AN ENHANCED SST COMPOSITE FOR WEATHER FORECASTING AND REGIONAL CLIMATE STUDIES

1

• AN ENHANCED SST COMPOSITE FOR WEATHER FORECASTING
  AND REGIONAL CLIMATE STUDIES
• Gary Jedlovec1, Jorge Vazquez2, and Ed Armstrong2
• NASA/MSFC Earth Science Office, Huntsville,
  Alabama
• NASA/JPL, Pasadena, California

2

• Motivation
• Need for high-resolution SST datasets for coastal
  applications and modeling
• fluxes of heat and moisture from ocean to
  atmosphere is closely coupled to SST
• coupling poorly represented in models used in
  coastal weather studies (Chelton et al. 2007
  Lacasse et al. 2008) due to failure to resolve
  areas of large SST gradients
• Current datasets insufficient in coastal or
  regions of large gradients in SST
• Presentation describes the collaborative research
  of scientists with the GHRSST-PP and the SPoRT
  program to
• produce an enhanced high-resolution (1km) SST
  product based on a proven current composite
  approach. Enhancement will include the
  incorporation of Single Sensor Error
  characteristics contained in the GHRSST-PP
  products.
• provide a near real time product from Level 2P
  data for distribution to user community

3

• GHRSST-PP
• Global Data Assimilation Experiment (GODAE) High
  Resolution Sea Surface Temperature Pilot Project
  (GHRSST-PP) operationally produces improved high
  and medium resolution global SST products from a
  number of different satellite sensors
• the MODIS and the AATSR derived SSTs both have
  1km spatial resolutions
• Microwave derived SSTs from AMSR-E and TMI
  produce SSTs at lower resolutions (25km) - not
  impacted by clouds.
• Background
• Short-term Prediction Research and Transition
  (SPoRT) project NASA / MSFC activity to
  transition unique NASA observations and research
  capabilities to the operational weather community
• focus on improvements in regional, short-term
  weather forecasts
• primary end users are NWS Forecast offices in
  Southern Region
• MODIS, AMSE-E, AIRS data and associated products
• nowcasting products such as total lightning,
  convective initiation indices, GOES aviation
  products
• unique regional weather model predictions (driven
  by NASA data)

4

• Current technique
• EOS science team algorithm used to process MODIS
  direct broadcast data at Univ. of South Florida
  (or archived data from DAAC) Terra and Aqua,
  day and night
• Assume day-to-day changes in SST are relatively
  small and preceding days values can be used to
  fill in cloudy regions
• Remap SST data to 1km grid at each time
• Apply cloud mask to filter cloud-free data
  (Jedlovec et al. 2008)
• Consider three most recent cloud-free SST values
  for each pixel (from the past week of data)
  call this a collection
• For each collection, exclude coldest value (bad
  data and extra cloud filter) and average other
  two to produce a composite SST value for each
  pixel
• 4x daily SST composite for Gulf of Mexico and
  near Atlantic region

5

• Research approach
• Enhancement to the current MODIS 1km SST
  composite product (Haines et al. 2007). Three
  primary aspects goal of the work.
• add 1km AATSR data and integrate AMSR-E
  (microwave) data to reduce the latency of the
  composite
• extend coverage of composite SST product for both
  the West and East Coasts of the United States,
  including the Gulf of Mexico
• incorporate error GHRSST-PP data / source error
  characteristics to the composite maps
• Validate approach with in-situ data from the
  World Ocean Database (WOD) and other sources and
  determine improvements of the enhanced composites
  (i.e., accuracy, reduced latency. etc.)
• Transition products to SPoRT program to support
  operational activities which include numerical
  weather forecasting and fisheries, and regional
  climate studies

6
MODIS

• MODIS
• 1km SST composite using 3 days of data (similar
  to Haines and Jedlovec approach).
• data from relatively clear days, some cloudy
  regions (black)
• large amplitude small spatial scale gradients in
  SST field are important and need to be preserved
• AMSR-E data
• AMSR-E SST 25km resolution at same composite
  length as MODIS. Only large scale events captured
  and missing data along the coast.
• MODIS and AMSR-E
• intercalibration accuracy of data allows for a
  combination of data from different instruments
• AMSR-E provides accurate information in cloudy
  regions
• new composite can utilize error estimates and
  latency to form a weighted composite product
• maintains accuracy and resolution
• reduces MODIS latency

AMSR-E
MODIS AMSR-E
7

• Advanced composing technique
• 4x daily composites MODIS (Terra Aqua), AATSR
  (morning), and AMSR-E (Aqua)
• Procedure
• remap SST data to common 1km grid at each time
• apply cloud mask to filter cloud-free data
• implement bias removal between instrument SSTs
  with PDFs
• combine AATSR and MODIS data at common time
  (morning)
• where MODIS data missing, use microwave SST to
  fill in (nearest neighbor or bilinear
  interpolation)
• consider three most recent cloud-free
  MODIS/AASTR/AMSR-E SST values for each pixel
  (from the past week of data) call this a
  collection
• process each collection using error weights
• SSTwgt (SST1xrmsr1 SST2xrmsr2 SST3xrmsr3) /
  (rmsr1 rmsr2 rmsr3)

8

• Summary
• Results have shown that low resolution SST data
  sets are insufficient to resolve air-sea coupling
  dynamics in areas of high SST gradients such as
  the Gulf Stream and along upwelling areas
  associated with the Eastern Boundaries of ocean
  basins.
• In a collaboration between SPoRT and the
  NASA/PODAAC/GDAC a new enhanced MODIS 1km
  coastal, that covers the US coasts, will be
  produced/
• simple methodologies will be implemented to use
  the error characteristics contained within the
  GHRSST data sets
• If successful the potential exists that these
  higher resolution data sets and composites will
  significantly impact weather prediction
  forecasting and fisheries management