A NEW METHOD TO DETERMINE NEAR-SEA SURFACE AIR-TEMPERATURE BY USING SATELLITE DATA

We present a new algorithm with which to determine near-sea surface ai r temperature from satellite observations. A relationship between air temperature, sea surface temperature (SST), wind speed, and humidity i s obtained from the aerodynamical equation and the bulk formula. We so lve air temperature from this relationship by giving other variables w ith the observations and without explicit assumption of boundary layer parameters. Our new method is validated using observed monthly mean d ata at the Japan Meteorological Agency (JMA) and Tropical Ocean-Global Atmosphere (TOGA)-Tropical Atmosphere Ocean (TAO) buoys. Air temperat ure and sensible heat flux can be determined with accuracies of 0.0 de grees +/- 1.2 degrees C and 0.1 +/- 8.7 W/m(2), respectively, when SST , wind speed, and humidity are given by in situ observations. In order to retrieve near-sea surface air temperature and sensible heat flux f rom satellite data, SST is obtained from multi channel sea surface tem perature (MCSST) of advanced very high resolution radiometer (AVHRR), wind speed is obtained from special sensor microwave/imager (SSM/I), a nd humidity is obtained from SSM/I together with the empirical equatio n proposed by Liu [1986]. An error of monthly mean satellite-derived a ir temperature is -0.3 degrees +/- 3.1 degrees C, and that of sensible heat flux is 10.0 +/- 37.6 W/m(2). Errors of both satellite-derived v alues are enlarged, possibly because the satellite observations have s easonably systematic error. Time evolution of the interannual variatio n of air temperature estimated by satellite agrees well with that of i n situ measurements. The estimation error of the interannual variation of air temperature is 0.0 degrees +/- 1.5 degrees C, and that of sens ible heat flux is 0.3 +/- 19.0 W/m(2). At an interannual timescale the satellite-derived sensible heat flux catches the time evolution of th e in situ observation. Both the air temperature and the sensible heat flux obtained by using our method have fewer errors than estimations m ade by assuming relative humidity as the climatological mean value.