Study On The Evaluation Of The FY-3B Microwave Radiation Imager On-orbit Data And The Retrieval Of Sea Surface Temperature, Wind Speed

Microwave remote sensing, could get through clouds, work in the all-time, andall-weather, and has been one of main technologies of observation and study of theocean. The application of satellite microwave radiometer, not only can make real-timemonitoring of the ocean come true, but also may obtain long-time and wide range ofsea observations. FY-3B MWRI is a satellite microwave scanning radiometer, whichwas developed by China independently, and was launched in November2010.Nowadays, the study of the FY-3B MWRI on-orbit data verification, retrieval andapplication research of oceanic parameters, has been in its infancy. Therefore, thepaper would primarily have carried out the study of the assessment of MWRI on-orbitdata, and the retrieval of sea surface temperature, wind speed.The main contents and conclusions of this study are as follows:1、Microwave radiative transfer equation, not only is the theoretical basis of themicrowave radiometer observations, but also is the physical foundation ofSpace-borne Microwave Radiometer in-orbit data quality assessment and marinephysical parameters retrieval.First, the paper has systematically described theocean-atmosphere microwave radiative transfer model, carried out the theoreticalsimulation and analyzed the effect of oceanic-atmospheric parameters, physicalmodel and frequency and on. And then, after the matched data sets of between FY-3Band AMSR-E observations、NCEP reanalysis data were established, combined withocean-atmosphere microwave radiative transfer theory, the paper has carried outcomparative analysis on the measurement of each channel, and completed theevaluation of FY-3B MWRI on-orbit data. In the global ocean and land surface, theobservations of FY-3B MWRI and AMSR-E are consistent. The correlationcoefficients of the data in land area were lager than0.97all channels, and correlation coefficients were larger than0.90in marine area, however, the RMS of marine areadata were smaller, about2.0K. Considering the effects including incidence angle, theFY-3B MWRI observations and its reference values in the marine areas have showngood consistency, correlation coefficient of all channel data were above0.91, and theRMS were about1.5K.2、Based on FY-3B MWRI observations and NDBC buoy measurement, thepaper has made use of multivariate statistical regression methods and neural networkmethods to study the retrieval of sea surface temperature and wind speed, thenanalyzed the differences among the seasons, and at last retrieved real-time globalocean SST and Wind. Compared with NDBC buoy data, the results of the full channelneural network algorithm for SST and Wind have shown best, respectively, the RMSwere1.42°C and1.33m/s; multivariate statistical regression retrieval algorithm,which is a very fast and effective way, the RMS of full channel algorithm were2.84°C and2.06m/s for SST and Wind. The results of the retrieval of SST and Wind weredifferent in different seasons. For SST, the retrievals in autumn and summer werebetter than spring and winter, and the RMS were smaller; for Wind, the retrievalswere same as wind.