ID>_SEC_AXIS       = 'X'
   FRAME_<frame_ID>_SEC_VECTOR_DEF = 'OBSERVER_TARGET_POSITION'
   FRAME_<frame_ID>_SEC_OBSERVER   = 'MERCURY'
   FRAME_<frame_ID>_SEC_TARGET     = 'SUN'
   FRAME_<frame_ID>_SEC_ABCORR     = 'NONE'


Top

Example: Nadir Frame for Mars Orbiting Spacecraft



Definition of the nadir frame:

    All vectors are geometric: no aberration corrections are used.

    The Z axis points from the spacecraft to the closest point on Mars.

    The component of inertially referenced spacecraft velocity vector orthogonal to Z is aligned with the -X axis.

    The Y axis is the cross product of the Z axis and the X axis.

This nadir frame can be defined using the following assignments, where

<frame_name>

should be replaced by an actual frame name
<orbiter_ID>

must be replaced with the integer ID code of the orbiter
<orbiter_ID/name>

must be replaced with either the integer ID code of the orbiter or the name of the orbiter
<frame_ID>

must be replaced by an integer ID code
   FRAME_<frame_name>              = <frame_ID>
   FRAME_<frame_ID>_NAME           = <frame_name>
   FRAME_<frame_ID>_CLASS          = 5
   FRAME_<frame_ID>_CLASS_ID       = <frame_ID>
   FRAME_<frame_ID>_CENTER         = <orbiter_ID>
   FRAME_<frame_ID>_RELATIVE       = 'J2000'
   FRAME_<frame_ID>_DEF_STYLE      = 'PARAMETERIZED'
   FRAME_<frame_ID>_FAMILY         = 'TWO-VECTOR'
   FRAME_<frame_ID>_PRI_AXIS       = 'Z'
   FRAME_<frame_ID>_PRI_VECTOR_DEF = 'TARGET_NEAR_POINT'
   FRAME_<frame_ID>_PRI_OBSERVER   = <orbiter_ID/name>
   FRAME_<frame_ID>_PRI_TARGET     = 'MARS'
   FRAME_<frame_ID>_PRI_ABCORR     = 'NONE'
   FRAME_<frame_ID>_SEC_AXIS       = '-X'
   FRAME_<frame_ID>_SEC_VECTOR_DEF = 'OBSERVER_TARGET_VELOCITY'
   FRAME_<frame_ID>_SEC_OBSERVER   = <orbiter_ID/name>
   FRAME_<frame_ID>_SEC_TARGET     = 'MARS'
   FRAME_<frame_ID>_SEC_ABCORR     = 'NONE'
   FRAME_<frame_ID>_SEC_FRAME      = 'J2000'


Top

Example: Roll-Celestial Spacecraft Frame



There are a variety of roll-celestial frames in use. This example may not match frame definitions used for any specific flight project; it is intended to demonstrate how to define this category of frame.

Definition of the roll-celestial frame:

    The Z axis points from the spacecraft to the earth. This vector is geometric (uncorrected).

    The component of an inertially referenced star direction vector orthogonal to the Z axis is the X axis. The star direction is provided by a specified star catalog in the form of right ascension and declination relative to the J2000 frame. If necessary, the RA/DEC coordinates should be adjusted for proper motion and parallax. This star direction vector is corrected for stellar aberration using the spacecraft as the observer.

    The Y axis is the cross product of the Z axis and the X axis.

This roll-celestial frame can be defined using the following assignments, where

<frame_name>

should be replaced by an actual frame name
<spacecraft_ID>

must be replaced with the integer ID code of the spacecraft
<spacecraft_ID/name>

must be replaced with either the integer ID code of the spacecraft or the name of the spacecraft
<frame_ID>

