Arctic polar algorithms for atmospheric water parameter retrievals from satellite passive microwave data

Abstract

The new algorithms for retrieval of total atmospheric water vapor content (Q) and total cloud liquid water content (W) from satellite microwave radiometer data, applicable for the Arctic Basin. These algorithms are based on the neural networks (NNs) regression technique employed for the inversion of a radiative transfer equation (RTE). For the algorithm development the numerical integration of RTE was carried out for the channel characteristics of a Special Sensor Microwave/Imager (SSM/I) and Advanced Microwave Scanning Radiometer (AMSR-E), and brightness temperatures (TB) were simulated for non-precipitating conditions over the open ocean. Sets of sea surface temperatures (Ts), surface winds and radiosonde reports collected by Russian research vessels served as input data for the integration. Only data with Ts less than 15degC were selected for algorithm development. Simulated radiometer noise was added to the calculated values of TB. Once developed, using theoretically simulated values of TB, the Q algorithms were then validated both for SSM/I and AMSR-E retrievals using satellite radiometric measurements collocated in space and time with polar station radiosonde data. The resulting SSM retrieval error proved to be 1.1 kg/m2, AMSR-E retrieval error -0.9 kg/m2. Considered case study was the polar low in the Norwegian Sea occurred 30-31 January 2008. NOAA AVHRR, Terra and Aqua MODIS images, QuikSCAT-retrieved wind fields, Envisat ASAR images as well as weather maps were used as ancillary data to passive microwave retrievals to study this phenomenon.