Submitted to: American Geophysical Union
Publication Type: Abstract only
Publication Acceptance Date: 5/27/2002
Publication Date: 5/29/2002
Citation: Kustas, W., Norman, J., Anderson, M., French, A. The utility of higher resolution surface temperature imagery derived from coarser resolution vegetation index-surface temperature data. EOS Transaction of the American Geophysical Union. 2003. v. 83. http://www.agu.org/meetings/waissm01. html. Interpretive Summary:
Technical Abstract: Routine estimation of the land surface energy balance of a region with satellite remote sensing of land surface temperature at high spatial resolutions (i.e., 100s of meters) has not been possible due to low frequency in repeated satellite coverage and cloud cover. More frequent coverage from weather satellites such as GOES and AVHRR, and the EOS MODIS afford the opportunity for routine surface energy flux monitoring, but the surface temperature products are at coarser spatial resolutions of 1 to 5 km. At these pixel resolutions, much of the information on spatial variability in soil and vegetation states useful in precision agriculture and distributed hydrologic modeling is lost. Due to engineering and instrument design, some of these satellite sensors can provide a vegetation index product at an order of magnitude smaller pixel resolution than the land surface temperature. For example, the MODIS vegetation index product is at 250 m resolution whereas surface temperature is at 1000 m. A disaggregation procedure is developed to take advantage of this difference in pixel resolution by using the Normalized Vegetation Difference Index-Radiometric Surface Temperature (NDVI-Ts) relation at the coarser Ts scale (e.g., 1000 m) and applying it to the finer resolution NDVI data (e.g., 250 m) comprising each Ts pixel. High resolution aircraft-based remote sensing data from the 1997 Southern Great Plains Experiment (SGP97) are used to evaluate this technique over a range of spatial resolutions from ~ 20 to 1500 m. The agreement in energy flux estimates at the various resolutions using measured versus estimated Ts in a two-source-energy-balance model is discussed.