must be replaced by an integer ID code
   FRAME_<frame_name>              = <frame_ID>
   FRAME_<frame_ID>_NAME           = <frame_name>
   FRAME_<frame_ID>_CLASS          = 5
   FRAME_<frame_ID>_CLASS_ID       = <frame_ID>
   FRAME_<frame_ID>_CENTER         = <spacecraft_ID>
   FRAME_<frame_ID>_RELATIVE       = 'J2000'
   FRAME_<frame_ID>_DEF_STYLE      = 'PARAMETERIZED'
   FRAME_<frame_ID>_FAMILY         = 'TWO-VECTOR'
   FRAME_<frame_ID>_PRI_AXIS       = 'Z'
   FRAME_<frame_ID>_PRI_VECTOR_DEF = 'OBSERVER_TARGET_POSITION'
   FRAME_<frame_ID>_PRI_OBSERVER   = <spacecraft_ID/name>
   FRAME_<frame_ID>_PRI_TARGET     = 'EARTH'
   FRAME_<frame_ID>_PRI_ABCORR     = 'NONE'
   FRAME_<frame_ID>_SEC_AXIS       = 'X'
   FRAME_<frame_ID>_SEC_VECTOR_DEF = 'CONSTANT'
   FRAME_<frame_ID>_SEC_FRAME      = 'J2000'
   FRAME_<frame_ID>_SEC_SPEC       = 'RA/DEC'
   FRAME_<frame_ID>_SEC_UNITS      = 'DEGREES'
   FRAME_<frame_ID>_SEC_RA         = <star RA in degrees>
   FRAME_<frame_ID>_SEC_DEC        = <star DEC in degrees>
   FRAME_<frame_ID>_SEC_ABCORR     = 'S'
   FRAME_<frame_ID>_SEC_OBSERVER   = <spacecraft_ID/name>


Top

Examples of Mean Equator and Equinox of Date Frames






Top

Earth Mean Equator and Equinox of Date Frames



Definition of a non-inertial Earth Mean Equator and Equinox of Date frame using 1976 IAU precession model. Here <frame_name> must be replaced by a string containing the name of the frame, and <frame_ID> must be replaced by an integer ID code:

   FRAME_<frame_name>             =  <frame_ID>
   FRAME_<frame_ID>_NAME          =  <frame_name>
   FRAME_<frame_ID>_CLASS         =  5
   FRAME_<frame_ID>_CLASS_ID      =  <frame_ID>
   FRAME_<frame_ID>_CENTER        =  399
   FRAME_<frame_ID>_RELATIVE      = 'J2000'
   FRAME_<frame_ID>_DEF_STYLE     = 'PARAMETERIZED'
   FRAME_<frame_ID>_FAMILY        = 'MEAN_EQUATOR_AND_EQUINOX_OF_DATE'
   FRAME_<frame_ID>_PREC_MODEL    = 'EARTH_IAU_1976'
   FRAME_<frame_ID>_ROTATION_STATE= 'ROTATING'
Definition for the inertial version of the above frame:

   FRAME_<frame_name>             =  <frame_ID>
   FRAME_<frame_ID>_NAME          =  <frame_name>
   FRAME_<frame_ID>_CLASS         =  5
   FRAME_<frame_ID>_CLASS_ID      =  <frame_ID>
   FRAME_<frame_ID>_CENTER        =  399
   FRAME_<frame_ID>_RELATIVE      = 'J2000'
   FRAME_<frame_ID>_DEF_STYLE     = 'PARAMETERIZED'
   FRAME_<frame_ID>_FAMILY        = 'MEAN_EQUATOR_AND_EQUINOX_OF_DATE'
   FRAME_<frame_ID>_PREC_MODEL    = 'EARTH_IAU_1976'
   FRAME_<frame_ID>_ROTATION_STATE= 'INERTIAL'
Definition for the frozen version of the above frame, where the ``freeze epoch'' is B1950 TDB. The resulting frame should match the inertial frame B1950 to round-off level:

   FRAME_<frame_name>             =  <frame_ID>
   FRAME_<frame_ID>_NAME          =  <frame_name>
   FRAME_<frame_ID>_CLASS         =  5
   FRAME_<frame_ID>_CLASS_ID      =  <frame_ID>
   FRAME_<frame_ID>_CENTER        =  399
   FRAME_<frame_ID>_RELATIVE      = 'J2000'
   FRAME_<frame_ID>_DEF_STYLE     = 'PARAMETERIZED'
   FRAME_<frame_ID>_FAMILY        = 'MEAN_EQUATOR_AND_EQUINOX_OF_DATE'
   FRAME_<frame_ID>_PREC_MODEL    = 'EARTH_IAU_1976'
   FRAME_<frame_ID>_FREEZE_EPOCH  = @1949-DEC-31/22:09:46.861901


Top

Examples of True Equator and Equinox of Date Frames




Definition of the Earth True Equator and Equinox of Date frame:

    The earth precession model is the 1976 IAU model.

