Artificial neural network based microwave precipitation estimation using scattering index and polarization corrected temperature

An Artificial Neural Network (ANN) based technique is proposed for estimating precipitation over Indian land and oceanic regions [30° S – 40° N and 30° E – 120° E] using Scattering Index (SI) and Polarization Corrected Temperature (PCT) derived from Special Sensor Microwave Imager (SSM/I) measurements. This rainfall retrieval algorithm is designed to estimate rainfall using a combination of SSM/I and Tropical Rainfall Measuring Mission (TRMM) Precipitation Radar (PR) measurements. For training the ANN, SI and PCT (which signify rain signatures in a better way) calculated from SSM/I brightness temperature are considered as inputs and Precipitation Radar (PR) rain rate as output. SI is computed using 19.35 GHz, 22.235 GHz and 85.5 GHz Vertical channels and PCT is computed using 85.5 GHz Vertical and Horizontal channels. Once the training is completed, the independent data sets (which were not included in the training) were used to test the performance of the network. Instantaneous precipitation estimates with independent test data sets are validated with PR surface rain rate measurements. The results are compared with precipitation estimated using power law based (i) global algorithm and (ii) regional algorithm. Overall results show that ANN based present algorithm shows better agreement with PR rain rate. This study is aimed at developing a more accurate operational rainfall retrieval algorithm for Indo-French Megha-Tropiques Microwave Analysis and Detection of Rain and Atmospheric Structures (MADRAS) radiometer.