NOAA KLM User's Guide Appendix M

Introduction Page, NOAA KLM TOC, Acronyms
Previous Section, Next Section

APPENDIX M: Implementation of AMSU-B Correction Algorithm

The NOAA-15 S-band antenna anomaly was attributed to a mismatch of materials. The STX-1, STX-2, and STX-3 high gain antennas were fabricated using a non-conductive center core with four conductive strips coiled around it. A metal transition tab was attached to an end of each strip and is used to complete the antenna's electrical path. Regrettably, the metal transition tab and the center core had very different thermal expansion properties. As the STX antennas went in and out of sun conditions the material mismatch condition caused these tabs to experience a small movement. These movements introduced stresses and the tabs eventually cracked due to metal fatigue. The antenna's electrical path was then compromised.

The lost NOAA-15 STX down link capability was successfully recovered through the use of an operational work-around (as described in this appendix). The satellite's data sources were switched to use differently designed antennas, the STX-2 omni and the STX-4 omni.

A corrective redesign to future NOAA satellite antennas was performed. Starting with NOAA-16 the antennas now include stress relief in critical areas and a change of materials to better match the thermal expansion properties. A flight quality antenna was fabricated and subjected to 3675 thermal cycles. No electrical performance degradation was noted. The tested antenna was then disassembled, inspected, and the design was confirmed for flight.

The AMSU-B instruments on NOAA-16 and NOAA-17 were modified before launch to augment the radio frequency shielding within the receiver. Therefore their susceptibility to interference is much reduced compared with NOAA-15.

For NOAA-16 any interference effects are below the noise level of the instrument, so no correction tables are needed.

For NOAA-17 there is low-level interference on Channels 18 and 19. These biases are associated with STX-2 and STX-3. The maximum Earth view bias is 2K for Channel 19 and 1K for Channel 18.

The biases were characterized during the post-launch checkout period, and correction tables issued. Because the spacecraft transmitters are stable there is no need for any additional correction tables or bias detection schemes.

This appendix is an attempt to document the bias characteristics of the AMSU-B instrument for the NOAA KLM spacecraft. The changes are ordered chronologically in this appendix, beginning with the most recent to the earliest changes (note that the tables are numbered in reverse order because of this arrangement).

• Appendix M.1 Discussion
• Appendix M.2 Information to be Stored in the Header
• Appendix M.3 Additional Tasks for the Level 1a - Level 1b Calibration Program
• Appendix M.4 Additional Tasks for the Level 1b User
• Appendix M.5 Derivation of Correction Data
• Appendix M.6 Interpolation Methods
• Appendix M.7 FORTRAN Examples
• Appendix M.8 Location of Transmitter Power Data
• Appendix M.9 Bias Correction Tables
• Appendix M.10 Implementation of New AMSU-B Bias Correction Changes (after October 13, 1998)

August 13, 2004

RFI corrections for the NOAA-15 SARR transmitter were updated in August of 2004. Corrections for the other transmitters remain unchanged. Count corrections for Earth views 1,5,10...90, Space view (91) and Target view (92).

July 12, 2002

March 15, 2001

An AMSU-B trending test was conducted on 3 March 2001. Channels 19 and 20 of the SARR table were affected, as well as Channel 19 of the STX-2 omni table. Transmitters STX-2 omni, STX-4 omni and SARR(A) are active. Count corrections for Earth views 1,5,10...90, Space view (91) and Target view (92).

December 8, 2000

An AMSU-B trending test was conducted on 4 December 2000 and was based on internal consistency and comparisons with NOAA-16. This revision includes corrections for anomalous bias, revised following trending tests of 3-4 November 2000 and updated for November drift. Only the channel 19 Earth-view SARR table was affected. The STX-2 omni antenna was substituted for the STX-2 high-gain antenna. Transmitters STX1, STX2, STX3 and SARR(A) are active. Counts corrections for Earth views 1,5,10...90, Space view (91) and Target view (92).

November 9, 2000

An AMSU-B trending test was conducted on 3-4 November 2000. Results were as expected, i.e. a small change in the Channel 19 SARR interference, and otherwise no significant change. Results agree well with those obtained by other methods, including comparison with NOAA-16.

The only change in the new bias correction is the SARR table for channel 19. Changes range from -0.8K (around pixel 50) to +1.7K (pixel 20). The last update was issued in June 2000.

