Skip to main content
ARS Home » Pacific West Area » Maricopa, Arizona » U.S. Arid Land Agricultural Research Center » Water Management and Conservation Research » Research » Publications at this Location » Publication #207912


item Inamdar, Anand
item French, Andrew

Submitted to: Journal of Geophysical Research
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 12/11/2007
Publication Date: 4/9/2008
Citation: Inamdar, A.K., French, A.N. 2008. Land surface temperature retrieval at high spatial and temporal resolutions over the southwestern United States. Journal of Geophysical Research. 113:1-18.

Interpretive Summary: Land surface temperature (LST) and its daily cycle of variation are important elements of the earth’s climate system and influence the earth’s water cycle, day-to-day weather and climate profoundly. Thorough knowledge of LST is useful over agricultural areas to help determine crop water demands and facilitate water management decisions. LST is measured routinely by ground meteorological stations, but such measurements are not available at many places worldwide. In order to assess water needs, monitor crop performance and production estimates, etc., we need LST measurements at every 1 km interval or less and every hour of the day and night. This is possible only through satellite remote sensing. In the present study, we utilize measurements from two different kinds of satellites: polar orbiting and geostationary. Polar orbiting satellites circle the earth in the north-south polar direction from an orbit height of about 800 km above the earth’s surface, and provides a global view of the earth's surface. Every area on the earth’s surface is observed at least once and at most twice per day. Geostationary satellites, on the other hand, are parked at a fixed location above the earth’s surface and make frequent measurements (every half hour) of a limited area of the earth’s surface. In this study, we combine remote sensing measurements from the NASA (National Aeronautics and Space Administration)-launched polar orbiters TERRA and AQUA and also NOAA (National Oceanic and Atmospheric Administration)-launched Geostationary Environmental Satellite (GOES) satellite, to derive LST on a 1 km spatial scale at half hour interval.

Technical Abstract: Land surface temperature (LST) and its diurnal variation are important observable characteristics when evaluating climate change, land-atmosphere energy exchange processes and the global hydrological cycle. These characteristics are observable from satellite platforms using thermal infrared, but doing so at both high spatial and high temporal resolutions has been difficult to achieve. Accurate temporal and spatial knowledge of LST is critical in global scale hydrological assimilation to improve estimates of soil moisture and evapotranspiration and to help monitor climate change. High temporal sampling of LST is achievable with geo-stationary satellites, but at spatial resolutions (> 4 km) often too coarse to accurately distinguish significantly different land surface types. In the past, satellite retrieval of good quality global LST at 1 km scales has relied upon NOAA polar orbiting satellites. Though augmented by MODIS (Moderate Resolution Imaging Spectroradiometer) data on board the EOS (Earth Observing System) Terra satellite since 2000, these data provide, at best, only one or two instantaneous observations per day. High temporal sampling of LST, on the other hand, is achievable with geostationary satellites, but with 4-5 km spatial resolution and lower accuracy. In our study, we employ MODIS as a calibration source for the GOES (Geostationary Environmental Satellite) satellite. Methodologies are presented to combine data from both types of satellite observations to yield half-hourly LST values, at 1 km spatial resolution. To accomplish this task, and return LST accurate to better than 2 C, minimally requires good cloud clearing and atmospheric correction algorithms. Also required is an underlying LST model to propagate values between satellite observations. Thus, the methodology we developed involves a complex sequence of tasks : a) a multi-stage cloud clearing approach; b) matching and merging of cloud-cleared GOES pixels with aggregated MODIS pixels; c) application of a generalized split-window scheme for performing atmospheric correction; and d) the use of an empirical harmonic model for the diurnal temperature cycle to improve the quality of the final LST product. The resulting 4 – 5 km LST is reduced to 1 km employing the correlation between the Normalized Difference Vegetation Index (NDVI) and LST. Evaluation of the retrieved LST against ground truth data indicated that the method yielded an accuracy of about 2 C.