Research On Geophysical Parameters Retrievals Under Rain With Microwave Radiometers

The microwave radiometer which is one kind of passive microwave remote sensors, has been widely used in many fields of microwave remote sensing of atmosphere, ocean and land. Since the sea surface is the interface where the energy of ocean and atmosphere is exchanged, it is an important issue of ocean research and development.In this paper, aiming at the problem that geophysical parameters retrievals are more difficult in rainfall conditions, based on the AMSR2 brightness temperature data, semi-statistical model and statistical model method are applied to the development of a hurricane wind retrieval algorithm and global retrieval algorithms of wind speed and sea surface temperature under rain. The present work will make up for the shortage of sea surface wind speed and temperature of AMSR2 in rainfall conditions and provide data products for the forecasting and assessment of marine disasters such as the hurricane.The research content mainly includes:(1)The stability evaluation of AMSR2 brightness temperature data. The basic principle of the lowest brightness temperature analysis method is expounded. The lowest brightness temperature values and their changes in each channel of AMSR2 have been calculated and analyzed The results showed that the standard deviation of lowest brightness temperature for each channel is 0.2K to 1.3K, which is comparable to other sensors. It showed that the stability and reliability of brightness temperature data of AMSR2 radiometer have been evaluated.(2) Hurricane wind retrieval algorithm under rain. Based on radiative transfer model, considering the impact on the atmospheric absorption of rainfall, brightness temperatures sensitivity of different frequencies to the rainfall have been studied, and find a channel combination of brightness temperatureTB6.8H-0.39*TB10.7H,which is not sensitive to rain but sensitive to other geophysical parameters. Based on the channel combination, wind speed algorithm is developed and validated. Compared with the HRD wind field data, standard deviation of linear regression algorithm results is 2.7 m/s, the relative error is 11%; standard deviation of BP neural network algorithm results is 2.2 m/s, the relative error is 8%. The results of BP neural network algorithm are superior to that of the linear regression algorithm.(3) Global retrieval algorithm of the geophysical parameters under rain. The data under rainy conditions has been extracted from AMSR2 L1 B brightness temperature data, which has matched with NDBC buoy data. For the linear regression algorithm results, standard deviation of sea surface temperature is 1.9℃, and the correlation coefficient is 0.97; standard deviation of sea surface wind is3.0 m/s, and the correlation coefficient is 0.70;For the BP neural network algorithm results, standard deviation of sea surface temperature is 1.6℃, and the correlation coefficient is 0.98; standard deviation of sea surface wind is 2.4m/s, and the correlation coefficient is 0.81. The results of BP neural network algorithm are superior to that of the linear regression algorithm.