Advanced of Mathematics-Statistics Methods to Radar Calibration for Rainfall Estimation; A Review

Abstract

Ground-based radar is known as one of the most important systems for precipitation measurement at high spatial and temporal resolutions. Radar data are recorded in digital manner and readily ingested to any statistical analyses. These measurements are subjected to specific calibration to eliminate systematic errors as well as minimizing the random errors, respectively. Since statistical methods are based on mathematics, they offer more precise results and easy interpretation with lower data detail. Although they have challenge to interpret due to their mathematical structure, but the accuracy of the conclusions and the interpretation of the output are appropriate. This article reviews the advanced methods in using the calibration of ground-based radar for forecasting meteorological events include two aspects: statistical techniques and data mining. Statistical techniques refer to empirical analyses such as regression, while data mining includes the Artificial Neural Network (ANN), data Kriging, Nearest Neighbour (NN), Decision Tree (DT) and fuzzy logic. The results show that Kriging is more applicable for interpolation. Regression methods are simple to use and data mining based on Artificial Intelligence is very precise. Thus, this review explores the characteristics of the statistical parameters in the field of radar applications and shows which parameters give the best results for undefined cases. Keywordsradar, statistical techniques, data mining, rainfall Stimation __________________________________________________*****_________________________________________________