Abstract
Geostationary Meteorological Satellite Visible/Infrared Spin-Scan Radiometer (GMS VISSR) images have been used to estimate diurnal variations of land surface temperature distributions over the Tibetan Plateau. The infrared split-window algorithm developed for NOAA Advanced Very High Resolution Radiometer (AVHRR) has been adapted for this purpose. Radiative transfer simulations are carried out to obtain the atmospheric transmittances and the difference temperatures that are involved in the internal coefficients of the split-window algorithm. Precipitable water distribution that is required by this algorithm is estimated from 6.7-μm brightness temperature utilizing spectral characteristics of the GMS water vapor channel. Cloud removal plays an important role in the surface temperature retrieval process. To identify convective cloud activity, many researchers use satellite infrared measurements with a fixed threshold technique. In this study, it is necessary to remove not only convective clouds but also warm clouds. For this purpose, a variable threshold technique is proposed. The threshold varies both seasonally and diurnally, and its value is determined on the basis of surface observations. With a variable threshold, it becomes possible to remove relatively warmer clouds in summer and detect colder ground surfaces at nighttime in the winter. The results of comparing estimated surface temperature from GMS data using this algorithm with in situ surface measurements show correlations around 0.8.
Corresponding author address: Yuichiro Oku, Disaster Prevention Research Institute, Kyoto University, Gokasho, Uji, Kyoto 611-0011, Japan. okusan@storm.dpri.kyoto-u.ac.jp