Land surface temperature estimation from AVHRR thermal infrared measurements: An assessment for the AVHRR Land Pathfinder II data set

Improved processing algorithms are being developed, under the AVHRR Land Pathfinder II (ALP-II) Project, to process the entire Advanced Very High Resolution Radiometer (AVHRR) archive, including methods to produce surface temperature (Ts) measurements from the thermal infrared portion of the observation record. This new data set should assist in the monitoring of global terrestrial environmental conditions on seasonal to interannual time scales. The impact of both the calibration and filter function variations between National Oceanic and Atmospheric Administration (NOAA)-AVHRR sensors is addressed in this proposed ALP-II approach. Accounting for these sensor factors reduces the error of intersatellite comparison of surface temperatures by as much as 2.3 K. The environmental factors were also addressed, including the surface emissivity, large range of atmospheric water vapor (WV) across the globe, and the potential for large surface-to-air temperature differences for land. To account for both the sensor and environmental factors in the ALP-II data set, we derive, from the MODTRAN-3 model, two sets of new coefficients, for low and high WV in the atmosphere. Two types of split-window equations were analyzed, the linear and quadratic forms of the land split-window equations. Our MODTRAN-based analysis suggests that the quadratic form is more accurate than the linear form, especially for high WV contents. These coefficients can be used in our LST algorithm in combination with a global emissivity map and WV data from Global Circulation Model (GCM).