 |
 |
 |
RSS Climate Data Record Updates |
The September 2006 Update to RSS Climate Data Records
Introduction
A major update to the SSM/I and TMI datasets has been implemented.
SSM/I Version 5 (V05) has been updated to Version 6 (V06).
TMI Version 3 (V03) has been updated to Version 4 (V04).
The major objectives of the updates were to:
1. |
Remove spurious trends in the wind speed retrievals. |
2. |
Implement a much-improved rain rate algorithm. |
3. |
Achieve better consistency for all retrievals over the 8 satellite platforms
(i.e, 6 SSM/I, TMI, and AMSR-E). |
4. |
Make minor improvements in other retrievals, namely SST and vapor. |
In addition to the SSM/I and TMI updates, a minor update to the AMSR-E rain rates has also been implemented. There was a small bug in the calculation of the composite maps (3-day, weekly, and monthly) that mostly affected the monthly maps. As a result, the monthly rain rates were biased low by about 0.03 mm/hr relative to the daily maps. Also, in the daily maps, extremely light rain (<0.025 mm/h) is now set to zero. This change has little effect on the mean rain rates, but does affect the aerial coverage. Since this is a minor update, we denote it as V05a, as compared to the preceding V05 dataset.
Two scientific papers describe in detail the new wind products and rain products:
 How Much More Rain Will Global Warming Bring?
Intercalibrated Passive Microwave Rain Products from the Unified Microwave OceanRetrieval Algorithm (UMORA)
For all of these updates, the satellite datasets have been completely reprocessed and the new versions, in their entirety, are now available for downloading. Any users doing climate work with our wind speeds or rain rates should definitely replace their earlier versions with SSM/I V06 and TMI V04. We also suggest that the AMSR-E V05a datasets be downloaded to ensure complete consistency among the various datasets. There are no changes to the file format, so read routines do not need to be changed.
RSS considers the September 2006 Update to be a very important milestone in our production of Climate Data Records (CDR). Investigators can confidently use these new satellite datasets for detailed interannual and decadal trend studies. As always, continued funding for production and dissemination of these CDR is dependant upon you, the users, to let us know how you have used these data and their value to your research. Please make sure you provide us with information about your work, send results and/or papers to our office, and always provide your correct email when accessing the data by ftp.
Thank you
Algorithm Improvements for SSMI
1. |
The intersatellite calibration of the 6 SSMIs at the antenna-temperature (TA) level has been completely redone. There were problems with the V05 calibration procedure that was introducing spurious trends into the wind speed retrievals. Fortunately these problems had only a small effect on the vapor retrievals. The new SSM/I wind retrievals are now consistent with buoys and scatterometers over the 19-year SSM/I period (1987-2006). The intersatellite agreement in the water vapor retrievals is also improved for V06. |
2. |
A much improved rain algorithm has been implemented. More realistic freezing level heights and beam-filling corrections are used. We rederived the relationship between rain column height and sea surface temperature (Reynold's SST) using NCEP freezing level heights in raining conditions. The new relationship produces freezing level heights that are more realistic and spatially representative. In addition, we have made two major changes to the beamfilling correction. 1) Previously, we allowed the correction to "max-out" when the radiometer saturated. The new correction produces more realistic corrections when saturation occurs. 2) We now incorporate footprint size into the beamfilling correction in order to bring SSM/I, AMSR-E and TMI rain rates into better agreement. As a result, the cloud water and rain rate retrievals are very consistent across all 8 satellite platforms (i.e, 6 SSM/I, TMI, and AMSR-E). This consistency is also a result of using the same method of TA resampling for all satellites (see next item). |
3. |
The geophysical retrieval algorithms required that the TA measurements from different frequencies all be resampled to a common spatial resolution. For SSM/I, we had been using a very simple averaging technique to accomplish the resampling. In the V06 algorithm, a more precise TA resampling method is used to optimally interpolate the TA observations to a common footprint. This technique is now the same as used for TMI and AMSR-E. Use of the new technique had little impact on surface wind speed or columnar water vapor, but it did have a substantial impact on the rain retrievals. It increased the cloud water and rain rates making them more consistent with AMSR-E and TMI. |
4. |
The geophysical retrieval algorithms and data processing systems have been modified and restructured to bring a higher degree of commonality to the SSM/I, TMI, and AMSR-E. Essentially the same set of algorithms is now used to process all three types of satellite microwave radiometers. This will help to ensure greater consistency among the extended suite of satellite microwave radiometers. |
Algorithm Improvements for TMI
1. |
The generation of a CDR for TMI is particularly challenging due to the following:
a. |
The vapor-deposited aluminum (VDA) on the TMI antenna oxidized and/or cracked. As a result the antenna has an emissivity of graphite, which is 3 to 4%. |
b. |
The solar environment for TMI is constantly changing due to its near-equatorial orbit drifting through the diurnal cycle. Furthermore, the solar environment changes radically every month or so when a 180° yaw maneuver is completed. |
c. |
In September 2001, the TMI orbit was boosted from an altitude of 350 km to 400 km. |
d. |
There are small errors in the knowledge of the satellite roll and pitch, particularly right after the 2001 orbit boost. |
Our new calibration algorithms attempt to correct for all these effects. We found that the algorithm used to correct roll and pitch errors was introducing a small along-scan error in the SST and wind retrievals. This problem has been fixed in V04. We also found small systematic biases in the SST, wind, and vapor retrievals that were correlated with the yaw state (either 0° or 180°). Finally, we found small systematic biases in wind and vapor for the time period before the boost as compared to the time period after the boost. Both these biases have been removed in V04 products. |
2. |
The new rain rate algorithm was implemented, as described in Item 2 above for SSM/I. As a result, the cloud water and rain rate retrievals are now very consistent across all 8 satellite platforms (i.e, 6 SSM/I, TMI, and AMSR-E). |
3. |
TMI had been using the same type of TA resampling as we use for AMSR-E (see Item 3 above in the SSM/I description). However, the resampling was not correctly taking into account the change in geometry that occurs when TMI goes through a 180° yaw maneuver. The resampling weights were correct for 180° yaw orientation but were incorrect for the 0° yaw orientation. This error led to a distortion near the swath edges and produced a positive bias in cloud water and rain rate at the swath edges. We have corrected this problem and now use 4 separate sets of resampling weights: yaw 0°-preboost, : yaw 180°-preboost, yaw 0°-postboost, : yaw 180°-postboost. Use of the new weights makes little impact on SST, wind speed, or columnar water vapor, but does significantly affect cloud water and rain rate. |
4. |
The geophysical retrieval algorithm and data processing system was updated as described in item 4 above for SSM/I. |
|
|
 |
 |
Introduction