    The earth nutation model is the 1980 IAU model.

Here <frame_name> must be replaced by a string containing the name of the frame, and <frame_ID> must be replaced by an integer ID code:

   FRAME_<frame_name>             =  <frame_ID>
   FRAME_<frame_ID>_NAME          =  <frame_name>
   FRAME_<frame_ID>_CLASS         =  5
   FRAME_<frame_ID>_CLASS_ID      =  <frame_ID>
   FRAME_<frame_ID>_CENTER        =  399
   FRAME_<frame_ID>_RELATIVE      = 'J2000'
   FRAME_<frame_ID>_DEF_STYLE     = 'PARAMETERIZED'
   FRAME_<frame_ID>_FAMILY        = 'TRUE_EQUATOR_AND_EQUINOX_OF_DATE'
   FRAME_<frame_ID>_PREC_MODEL    = 'EARTH_IAU_1976'
   FRAME_<frame_ID>_NUT_MODEL     = 'EARTH_IAU_1980'
   FRAME_<frame_ID>_ROTATION_STATE= 'ROTATING'
Definition for the inertial version of the above frame:

   FRAME_<frame_name>             =  <frame_ID>
   FRAME_<frame_ID>_NAME          =  <frame_name>
   FRAME_<frame_ID>_CLASS         =  5
   FRAME_<frame_ID>_CLASS_ID      =  <frame_ID>
   FRAME_<frame_ID>_CENTER        =  399
   FRAME_<frame_ID>_RELATIVE      = 'J2000'
   FRAME_<frame_ID>_DEF_STYLE     = 'PARAMETERIZED'
   FRAME_<frame_ID>_FAMILY        = 'TRUE_EQUATOR_AND_EQUINOX_OF_DATE'
   FRAME_<frame_ID>_PREC_MODEL    = 'EARTH_IAU_1976'
   FRAME_<frame_ID>_NUT_MODEL     = 'EARTH_IAU_1980'
   FRAME_<frame_ID>_ROTATION_STATE= 'INERTIAL'
Definition for the frozen version of the above frame, where the ``freeze epoch'' is B1950 TDB.

   FRAME_<frame_name>             =  <frame_ID>
   FRAME_<frame_ID>_NAME          =  <frame_name>
   FRAME_<frame_ID>_CLASS         =  5
   FRAME_<frame_ID>_CLASS_ID      =  <frame_ID>
   FRAME_<frame_ID>_CENTER        =  399
   FRAME_<frame_ID>_RELATIVE      = 'J2000'
   FRAME_<frame_ID>_DEF_STYLE     = 'PARAMETERIZED'
   FRAME_<frame_ID>_FAMILY        = 'TRUE_EQUATOR_AND_EQUINOX_OF_DATE'
   FRAME_<frame_ID>_PREC_MODEL    = 'EARTH_IAU_1976'
   FRAME_<frame_ID>_NUT_MODEL     = 'EARTH_IAU_1980'
   FRAME_<frame_ID>_FREEZE_EPOCH  = @1949-DEC-31/22:09:46.861901


Top

Example of a Mean Ecliptic and Equinox of Date Frame




Definition of the Earth Mean Ecliptic and Equinox of Date frame:

    The earth precession model is the 1976 IAU model.

    The earth mean obliquity model is the 1980 IAU model.

