Submitted to: Remote Sensing of Environment
Publication Type: Proceedings
Publication Acceptance Date: 5/5/2000
Publication Date: 9/8/2000
Citation: Interpretive Summary: Reliably estimating surface energy fluxes from agricultural fields requires knowledge of surface cover to characterize surface roughness and resistance to heat transfer. Typical surfaces include green vegetation, senescent vegetation, and bare soil. Using spectral vegetation indices, remote sensing measurements in the visible and near-infrared bands (VNIR) can detect green vegetation but cannot distinguish senescent vegetation from bare soil. A method using multi-band thermal infrared data, known as thermal emissivity contrast, can make this distinction. The method and its application in a Southern Great Plains 1997 study at El Reno, Oklahoma, are described. By combining emissivity contrast data with a VNIR vegetation data, it is possible to distinguish between green vegetation, senescent vegetation, and bare soil.
Technical Abstract: A remote sensing method utilizing multi-band thermal infrared (8-12um) imagery is described that discriminates between senescent vegetation, and bare soil. This discrimination is achieved by computing thermal band emissivities from a temperature-emissivity separation algorithm and then classifying surface features based upon spectral emissivity contrast. In a study of rangelands and wheat fields in central Oklahoma, the range of these spectral emissivities is diagnostic of the presence, or the absence of, surface vegetative cover. A large range of emissivities, approximately greater than 0.03, is indicative of bare soil; while a low range, less than 0.02, is indicative of vegetative cover. When knowledge of the emissivity range is combined with a vegetation index, such as NDVI, the surface may be classified by a ternary system: bare soil, green vegetation, and senescent vegetation. Discrimination between bare soil and soil covered with senescent vegetation using emissivity contrast should be feasible in other settings. The benefit of this technique is that heat flux predictions can be based upon a more accurate surface representation than otherwise provided by visible and near-infrared land classification schemes.