Impact of AIRS Retrievals on Forecast Skill using the GEOS5 DAS Oreste Reale, Bob Rosenberg, Eugenia

1
Impact of AIRS Retrievals on Forecast Skill using
the GEOS-5 DASOreste Reale, Bob Rosenberg,
Eugenia Brin, Joe Terry, Juan Carlos Jusem, Joel
Susskind, Max Suarez, Larry Takacs, Ricardo
Todling, Lars Peter Riishojgaard

2

• Special Thanks
• Michele Rienecker and the GMAO for providing the
  GEOS-5 DAS and the technical support needed to
  set up and run these experiments.
• AIRS team at JPL
• Joel Susskind and his group
• Genia Brin for her tireless work in setting and
  running the experiments

3
The Experiment

• Control assimilation is the GEOS-5 DAS Version
  beta7P2 at 1x1 degree resolution
• AIRS assimilation is the 1x1-deg. GEOS-5 DAS
  Version beta7P2 with AIRS retrievals inserted as
  rawinsonde temperature profiles at 100 levels
• Retrievals used for this experiment are produced
  by the Version 4.7 AIRS algorithm with latest
  quality control, thinned by removing every other
  point

4
The Experiment

• Control assimilation started 12/15/2002 and run
  through 1/31/2003
• AIRS assimilation started 1/1/2003 and run
  through 1/31/2003
• Individual 5-day forecasts begun at 00Z each day
  starting 1/6/2003 and ending 1/31/2003.
• First 5 days in January discarded for spin-up,
  used 26 5-day forecasts.
• Both control and AIRS forecasts verified against
  the NCEP analysis

5
GEOS-5 AIRS vs. Control
6
Individual Forecast Anomaly Correlations
7
Case Selection

• Goal was to pick a forecast where the control did
  reasonably well and AIRS showed notable
  improvement
• Selected forecast from Jan. 25 (case 20)
• Looked in NH Extratropics for a link between the
  data and a synoptic event that was better
  forecast using AIRS data

8
AIRS data over N. Pacific used by assimilation on
00 GMT Jan. 25, 2003

• (will show plot of global coverage of AIRS data
  on January 25, 2003)

9
Hov. Z500 1/25-1/30 30-60N
10
500MB Height anomalies
11
700MB Height anomalies
12
Summary

• AIRS retrievals produced meaningful improvement
  in GEOS5 skill over a sample of 26 5-day
  forecasts in Jan. 2003
• A case was selected where an improved
  representation of geopotential height over the
  northeastern Pacific leads to a better forecast
  of deep cyclones over Canada and the N Atlantic.

13
Future Work

• We plan to evaluate all cases of meaningful
  impact, both positive and negative.
• We will document the impact and the manner of the
  dynamics related to that impact.
• We plan to re-run experiments using Version 5
  retrievals with specified error estimates.