Here <frame_name> must be replaced by a string containing the name of the frame, and <frame_ID> must be replaced by an integer ID code:

   FRAME_<frame_name>             =  <frame_ID>
   FRAME_<frame_ID>_NAME          =  <frame_name>
   FRAME_<frame_ID>_CLASS         =  5
   FRAME_<frame_ID>_CLASS_ID      =  <frame_ID>
   FRAME_<frame_ID>_CENTER        =  399
   FRAME_<frame_ID>_RELATIVE      = 'J2000'
   FRAME_<frame_ID>_DEF_STYLE     = 'PARAMETERIZED'
   FRAME_<frame_ID>_FAMILY        = 'MEAN_ECLIPTIC_AND_EQUINOX_OF_DATE'
   FRAME_<frame_ID>_PREC_MODEL    = 'EARTH_IAU_1976'
   FRAME_<frame_ID>_OBLIQ_MODEL   = 'EARTH_IAU_1980'
   FRAME_<frame_ID>_ROTATION_STATE= 'ROTATING'
Definition for the inertial version of the above frame:

   FRAME_<frame_name>             =  <frame_ID>
   FRAME_<frame_ID>_NAME          =  <frame_name>
   FRAME_<frame_ID>_CLASS         =  5
   FRAME_<frame_ID>_CLASS_ID      =  <frame_ID>
   FRAME_<frame_ID>_CENTER        =  399
   FRAME_<frame_ID>_RELATIVE      = 'J2000'
   FRAME_<frame_ID>_DEF_STYLE     = 'PARAMETERIZED'
   FRAME_<frame_ID>_FAMILY        = 'MEAN_ECLIPTIC_AND_EQUINOX_OF_DATE'
   FRAME_<frame_ID>_PREC_MODEL    = 'EARTH_IAU_1976'
   FRAME_<frame_ID>_OBLIQ_MODEL   = 'EARTH_IAU_1980'
   FRAME_<frame_ID>_ROTATION_STATE= 'INERTIAL'
Definition for the frozen version of the above frame, where the ``freeze epoch'' is B1950 TDB.

   FRAME_<frame_name>             =  <frame_ID>
   FRAME_<frame_ID>_NAME          =  <frame_name>
   FRAME_<frame_ID>_CLASS         =  5
   FRAME_<frame_ID>_CLASS_ID      =  <frame_ID>
   FRAME_<frame_ID>_CENTER        =  399
   FRAME_<frame_ID>_RELATIVE      = 'J2000'
   FRAME_<frame_ID>_DEF_STYLE     = 'PARAMETERIZED'
   FRAME_<frame_ID>_FAMILY        = 'MEAN_ECLIPTIC_AND_EQUINOX_OF_DATE'
   FRAME_<frame_ID>_PREC_MODEL    = 'EARTH_IAU_1976'
   FRAME_<frame_ID>_OBLIQ_MODEL   = 'EARTH_IAU_1980'
   FRAME_<frame_ID>_FREEZE_EPOCH  = @1949-DEC-31/22:09:46.861901


Top

Example of an Euler Frame




As an example, we construct an Euler frame called IAU_MARS_EULER. Frame IAU_MARS_EULER is mathematically identical to the PCK frame IAU_MARS.

The PCK data defining the underlying IAU_MARS frame are:

   BODY499_POLE_RA          = (  317.68143   -0.1061      0.  )
   BODY499_POLE_DEC         = (   52.88650   -0.0609      0.  )
   BODY499_PM               = (  176.630    350.89198226  0.  )
These values are from:

   Seidelmann, P.K., Abalakin, V.K., Bursa, M., Davies, M.E., Bergh, C.
   de, Lieske, J.H., Oberst, J., Simon, J.L., Standish, E.M.,
   Stooke, P., and Thomas, P.C. (2002). "Report of the IAU/IAG Working
   Group on Cartographic Coordinates and Rotational Elements of the
   Planets and Satellites: 2000," Celestial Mechanics and Dynamical
   Astronomy, v.82, Issue 1, pp. 83-111.
Here pole RA/Dec terms in the PCK are in degrees and degrees/century; the rates here have been converted to degrees/sec. Prime meridian terms in the PCK are in degrees and degrees/day; the rate here has been converted to degrees/sec.

The 3x3 transformation matrix M defined by the angles is

   M = [angle_1]   [angle_2]   [angle_3]
                3           1           3
Vectors are mapped from the J2000 base frame to the IAU_MARS frame via left multiplication by M.

