SeaWinds AMSRderived Impact Table

1
SeaWinds AMSR-derived Impact Table

• Jan 17, 2006
• Bryan Stiles and R. Scott Dunbar

2
Impact Table Overview

• Speed Bias and Cross Track Directional Bias are
  tabularized by
• Attenuation
• Backscatter ratio B/ ?0
• Cross Track Distance
• Average outer beam measured ?0
• Cross track bias is computed for
  Scatterometer-only DIRTH w.r.t NCEP
• Speed Bias is computed similarly below a ?0
  threshold but is computed directly from A and B
  above the threshold.

3
To compute speed impacts at high winds we

• Estimate retrieved wind speed from average
  measured s0. (using QSCAT1)
• An estimate of speed is derived for each beam.
• An average value is computed weighted by the
  number of measurements from each beam.
• Estimate true wind speed similarly from s0s
  corrected using A and B.
• Compute the difference between the two speeds.
• Accumulate difference for each WVC in SWS mission
  into bins in speed impact table.
• This resulting table is the GMF-derived table.

4
Why do we compute Speed Impact two different ways?

• The two methods agree well for moderate wind
  speeds (7-15 m/s).
• The A, B model of rain impact on s0 was estimated
  as a function of liquid, vapor, SST, and beam.
• To obtain agreement between the A, B model and
  the NWP speed biases at high winds.
• Rain backscatter would have to be a increasing
  function of wind speed. AND/OR
• Attenuation would have to decrease with wind
  speed.
• Because the required changes in A, B seem
  counterintuitive, we conclude that the NWP speed
  winds are systematically biased low for cases
  with high wind speed and high liquid.
• This could be explained by rain correlating with
  small scale high wind regions, poorly represented
  by low resolution NWP fields.
• For this reason we do not want to use NCEP speed
  impacts for high winds.

5
At low wind speeds computing speed impact
directly from A and B overestimates speed impact.

• Why?
• Errors in liquid estimate.
• Cases of truly high liquid and low measured s0 do
  not occur.
• Some bins in impact table are populated solely by
  the high end of the liquid error distribution.
• Errors in backscatter, B, estimate
• For very low winds where rain-free s0lt -25 dB, B
  estimates cannot meet the required precision to
  correct the speeds well.
• Solution?
• New table using NCEP-derived impact for average
  outer beam measured s0 lt 0.021.
• O.021 is the 99.9 percentile of the B value.

6
Speed impact as originally derived by
SWS-NWPHigh wind biases are wrong due to NCEP
error
7
Speed Biases Computed from A and BLow wind
biases are wrong due to errors in liquid and B
8
Final Hybrid table fixes high speed and low speed
problems
9
Speed Impact vs ECMWF wind and AMSR Liquid
10
Direction Impact vs ECMWF wind and AMSR Liquid