AIRS Profile Assimilation -

1
AIRS Profile Assimilation - Case Study results
Shih-Hung Chou, Brad Zavodsky Gary Jedlovec,
and Bill Lapenta
2
Motivation for Profile Assimilation at SPoRT

• The SPoRT Center seeks to improve short-term
  weather forecasts by the use of satellite-based
  observation.
• AIRS data complement traditional upper-air
  observations in data-sparse regions (both ocean
  and land)
• In contrast to AIRS radiances, profiles provide
  an easier assimilation method allowing regional
  and local end users (e.g. HUN WFO) to run NWP
  systems
• Hyperspectral nature of AIRS sounder allows for
  high-resolution data

3
AIRS Specifications

• Aboard Aqua polar orbiter
• Early afternoon equator crossing
• 2378 spectral channels
• 3.7 15.4 µm (650 2675 cm-1)
• 3 x 3 footprints (50 km spatial resolution)
• AMSU allows for retrievals in both clear and
  cloudy scenes
• Version 4.0 Error Estimates (Tobin et al. 2006)
• 0.6-1.0K over ocean ( 50o latitude)
• 0.9-1.3K global ocean and land (in 1 km layers)
• lt 15 RH (in 2 km layers)

4
AIRS Data Quality Indicators
0700 UTC 20 November 2005 AIRS swath

• Quality indicators (QIs) in prototype v5
• each profile contains level-specific QI
• level-by-level error estimates for each T and q
  profile
• QIs allow for the maximum amount of quality
  data to be assimilated
• optimal use of QIs should produce an analysis
  that provides better initial conditions for the
  WRF

5
Lessons Learned from Previous SAC

• 4 January, 2004
• Pacific storm stalled off shore limited its
  impact on land
• Difficult to evaluate AIRS impact due to
  insufficient RAOB stations and stage IV precip
  data for verification
• Mixed results for AIRS impact on forecast

6
Case Study November 20-22, 2005
Rapidly intensifying storm off the eastern
seaboard under forecasted by GFS, NAM, and SPoRT
operational WRF
Case Selection

• relevant to SPoRT interests in SEUS region
• ample verification data available over the
  Eastern US synoptic setting
• opportunity to eventually test both over-ocean
  and over-land AIRS profiles
• comparable CONUS domain to other SPoRT WRF for
  easy transfer to operational applications

7
Analysis and Forecast Model Configuration

• WRF Model Configuration
• 36km domain with 150x360 grid
• 37 vertical levels
• Initialized with NAM analysis, LBC updated every
  3 h
• ADAS Analysis Configuration
• Same horizontal domain as WRF
• 43 vertical levels separated by 500 m
• AIRS profiles are assimilated as RAOBs using QIs
  to determine highest quality data
• use Tobin et al. (2006) for observation error and
  standard model errors for background
• Assimilation / Forecast
• 7h forecast used as background for ADAS

AIRS valid at 0700 UTC
00 UTC
ADAS
7h FCST
11/21/05
00 UTC
00 UTC
Validation at 00 UTC and 12 UTC
11/20/05
11/22/05
8
Impact of AIRS Profiles on ADAS Analysis
700 hPa Temp Difference
AIRS data have an cooling impact over Atlantic,
but a warming impact on land
9
Impact of AIRS Profiles on ADAS Analysis
20 November 2005 Wallops Island, VA
07Z BKGD 07Z AIRS 07Z ADAS
10
Impact of AIRS Profiles on Initial Conditions
20 November 2005 Wallops Island, VA
07Z BKGD 07Z AIRS 07Z ADAS 00Z RAOB 12Z RAOB

• AIRS shows cooling in the lower and upper
  troposphere

• AIRS shows drying above 900 hPa

AIRS can spatially and temporally fill the gap
between conventional observations
11
Temperature and Moisture Impact

• Control is too warm and moist at all
  tropospheric levels

12
6-h Cumulative Precipitation Impact

• CNTL over-forecast over the low center and
  under forecast over TN/AL
• AIRS improves forecast compared to NCEP Stage
  IV data in region of heaviest precipitation

13
6-h Cumulative Precipitation Impact

• Qualitative Precipitation Forecast

• Bias Score
• a measure of precip coverage
• Precipitation under-forecasted
• CNTL better at middle threshold AIRS better at
  high
• Equital Threat Score
• a measure of precip loaction
• AIRS outperforms CNTL at most threshold
  similar at smallest threshold

14
Summary

• AIRS Level-2 profiles provide valuable data over
  regions otherwise devoid of upper-air
  observations they also fill the gap in time
  between the conventional observations
• Level-specific QIs for AIRS profiles allow for
  the assimilation of the largest volume of highest
  quality data
• AIRS data improves forecasts of T, q, and 6 h
  precip
• Future plans involving AIRS
• Real-time forecasts to evaluate long-term impact
• Select new case studies for in-depth analysis