The relationship of these Euler angles to RA/Dec/PM for the J2000-to-IAU Mars body-fixed transformation is as follows:

   angle_1 is        PM  * (radians/degree)
   angle_2 is pi/2 - Dec * (radians/degree)
   angle_3 is pi/2 + RA  * (radians/degree)
Since when we define the IAU_MARS_EULER frame we're defining the *inverse* of the above transformation, the angles for our Euler frame definition are reversed and the signs negated:

   angle_1 is -pi/2 - RA  * (radians/degree)
   angle_2 is -pi/2 + Dec * (radians/degree)
   angle_3 is       - PM  * (radians/degree)
Then our frame definition is:

   FRAME_IAU_MARS_EULER            =  <frame_ID>
   FRAME_<frame_ID>_NAME           = 'IAU_MARS_EULER'
   FRAME_<frame_ID>_CLASS          =  5
   FRAME_<frame_ID>_CLASS_ID       =  <frame_ID>
   FRAME_<frame_ID>_CENTER         =  499
   FRAME_<frame_ID>_RELATIVE       = 'J2000'
   FRAME_<frame_ID>_DEF_STYLE      = 'PARAMETERIZED'
   FRAME_<frame_ID>_FAMILY         = 'EULER'
   FRAME_<frame_ID>_EPOCH          =  @2000-JAN-1/12:00:00
   FRAME_<frame_ID>_AXES           =  ( 3  1  3 )
   FRAME_<frame_ID>_UNITS          = 'DEGREES'
   FRAME_<frame_ID>_ANGLE_1_COEFFS = (  -47.68143
                                          0.33621061170684714E-10 )
   FRAME_<frame_ID>_ANGLE_2_COEFFS = (  -37.1135
                                         -0.19298045478743630E-10 )
   FRAME_<frame_ID>_ANGLE_3_COEFFS = ( -176.630
                                         -0.40612497946759260E-02 )


Top

Examples of Product Frames






Top

IAU_EARTH Frame, Augmented with Nutation Model



The example shown here is not realistic; it is provided only to show how to create a product frame specification.

The EARTH_ROTATING frame defined uses the Earth spin angle relative to the mean equinox of date from the IAU_EARTH frame and pole and equinox from the Earth true equator and true equinox of date frame. While the pole direction of the IAU_EARTH frame reflects precession, the pole direction of this frame reflects both precession and nutation.

The transformation from the J2000 frame to the product frame is defined by:

     EARTH_ROTATING       IAU_EARTH       TETE
   T                =   T            *  T
     J2000                EME             J2000
where the notation

     B
   T
     A
indicates the transformation from frame A to frame B.

The specifications of the frame and of the two supporting frames EME and TETE are shown below.

   \begindata
 
   FRAME_EARTH_ROTATING          =  1890000
   FRAME_1890000_NAME            =  'EARTH_ROTATING'
   FRAME_1890000_CLASS           =  5
   FRAME_1890000_CLASS_ID        =  1890000
   FRAME_1890000_CENTER          =  399
   FRAME_1890000_RELATIVE        = 'J2000'
   FRAME_1890000_DEF_STYLE       = 'PARAMETERIZED'
   FRAME_1890000_FAMILY          = 'PRODUCT'
   FRAME_1890000_ROTATION_STATE  = 'ROTATING'
   FRAME_1890000_FROM_FRAMES     = ( 'EME',       'J2000' )
   FRAME_1890000_TO_FRAMES       = ( 'IAU_EARTH', 'TETE'  )
 
   \begintext
 
   Earth mean equator and mean equinox of date frame "EME":
 
   \begindata
 
   FRAME_EME                     =  1890001
   FRAME_1890001_NAME            =  'EME'
   FRAME_1890001_CLASS           =  5
   FRAME_1890001_CLASS_ID        =  1890001
   FRAME_1890001_CENTER          =  399
   FRAME_1890001_RELATIVE        = 'J2000'
   FRAME_1890001_DEF_STYLE       = 'PARAMETERIZED'
   FRAME_1890001_FAMILY          = 'MEAN_EQUATOR_AND_EQUINOX_OF_DATE'
   FRAME_1890001_PREC_MODEL      = 'EARTH_IAU_1976'
   FRAME_1890001_ROTATION_STATE  = 'ROTATING'
 