The STX-2 omni antenna was substituted for STX-2 high-gain. Transmitters STX1, STX2, STX3 and SARR(A) are active. Count corrections for Earth views 1, 5, 10...90, space view (91) and Target view (92).

July 10, 2000

An AMSU-B bias trending test was conducted on 27-28 June 2000. Results of this test, plus monitoring of daily mean brightness temperatures and NWP model comparisons, show that an update to the bias correction tables is required. The last update was issued in May 2000. Channel 19 bias due to SARR interference, and to a lesser extent channel 20, is continuing to drift at up to 1K per month, depending on scan position. This does not appear to be directly related to temperature change, or any other external factor.

This time only the SARR table has been updated. STX-2 has not changed.

AMSU-B Bias corrections, Version 2.0, 10 July 2000

STX-2 omni was substituted for STX-2 high-gain per Nigel Atkinson (UKMO). The other active transmitters include STX1, STX2, STX3 and SARR(A). Counts corrections for Earth views 1, 5, 10...90, space view (91) and Target view (92).

May 4, 2000

An AMSU-B bias trending test was conducted on 26-27 April 2000. Results of this test, plus monitoring of daily mean brightness temperatures, show that a minor update to the bias correction tables is required. The last update was issued in January 2000. Only the STX-2 and SARR tables have been updated.

Differences are as follows:

Channel 16: None

Channel 17: Maximum change 1.3K, mainly affecting pixels 1-30. This is probably not an instrument change but rather an improvement in the characterization method. Globally-averaged brightness temperatures were showing unrealistic structure between pixels 1-30, which has now been removed.

Channel 18: None

Channel 19: Maximum change 1.7K at pixels 1-10. This is real, and is probably associated with a change in thermal forcing. The bias pattern changed slightly around 13th April, coinciding with a change from falling instrument temperatures to rising. Note that channel 19 shows the largest EMI effects and is therefore the most sensitive to environment changes.

Channel 20: None

AMSU-B Bias corrections, Version 1.9, 4 May 2000 includes corrections for anomalous bias.

Revised following trending tests of 26-27 April 2000. STX-2 Omni substituted for STX-2 high-gain per Nigel Atkinson (UKMO). The other active transmitters include STX-4 Omni and SARR. The STX-4 Omni does not appear to cause any interference, so there is no table for it.

January 10, 2000:

Revised following trending tests of 23 December 1999. STX-2 Omni antenna substituted for STX-2 high gain antenna. Transmitters STX-2 Omni, STX-4 and SARR(A) are operational per Nigel Atkinson (UKMO). It should be noted that there is no correction table for STX-4 because it does not contribute to the interference with the AMSU-B instrument.

January 4, 2000:

After re-configuring the STX antennas on NOAA-15, the previous AMSU-B bias has been minimized and stabilized. The monitoring of this new configuration since September 28, 1999 has shown little change in the bias. Therefore, it has been decided to declare the NOAA-15 1b, 1b* AMSU-B Level 1b an operational product as of January 6, 2000. Bias correction updates will still be implemented to perform bias corrections as deemed necessary by the instrument scientists.

December 1, 1999:

Updated values for the antenna bias will be implemented for NOAA-15. The changes are in response to the request to update the AMSU-B RFI Correction tables with data submitted by the UK Meteorological Office (UKMO) on Nov. 12, 1999. The first orbits to be processed will be:

NSS.HRPT.NK.D99335.S1709.E1721.B0806464.GC scheduled to be ingested at 12:09 pm local time and NSS.GHRR.NK.D99335.S1535.E1707.B0806364.GC scheduled to be ingested at 12:23 pm local time.

November 24, 1999:

The following updates to the AMSU-B RFI bias corrections are based on the October 1999 trending tests (performed by the UKMO).

These are the following differences compared with the previous version:

1) Only STX-2 and SARR have changed (the other transmitters are not used operationally now, so the old values remain).

2) The largest change is for SARR in channel 19. Maximum change=3K. Would not expect any obvious difference in imagery products, but the new version gives a much flatter scan-dependence when compared with NWP model background.

3) Have left channel 18 completely unchanged. Changes in channels 17 and 20 are less than 1K.

