A New Method to Retrieve Near-sea Surface Meteorological Parameters

Abstract

A new method is presented to retrieve near-sea surface instantaneous air temperature and dew point temperature. Satellite data include the air temperature and point temperature at 1000 mb are inversed by TOVS data. Sea surface temperature is derived from AVHRR data. Ship-measured data come from the observations of NEAR-GOOS (North-East Asia Regional- Global Ocean observing) experiment. Then two different artificial neural network models are developed to retrieve near- sea surface instantaneous air temperature and dew point temperature. For the single parameter neural network model (SANN) model, the root mean square (RMS) error of near-sea surface instantaneous air temperature and dew point temperature are 1.85degC and 2.59degC respectively, but those for multi-parameter neural network (MANN) model are 1.92degcand 2.67degC respectively. The results indicate that the retrieval accuracy of the SANN model is superior to that of the MANN model and prove that there is a close relationship between near- sea surface air temperature and dew point temperature.