Artificial Neural Network Approach in Rain Rate Estimation by Using AMSU-B and MHS Data

Low-Earth Orbit (LEO) satellites provide passive microwave (PMW) information about the hydrometeors. PMW data, which can be obtained from Advanced Microwave Sounding Unit-B (AMSU-B) and Microwave Humidity Sounder (MHS), has a direct interaction with precipitation. The methods developed using PMW data are sensitive to the concentration of ice particles or water droplets associated with precipitation. Therefore, this data is more appropriate to estimate the rain rate.

Artificial Neural Network (ANN) is one of the techniques of Machine Learning (ML). It has been developed by imitating the stimulation and information received by the sense organs through the neurons in the computer environment. Today it is one of the technics of being used in almost all computational science fields. ANN can also be thought of as a black box that processes given inputs and generates outputs. This system process data in parallel, and it also learn coefficients among neurons with the principle of minimalizing and renewing mistake. In other words, ANN uses the method of trial and error.

The aim of this study is to develop an estimation model based on ANN by using brightness temperature data from five different channels (89, 150, 184, 186, and 190 GHz) of the AMSU-B and MHS mounted on NOAA 15-18 and METOP-A satellites. Brightness temperatures obtained by different channel frequencies are used as an input for the ANN model, and the rain rate values are tried to be estimated. In addition, the model results are compared with the rain gauge data and rain rate values of the 183 WSL Fast Rain Rate Retrieval Algorithm.

Keywords: Artificial Neural Networks, Precipitation, Rain rate, Turkey.