create new tag
view all tags

Status of standard pipeline pointing products: ground-based improvements (SPG 13.0 and beyond)

Scope of applicability

This page refers to the improved pointing information provided in products created in SPG 13.0 and later, derived from gyro-reconstructed attitude data. For data products created with SPG versions before 13.0, please refer to the page on improved pointing information from "simple" ground-based processing.

For known issues and pointing-related software updates, please refer to the pages: HIPE Known Issues, Data Products' Known Issues, and What's New in HIPE.

Background information

The accuracy of the Herschel's attitude measurements was notably improved along the operational phase of the mission, and further improved on-ground (see the pointing calibration sections in this web), but the attitude measurements derived on-board and down-linked (the so-called filtered attitude) were too heavily filtered and therefore resulted in a poor estimation of the high-frequency changes in the spacecraft attitude (the spacecraft pointing jitter/stability). The "gyro-based attitude reconstruction" method was devised to overcome this limitation. It combines the star tracker attitude measurements with the output from the Inertial Reference Unit (gyroscopes).

As the name implies, this method places much greater weight on the measurements made by the gyros, using attitude measurements constructed from star tracker data to provide an absolute reference and to account for gyro drifts. Although it has been demonstrated that this method is able to successfully reduce the high-frequency components of the Absolute Measurement Error (AME) it neverthess suffers from a couple of drawbacks (when compared with a conventional attitude estimator) related to limitations in the calibration of gyro biases and drift rates.

Software based on the gyro-based attitude reconstruction method was developed initially at the PACS ICC. The design and first prototype was implemented by H. Feuchtgruber in IDL and ported to Jython and integrated as toolbox package in HIPE 12.0, by B. Vandenbussche. Further refinements were included at the HSC and eventually ported to Java for inclusion in the Standard Product Generation (SPG) pipeline.

A users' guide including a description of the software algorithms can be found here.

Pointing information provided

The S/C attitude information is provided within the pointing product. The most relevant columns regarding the gyro-reconstructed attitude are:

  1. Filtered attitude quaternion (filterQuat): the attitude used by default to derive the astrometry. From SPG 13.0 onwards, contains the output of the gyro-based reconstruction ground-based processing.
  2. Simple-corrected filtered quaternion (simpleCorrFilterQuat): the attitude processed on-ground using the "simple-correction" algorithms.
  3. Uncorrected filtered quaternion (uncorrFilterQuat): the original filtered attitude computed on-board and down-linked in telemetry.
  4. Gyro attitude probability-X: (gyroAttProbX). Probability of such a large value of the minimized cost function (x-axis) occurring at random.
  5. Gyro attitude probability-Y: (gyroAttProbY). Idem for y-axis.
  6. Gyro attitude probability-Z: (gyroAttProbZ). Idem for z-axis.
  7. Attitude sigma-X (gyroAttSigmaX). Standard deviation of error about x-axis.
  8. Attitude sigma-Y (gyroAttSigmaY). Idem for y-axis.
  9. Attitude sigma-Y (gyroAttSigmaY). Idem for z-axis.

In addition, there are several meta-data keywords recording the parameters used in the generation of the gyro-reconstructed attitude, i.e. the parameters passed to the main task calcAttitude. Finally, two meta-data keywords have been included to ease the evaluation of the quality of the gyro- reconstructed attitude: gyroAttQuality and gyroAttCoverage.

The definitions of these two proxies of the gyro-reconstructed attitude quality are as follows:

  1. gyroAttQuality: it is computed as 1 − Nbad/Ngyr, where Nbad is the number of “bad samples”, such as the product gyroAttProbX[i] * gyroAttProbY[i] * gyroAttProbZ[i] is less than certain threshold value probThreshold (gyroAttProbX(Y,Z) is the probability that “the fit is good” in each axis, i.e. that min{χ2} > obtained value) and Ngyr is the total number of samples used for the computation.
  2. gyroAttCoverage: gyro-reconstructed attitude coverage as given by the fraction Ngyr/Ntot, where Ngyr is the number of samples used for the computation and Ntot is the total number of samples.

The "suspicious gyro attitude" flag can be activated on the condition:

if ((gyroAttQuality < probBad) or (gyroAttCoverage < coverageThresh)):  
  gyroAttSuspicious = 1

where probBad= is the recommended threshold value for gyroAttQuality= and coverageThresh is the recommended threshold for gyroAttCoverage. These parameters are also included as meta-data keywords in the pointing product. Sensible values are =probThreshold = 0.0001 and =probBad = 0.7.

Performance of the gyro-based ground processing.

The results of the tests carried out at the HSC can be found here.

-- MiguelSanchez - 08 Apr 2015

-- CraigStephenson - 08 Jul 2015

-- CraigStephenson - 04 Jan 2016

-- Miguel Sanchez - 2016-08-03

Edit | Attach | Watch | Print version | History: r7 < r6 < r5 < r4 < r3 | Backlinks | Raw View | Raw edit | More topic actions
Topic revision: r7 - 2016-08-03 - MiguelSanchez
This site is powered by the TWiki collaboration platform Powered by Perl