4) Channel 16 uses a new technique involving AMSU-A to remove the unwanted effects of scene variation, the RFI corrections for this channel should be more accurate than before (probably about 0.2K). Other channels should be within 1K. Channel 19 is the one most likely to drift with time.

For RFI corrections, the users don't have to use the values in the Level 1b header if they don't want to.

November 11, 1999:

AMSU-B Bias corrections, Version 1.7, includes corrections for anomalous bias revised following trending tests of Oct 1999. STX-2 omni substituted for STX-2 high-gain. Based on October 1999 data. Transmitters STX1, STX2, STX3 and SARR(A) counts corrections for Earth views 1,5,10...90, Space view (91) and Target view (92). Mean transmitter power in counts, ID=4.

November 4, 1999:

Since the launch of NOAA-15, the three high-gain antennas connected to three of the four NOAA-15 transmitters (STX), specifically STX-1, STX-2 and STX-3, have experienced increasing performance degradation. In addition, STX-1 and STX-3 radio frequency interference is being received by the AMSU-B instrument. The STX-1 high-gain HRPT antenna (1698.0 Mhz) is degraded to a level where small dish (1 m) users experience a significant number of HRPT dropouts, in many cases rendering the data unusable.

NOAA/NESDIS has reconfigured the spacecraft in an effort to maintain many of the missions of NOAA-15; i.e., readout of the full stored GAC data, HRPT direct readout (which also includes the HIRS and AMSU data), and HRPT Level 1b acquisitions at the NOAA CDAs.

At 0100 UTC on 28 September 1999, NOAA/NESDIS moved the HRPT service from STX-1 to the STX-2 OMNI antenna (1702.5 MHz). This transmitter/antenna combination has been tested with small dish receiving stations, and has been found to provide satisfactory reception under most conditions. (The EIRP for the STX-2 OMNI is equal to or greater than 24 dBm over 90% of a sphere). The STX-2 transmissions are Right Hand Circularly polarized and compatible with existing HRPT antenna systems. This change is permanent. APT service is not affected.

Because of these antenna interference problems, all Level 1b AMSU-B data from NOAA-15 received between launch and Sept. 28, 1999 should be considered unusable. A new bias correction table for the STX-2 OMNI antenna was included in the NOAA-K CPIDS AMSU-B preprocessor since Sept. 15, 1999, and is therefore included in the AMSU-B data set header record. These bias corrections must be applied by the user.

September 28, 1999:

Table M.9-2 gives the bias corrections for AMSU-B for transmitters: STX2 Omni, STX3 and SARR(A). The corrections (in counts) are provided for Earth views 1, 5, 10...90, Space view (91) and Target view (92) and the mean transmitter power is also given.

January 4, 1999:

The S band transmitters on NOAA-K have been shown to interfere with AMSU-B. The effect is largest on channels 17 and 19 but has been shown to be present in the other channels also. The errors are largest in the Earth views, but space views and to a lesser extent internal target views are also affected. This appendix describes the UK Meteorological Office (UKMO) recommendations for correcting the data.

The largest Earth view biases are in channels 17 and 19, with a strong scan-dependence and maximum errors of approximately 40K when the STX-1 and SARR transmitters are active. Since these transmitters are normally switched on all the time, the bias appears constant.

The largest space view effect is in channel 17 when the STX-3 transmitter is active. This transmitter is currently used for data dumps to a ground station, typically for 5-10 minutes per orbit (not every orbit). When the transmitter is on, the space view counts rise by the equivalent of approximately 5K and the internal target view counts rise by approximately 1K. These rises cause errors in the calibration and result in Earth view brightness temperature errors of typically 2.5K at a scene temperature of 200K, decreasing to1K at warm scene temperatures.

The correction scheme described below applies both to users of the NESDIS Level 1b product and to users who receive the HRPT transmissions directly. When the scheme is implemented, it is expected that the overall error will be reduced to approximately 1K for all channels.

Note that on 13^th October 1998, an additional bias appeared, which was largest on channel 17 (up to 60K additional error) but was present to some extent on all channels. This new bias has been present on every orbit since then, and is present for typically 75% of every orbit. See Section M.10 for further discussion of this problem.