   \begintext
 
   Earth true equator and true equinox of date frame "TETE":
 
   \begindata
 
   FRAME_TETE                    =  1890002
   FRAME_1890002_NAME            =  'TETE'
   FRAME_1890002_CLASS           =  5
   FRAME_1890002_CLASS_ID        =  1890002
   FRAME_1890002_CENTER          =  399
   FRAME_1890002_RELATIVE        = 'J2000'
   FRAME_1890002_DEF_STYLE       = 'PARAMETERIZED'
   FRAME_1890002_FAMILY          = 'TRUE_EQUATOR_AND_EQUINOX_OF_DATE'
   FRAME_1890002_PREC_MODEL      = 'EARTH_IAU_1976'
   FRAME_1890002_NUT_MODEL       = 'EARTH_IAU_1980'
   FRAME_1890002_ROTATION_STATE  = 'ROTATING'
 
   \begintext
In order for this frame to be generally useful, a more accurate Earth spin model than that provided by the IAU_EARTH frame would need to be used. In practice, high-precision binary Earth PCKs are more suitable as a source of accurate Earth orientation data.



Top

Dog-Leg Frame for Saturn



A ``Dog-Leg'' frame for Saturn is a realistic application of product frames. The specification of this frame is quite complex. Contact NAIF for details.



Top

Examples of Switch Frames




In the examples below, all frame names and ID codes are fictitious.



Top

Switch Frame Using Reconstructed and Predict CKs



In this example, a switch frame uses distinct CK base frames for reconstructed and predicted data. The two CK frames use different spacecraft clocks.

The specification for such a switch frame would have the form:

   \begindata
 
   FRAME_SWITCH1              = -123001
   FRAME_-123001_NAME         = 'SWITCH1'
   FRAME_-123001_CLASS        = 6
   FRAME_-123001_CLASS_ID     = -123001
   FRAME_-123001_CENTER       = -123
   FRAME_-123001_ALIGNED_WITH = (
                                   'CK_PREDICTED'
                                   'CK_RECONSTRUCTED'
                                )
   \begintext
The base frames, which are specified by the assignment of the kernel variable FRAME_-123001_ALIGNED_WITH, are listed in order of increasing priority: given a request time, the SWITCH1 frame first tries to get orientation of the frame CK_RECONSTRUCTED; if not found, it tries to get orientation of the frame CK_PREDICTED.

The CK frames referenced by this switch frame must be defined, and the corresponding CKs loaded, along with associated SCLK kernels and a leapseconds kernel, for the switch frame to be usable.

Loading the CKs without loading an FK that defines the CK frames would not work. Examples of the CK frame definitions are shown below.

Reconstructed attitude CK frame:

   \begindata
 
   FRAME_CK_RECONSTRUCTED = -123501
   FRAME_-123501_NAME     = 'CK_RECONSTRUCTED'
   FRAME_-123501_CLASS    = 3
   FRAME_-123501_CLASS_ID = -123601
   FRAME_-123501_CENTER   = -123
   CK_-123501_SCLK        = -123
 
   \begintext
Predicted attitude CK frame:

   \begindata
 
   FRAME_CK_PREDICTED     = -123502
   FRAME_-123502_NAME     = 'CK_PREDICTED'
   FRAME_-123502_CLASS    = 3
   FRAME_-123502_CLASS_ID = -123602
   FRAME_-123502_CENTER   = -123
   CK_-123502_SCLK        = -123000
 
   \begintext
Base frames may also be specified by frame ID, so the ``aligned with'' assignment may be written as

   \begindata
 
   FRAME_-123001_ALIGNED_WITH = (
                                   -123502
                                   -123501
                                )
   \begintext
Note that the frame ID of a CK frame might not match the frame's frame class ID, which is the ID used in CKs providing data for that frame. In this case, the IDs used in reconstructed and predicted CKs would be -123601 and -123602 respectively. Using those IDs in the ``aligned with'' assignment would not work.



Top

Switch Frame Using CK and Dynamic Frames



This switch frame uses CK frames for reconstructed and predicted data as in the previous example, and it uses dynamic frames providing nominal attitude for the cruise and orbit mission phases, for times when no CK data are available.

