Comparison of Land Surface Skin Temperature derived from NCEP operational NWP models, LDAS, satellit

1
Comparison of Land Surface Skin Temperature
derived from NCEP operational NWP models,LDAS,
satellites, and ground observationsJesse Meng
and Ken MitchellNCEP/EMC (NOAA)
2
Acknowledgements

• NCEP Environmental Modeling Center (EMC)
• Land Team
• K. Mitchell, H.Wei, G.Gayno, J.Meng, Y.Xia,
  C.Marshall, V.Wong, M.Ek. R.Yang
• Mesoscale Modeling Branch
• B. Ferrier, E. Rogers
• Global Modeling Branch
• S. Saha, G. White, Y. Hou, F. Yang
• NESDIS Office of Research Applications (ORA)
• Land Team
• D. Tarpley, I. Laszlo
• NOAA Climate Program Office (NCPO)
• CPPA Climate Prediction Program for America (Jin
  Huang)
• CPPA-funded PIs R. Pinker, B. Cosgrove, J.
  Augustine, T. Meyers,

3
LST Land Surface Skin Temperature

• Satellite sources of LST
• Geostationary (split window, sounder)
• AVHRR
• MODIS
• SSMI, AMSU, other microwave platforms
• NCEP to date has utilized mostly GOES-based LST
• Operational in NESDIS hourly, 1/2-deg resolution
• Test Bed at U.Md (R. Pinker) hourly, 1/8th-deg
  resolution
• Split window technique through GOES (I-M) (before
  2003)
• Sounder technique most recently (since 2003)
• Good definition of diurnal cycle

4
LST Data and Acronyms

• NAM regional forecast
• NDAS regional 4DDA
• GFS global forecast
• GDAS global 4DDA
• LDAS uncoupled Land DA
• GOES satellite retrieval
• OBS surface observation

5
LST in NCEP NWP models

• Upward longwave radiation
• LW? e s (LST)4
• Sensible heat flux
• SH Ch (LST Tair)

6
NCEP 2-m air temperature verification
Eastern US
Western US
Green Observation Blue OPS GFS Pink PRY GFS
7
Monthly Mean 18Z LST K Feb2006
12 km
0.125 lat/lon
8
Monthly Mean 18Z LST Diff K Feb2006
12 km
0.125 lat/lon
Model/DAS are warmer than satellite estimates
over snow, cooler over non-snow area.
9
Monthly Mean 18Z LST Diff K Feb2006
12 km
0.125 lat/lon
NDAS tends to increase LST in vegetated
area. LDAS is forced with GOES estimated SW
radiation and gauge precipitation.
10
Monthly Mean 18Z LST K May2006
12 km
0.125 lat/lon
11
Monthly Mean 18Z LST Diff K May2006
12 km
0.125 lat/lon
Model/DAS are warmer than satellite estimates in
wet area (ET), cooler in dry area (e).
12
Monthly Mean 18Z LST Diff K May2006
12 km
0.125 lat/lon
NDAS tends to increase LST in vegetated
area. LDAS is forced with GOES estimated SW
radiation and gauge precipitation.
13
Monthly Mean 18Z LST K May2006
0.3125 lat/lon
12 km
GFS/GDAS are cooler.
14
Monthly Mean 18Z LST K May2006
0.3125 lat/lon
0.125 lat/lon
GFS/GDAS are cooler.
15
LST verification May2006 Mean
16
LST verification 07May2006
310
300
17
LST verification May2006 18Z
An emissivity reduction from 1.0 to 0.95 can
reduce LST by 4K.
18
Examples of sfc emissivity in Noah LSM

• Operational
• No snow 1.00
• Snowpack 0.95
• Test Bed
• Cropland .985 (summer) .92 (winter)
• Grassland .960 (summer) .92 (winter)
• Tropical forest .95 always
• Savanna .92 always
• Tundra .92 always
• Bare ground/sparse veg .90 always

19
Summary

• LST is critical to remote sensing and data
  assimilation.
• LST in land surface modeling is sensitive to
  emissivity, radiation, snowpack,
  evapotranspiration (soil/vegetation type, soil
  moisture/temperature).
• LST validation is challenging due to mismatched
  spatial/temporal scales, especially, subgrid
  variability.
• Broadband IR emissivity is needed to improve
  LST modeling, satellite retrieval, and
  assimilation.