Development and Preliminary Results of Image Processing Tools for Meteorology and Air Quality Modeling

1
Development and Preliminary Results of Image
Processing Tools for Meteorology and Air Quality
Modeling

• Limei Ran
• Center for Environmental Modeling for Policy
  Development
• Institute for the Environment, UNC-Chapel Hill
• Jonathan Pleim and Robert Gilliam
• Atmospheric Modeling and Analysis Division
• USEPA/ORD/NERL, RTP, NC

2
Outline of the Presentation

• Spatial Allocator (SA) Raster Tools
• GOES, OMI, and MODIS Tools
• WRF and CMAQ Outputs, 08/2006
• Conclusions and Future Work
• Acknowledgements

3
Raster tools in Spatial Allocator (SA)

• The SA Raster tools are developed to compute
  image/satellite data on model grids
• The tools can process
• 2001 30m NLCD and 1km MODIS land cover data
• GOES satellite data (new)
• OMI Level 2G and 3 data (new)
• MODIS Level 2 cloud product (under development)
• SA Web site http//www.ie.unc.edu/cempd/projects/
  mims/spatial/
• Plan to release new SA 11/2009

4
Model Grid Satellite Data Computation

• Three steps in computing model grid satellite
  data

3. Compute model grid satellite value average or
total
3a. Project sat. points to grid cells
1. Create model grid shapeFile or raster file
2. Rasterize model grids
3a
3b
3b.Project grid cells to Sat. data projection
5
GOES Satellite Data

• GOES satellite data from National Space Science
  and Technology Center (NSSTC) at Huntsville, AL
• http//satdas.nsstc.nasa.gov/index.html
• GOES data contain two types of hourly data
• GOES Imager 4km
• Cloud Albedo, Insolation, Surface Albedo, Cloud
  Top Pressure, Infrared Temperature
• GOES Sounder 10km
• Cloud Top Pressure, Skin Temperature, Total
  Precipitable Water
• Data are in GRIB format
• Two script files
• Compute GOES data on model grids and create a
  WRF-ready NetCDF file (1 hour time step)
• Convert NetCDF file into WRF data assimilation
  format

6
GOES Imager Surface Albedo
East US 12km Domain
GOES Data
()
()
7
GOES Sounder Skin Temperature
East US 12km Domain
GOES Data
8
OMI Level 2G and 3 Product Tool

• Aura OMI daily global data from NASA GIOVANNI web
  site
• http//disc.sci.gsfc.nasa.gov/giovanni/overview/in
  stances_atmospheric.html
• Level 3 ozone, aerosol, and cloud (0.25 or 1
  degree resolution)
• Level 2G ozone, aerosol, SO2 and NO2 (0.125 or
  0.25 degree resolution)
• Data are in HDF4 format
• One script file to compute grid OMI data

9
OMI L2G NO2 Tropospheric Column Density 1015
molec/cm2
10
MODIS L2 Cloud Products

• Each MODIS L2 cloud product file (one granule)
  contains
• 39 cloud variables in 5X5km or 1X1km arrays (at
  nadir)
• five-minute time interval data for area
  1354X2030km
• Data are in HDF4 format
• http//modis-atmos.gsfc.nasa.gov/MOD06_L2/index.ht
  ml
• http//ladsweb.nascom.nasa.gov/data/search.html
• Two tools
• Compute model grid one cloud variable from one
  MODIS L2 cloud product file
• Compute model grid multiple cloud variables from
  given periods MODIS L2 cloud product files
  (under development)

11
NLCD and GLCC Evergreen Forest
East US 12km Domain
Texas 4km Domain
Houston 1km Domain
12
US Eastern 12km Grid Domain small difference
GLCC
13
Difference in NH4 aerosol concentration error for
August 2006 NLCD MAE GLCC MAE
14
Houston 4km ComparisonsNLCD versus USGS
Example of how the more finely resolved NLCD
landuse represents areas along complex coastlines
more accurately. The NLCD simulation represents
diurnal temperature better than the 1-km base
USGS. Landmask to the right is more refined when
NLCD is used.
NLCD
USGS
15
2-m Temperature Time Series Average of all sites
around Houston
Mean absolute error (MAE) for NLCD simulation is
0.69 K and the USGS is 0.83 K. Mean BIAS is near
zero for NLCD and -0.18 for USGS. Index of
Agreement (IOA) is 0.94 versus 0.91.
16
Houston area 1-km grid resolution WRF model
simulation
2-m Temperature and 10-m Winds Aug 5, 2006 at
21 UT (4 PM LT) NLCD LU data

• Fine scale effects of urban heat island
• Bay breeze from Galveston Bay

17
Conclusions

• WRF runs
• Preliminary comparisons show small differences at
  12-km grid resolution
• Improved results from NLCD runs at 4-km and 1-km
  grid resolutions, particularly near coastlines
• CMAQ runs
• Largest difference is in the bi-directional NH3
  surface flux because it is closely related to
  landuse, especially crops.

18
Future Work

• Run CMAQ for 4-km and 1-km Texas domains
• Compare results with WRF MODIS landuse data
• Use BELD4 from 2001 NLCD and MODIS land cover,
  FIA, and NASS for biogenic emission
• Develop tools to process and analyze A-Train
  satellite products (MODIS Aqua, Aura, PARASOL,
  CALIPSO, and CloudSAT) for aerosol and cloud
  interaction in air quality and regional climate
  chemistry research

19
Acknowledgements

• This project is funded by
• US EPA under the Contract No. EP-W-095-023,
  Operation of the Center for Community Air
  Quality Modeling and Analysis.
• NASA ROSE Grants UAH and NNX08AL28G
• We gratefully acknowledge the support of
• Arastoo Pour Biazar and Bill Crosson from the
  National Space Science and Technology Center at
  UA Huntsville, AL
• Aijun Xiu, Uma Shankar from IE, UNC-Chapel Hill
• William Benjey, Ellen Cooter, and Wyat Appel from
  US EPA