This example is for a seven year long mission:

   launch          -- 2018-01-01
   orbit insertion -- 2018-10-01
   end of mission  -- 2025-01-01
The CK frames are applicable for the entire mission.

The nominal cruise attitude is implemented by the dynamic frame DYN_CRUISE (definition not shown). This frame is applicable only for the cruise phase of the mission.

The nominal orbital attitude is implemented by the dynamic frame DYN_ORBIT (definition not shown). This frame is applicable only for the orbital phase of the mission.

The CK frames are listed last in the set of base frame names, so they have highest priority.

Start and stop times below are expressed in text kernel format. The times are interpreted as TDB calendar dates.

   \begindata
 
   FRAME_SWITCH2              = -123002
   FRAME_-123002_NAME         = 'SWITCH2'
   FRAME_-123002_CLASS        = 6
   FRAME_-123002_CLASS_ID     = -123001
   FRAME_-123002_CENTER       = -123
 
   FRAME_-123002_ALIGNED_WITH = (
                                   'DYN_CRUISE'
                                   'DYN_ORBIT'
                                   'CK_PREDICTED'
                                   'CK_RECONSTRUCTED'
                                )
 
   FRAME_-123002_START        = (
                                   @2018-01-01
                                   @2018-10-01
                                   @2018-01-01
                                   @2018-01-01
                                )
 
   FRAME_-123002_STOP         = (
                                   @2018-10-01
                                   @2025-01-01
                                   @2025-01-01
                                   @2025-01-01
                                )
   \begintext
Time strings recognized by the CSPICE function str2et_c also may be used. We could define the interval start times using the assignment

   \begindata
 
   FRAME_-123002_START        = (
                                   '2018 JAN 1 00:00:00.000 TDB'
                                   '2018 OCT 1 00:00:00.000 TDB'
                                   '2018 JAN 1 00:00:00.000 TDB'
                                   '2018 JAN 1 00:00:00.000 TDB'
                                )
   \begintext


Top

Predicted Attitude Profile for Observation Planning



In this example, a sequence of base frames provides nominal predicted pointing, over a short time period, for a spacecraft observing a target object.

This example is for two six-hour long orbits, each broken into equal chunks for Sun-pointing, Earth-pointing, and observation modes. Pointing transitions are abrupt: at each transition time, the switch frame instantaneously changes its orientation from that of one base frame that of the next.

The nominal Sun-pointing attitude is implemented by the dynamic frame DYN_SUN_POINTING (definition not shown).

The nominal Earth-pointing attitude is implemented by the dynamic frame DYN_EARTH_POINTING (definition not shown).

The nominal observation attitude is implemented by the dynamic frame DYN_OBSERVATION (definition not shown).

   \begindata
 
      FRAME_SWITCH3              = -123003
      FRAME_-123003_NAME         = 'SWITCH3'
      FRAME_-123003_CLASS        = 6
      FRAME_-123003_CLASS_ID     = -123003
      FRAME_-123003_CENTER       = -123
      FRAME_-123003_ALIGNED_WITH = (
                                      'DYN_SUN_POINTING'
                                      'DYN_EARTH_POINTING'
                                      'DYN_OBSERVATION'
                                      'DYN_SUN_POINTING'
                                      'DYN_EARTH_POINTING'
                                      'DYN_OBSERVATION'
                                   )
 
      FRAME_-123003_START        = (
                                      @2018-01-01/00:00:00
                                      @2018-01-01/02:00:00
                                      @2018-01-01/04:00:00
                                      @2018-01-01/06:00:00
                                      @2018-01-01/08:00:00
                                      @2018-01-01/10:00:00
                                   )
 
      FRAME_-123003_STOP         = (
                                      @2018-01-01/02:00:00
                                      @2018-01-01/04:00:00
                                      @2018-01-01/06:00:00
                                      @2018-01-01/08:00:00
                                      @2018-01-01/10:00:00
                                      @2018-01-01/12:00:00
                                   )
   \begintext
Because the time intervals associated with the base frames are listed in increasing time order and overlap only at their endpoints, request times will be mapped to time intervals by binary search. If the time intervals were listed in any other order, a linear search would be used.