M.1 DISCUSSION

The AMSU-B Level 1b dataset currently provides the following primary information:

• Raw instrument counts for Earth views. 90 values per channel per scan line.
• Polynomial coefficients for converting the raw counts to 'antenna temperature', i.e. the brightness temperature seen by the instrument with no correction for beam side lobes. Three coefficients per channel per scan line.

Space view counts and internal target counts are also present in the Level 1b dataset but are not required by the user once the polynomial coefficients have been calculated.

The Level 1b dataset is derived from the Level 1a dataset, which contains only raw instrument telemetry (i.e. no conversion coefficients).

In August 1998, a modification was made to the NESDIS Level 1b format to include information on the S-band transmitter status and output power levels. Thus, for any scan line it is possible to determine which transmitters were on. This information is stored in the analog telemetry data field. In general, some of the transmitters are only used during data downlinks (currently, STX-2 and STX-3) while others are on continuously (currently, STX-1 and SARR). The information available in the Level 1b dataset is as follows: STX-1, STX-2, STX-3 and STX-4 status; STX-1, STX-2 and STX-3 output power; SARR-A and SARR-B output power. All values are in the range 0 to 255 counts (representing analog voltages in the range 0 to 5.1 volts). For users of direct HRPT transmissions, the transmitter information may be extracted from the TIP data which is embedded in the HRPT data frame - see Section M.8 for details.

The correction scheme may be summarized as follows:

• Information on the required Earth view, space view and internal target view corrections are to be included in the Level 1b header, i.e. applying to the whole orbit.
• The polynomial coefficients are calculated based on corrected space view and target view counts (though the counts values stored in the Level 1b dataset are not changed).
• Scan-dependent Earth view corrections are applied by the Level 1b user before converting to radiance.

The advantages of including the correction information in the Level 1b header are as follows:

• It guarantees that the information is available to all Level 1b users
• The user does not have to keep track of any updates to the corrections

The advantages of making the space view and internal target view corrections at the Level 1A-Level 1b stage are as follows:

• Fixed space view and internal target view corrections are already applied at the Level 1A-Level 1b stage, so it is logical to apply the new correction here.
• The Level 1b user only needs to apply an Earth view correction.
• It allows the correction to be made before the quality control procedures are carried out and before the 7-line calibration convolution function is applied. This improves the results at times of transmitter change.

Strictly speaking the relationship between analog counts and actual transmitter power level is not linear. However, for the purpose of AMSU-B bias removal, a linear relationship may be assumed, particularly since the transmitters are normally either fully on (with nearly constant power level) or fully off (with zero counts).

M.2 INFORMATION TO BE STORED IN THE HEADER

Rather than attempting to identify which channels are susceptible to which transmitters, corrections are provided for all channels and transmitters. Thus, some corrections may be zero.

Transmitter - STX-1, STX-2, STX-3 and SARR. (STX-4 is not normally used, so may be ignored. Also, no distinction is made between SARR-A and SARR-B: only one of these will be active at any given time).

The count corrections are stored in the AMSU-B header record at byte offset 1001 (420 2-byte integers). The bias corrections in the header are given for 5 channels, 21 views, and 4 transmitters, in that order. The 21 views include 19 Earth scene views (views 1, 5, 10, 15, 20, ..., 85, 90) followed by cold and warm calibration views. The order of the 4 transmitters is STX-1, STX-2, STX-3, and SAR.

Transmitter reference powers - mean power at the time the above corrections were derived.

The reference transmitter powers are stored at byte offset 1849 (four 2-byte integers, scaled by 10 x value in counts, i.e. a precision of 0.1 counts). The mean transmitter power at which the bias corrections were determined are given for each transmitter, in the order STX-1, STX-2, STX-3 and SAR.

M.3 ADDITIONAL TASKS FOR THE LEVEL 1A-LEVEL 1b CALIBRATION PROGRAM

This section applies to users of HRPT direct transmissions, or to users of raw GAC or LAC data.

• Read the bias correction information and 'reference' transmitter powers from an external file (containing data generated by the UKMO - see Section M.9) and incorporate into the Level 1b header.
• For each scan line, extract the transmitter powers from the analog telemetry. Divide each transmitter power by the reference transmitter power.
• For each transmitter that has non-zero power, obtain the space view and target view corrections and multiply by the power ratio. Correct the space view and internal target view counts before doing the space view / target view counts convolution process. If more than one transmitter is on, the combined correction is the sum of the individual corrections.