Satellite estimate of specific humidity over the ocean using microwave radiometry

Aim of this work is to retrieve specific humidity over ocean using microwave data. The study focuses on 12 months period of year 2004, at daily scale. First, we give an algorithm which links brightness temperatures provided by microwave sounder radiometers AMSU-A and AMSU-B on board NOAA satellites, to PIRATA buoys (located in differents points over the Atlantic) data, using linear regression method firstly, and neural networks method after. After, we do same work but using brightness temperatures of SSMI radiometers, in order to compare performances between AMSU et SSMI radiometers. Third phase consist to add sea surface temperature (SST) provided by MODIS satellite, to brightness temperature in order to improve algorithms, and we do comparison between results giving by the two instruments. Comparison shows that AMSU radiometer is most performant than SSMI. The retrieval regression method had rms differences of 0.9g/kg for AMSU. When we add SST rms differences is 0.83g/kg. Neural networks methods had rms differences of 0.75g/kg. These results show the important capacity of AMSU radiometer to restitute specific humidity over the ocean.