Improvements in atmospheric water vapor content retrievals over open oceans from satellite passive microwave radiometers

摘要

High accuracy retrieval algorithm is developed for integrated water vapor content (WVC) retrieval from Advanced Microwave Sounding Radiometer 2 (AMSR2) measurements over open ocean areas. The algorithm is based on physical modeling of the brightness temperature (BT) of the upwelling radiation of the atmosphere - ocean system. The brightness temperature inversion is carried out with Neural Networks (NNs), trained on an ensemble of BTs, calculated using the data set of the atmospheric and oceanic parameters, governing passive microwave radiation of the atmosphere - ocean system. The improvement in the WVC retrieval is associated with the usage of the new ocean emissivity empirical model and exclusion of vertically polarized BT from the NNs inputs. The new model was incorporated into BT geophysical model and a new set of NNs coefficients was obtained for WVC algorithm. Though the results of the numerical experiment demonstrate higher retrieval accuracy for the old version of the NN model, real-valued validation with Global Positioning System WVC shows better performance of the new NN model with excluded vertically polarized measurements from NN inputs. The advantages of the new algorithm are the most remarkable under conditions of high winds. This is especially important in the studies of extreme weather events such as Polar Lows, Extratropical and Tropical Cyclones associated with high winds. Examples of WVC retrievals under extreme events are given.