AMSU ALGORITHM CHANGE LOG

Spacecraft status as it relates to CIMSS AMSU algorithm
NOAA-15 - Launched 05/13/1998
   - AMSU-B interference until Sep 1999
   - Channels 18,19 and 20 turned off on AMSU-B Sep 16, 2010
   - AMSU-B has failed 03/29/2011

NOAA-16 - Launched 09/21/2000 channel 8 is noisy.
   - Channel 4 failed 03/04/2008 and since this impacts limb correction of channel 5 that channel is noisy
   - NOAA-16 decommissioned June 10, 2014 following unrecoverable spacecraft anomaly

NOAA-17 - Launched 06/24/2002.
   - AMSU-A failed (scan motor) 10/30/2003

NOAA-18 - Launched 05/20/2005. No isssues

NOAA-19 - Launched 02/06/2009.
   - Channel 8 became noisy 12/22/2009. Noise has become signficant enough to impact estimates.

METOP-A - Launched 10/19/2006. Channel 7 failed 12/16/2008. No estimates being produced in real-time for this reason.

METOP-B - Launched 09/17/2012

_________________________________________________________________________________________________________________________
February 08, 2017 - Added two new QC flags. One checks to see if the FOV used for the channel 8 Tb max is near the scan edge (FOV 1-2 and 29-30). Previously only CH7 FOV was used but occationally these two values may be different. Thus a CH7 FOV 3 estimate would give a valid result even if FOV 8 was closer to the scan edge. The other flag is to flag estimates where the difference between rmw and FOV resolution is < -40 km.

October 03, 2016 - Made a change to how ARCHER input is used in AMSU and ATMS. Instead of using the most recent eye size and ARCHER score use the smallest eye size and strongest ARCHER score within 3 hours of the pass.

October 02, 2016 - Complete re-work of AMSU regressions using 1998-2015 data. The main changes include accounting for TC eyewall slope (45 degs) in the three channels (6-8), accounting for loss of AMSU-B 89 Ghz, different calculations of the motion component to add to Vmax, and dealing with the noise on ch8 on N-19. Bulk stats for the indenpendent sample are N=773 MSLP Bias = 0.5 hPa Abs Error = 5.8 RMSE = 7.9 Vmax Bias = 0.0 kts Abs Error = 9.2 RMSE = 11.8

November 01, 2015 - Added Metop-B to processing starting with Cyclone Chapala 04A.

August 09, 2014 - N15 noise appears to have abated. Storms impacted are 07E (Genevieve), 10E (Julio), 09E (Iselle) and 11W (Halong)

July 29, 2014 - N15: Severe noise noted in all channels of N15 last 48 hours for some passes

June 10, 2014 - N16 decommisioned

May 28, 2011 - Due to loss of AMSU-B on N15 the bracket correction that accounts for the AMSU FOV position relative to the true TC position now uses the difference between the scattering-corrected channel 6 and non-scattering corrected channel 6 to aproximate the 89 Ghz signal.

April 10, 2011 - AMSU-B on NOAA-15 has failed. AMSU estimates will be impacted as this channel provides the bracket correction. A fix is being worked on.

September 14, 2010 - N15 AMSU-B channel 16 not available for AMSU intensity estimates due to loss of that channel.

September 21, 2010 - N15 AMSU-B channel 16 restored. Channels 18-20 are no longer available due to failed oscillator (see POES status reports for more information).

July 27, 2010 - Excessive noise continues to occur in channel 8 on NOAA 19. If noise continues to increase this channel may need to be removed from AMSU algorithm for N19. This would result in degraded algorithm performamce for N19.

March 11, 2010 - Changed the vertical cross sections to eliminate the noisy anomalies in the lowest channels below channel 5. Smoothing is done by using 75% of channel 5 for channel 4 and 50% of channel 5 for channel 3. Typically these channels have a significant contribution from the surface thus if any portion of the pass is over land the result is unrealistic anomalies near the TC center. The same can occur near strong ocean sst gradients. An example of the change can be seen here:

Noisy cross section



Clean cross section:



February 1, 2010 - Adjusted the limb correction coefficients from NOAA-15 to remove some residual limb brightening/cooling that was apparent. Impact should be minimal, perhaps 0.3K change for the dervived anomaly magnitiudes.