A Method for Correction of Tropical Rainfall Measuring Mission Satellite Temperature Network in Mazandaran Province

Abstract

Accurate spatial estimation of temperature is very important in meteorological research. This study investigated the efficiency of temperature products of the Tropical Rainfall Measuring Mission (TRMM) satellite in estimating temperature in Mazandaran Province, and its accuracy were compared with inverse distance weighting and Co-Kriging interpolation methods. Finally, a new method was proposed to improve the accuracy of temperature estimation by combining the TRMM temperature products and terrains. Data recorded at 25 meteorological stations and 26 monthly and annual TRMM satellite images in 2012 and 2013 were used. The results showed a significant correlation between temperature data and satellite products, latitude, and altitude in significance level of 95%. Analyzing error indices showed that TRMM products have underestimation error that this bias error contributed to about 60% of error in these satellite images. Despite the larger error of TRMM products than interpolation methods, the regression analysis results demonstrated the superiority of satellite temperature products over interpolation methods. Furthermore, higher correlation of observed and estimated data showing that satellite products give a better understanding of cold and hot points of the study area despite its underestimation error. Combining satellite temperature products with influential covariates of altitude and latitude in the regression equation reduced the temperature estimation error of the TRMM products by 80%. The estimation precision increased over 70% compared to other temperature interpolation methods. Analyzing isotherm maps indicate the higher temperature of eastern coasts than western coasts. Moreover, evaluating different temperature estimation methods showed the higher precision of the methods that involved covariates than other methods.