Estimation of standard duration maximum rainfall by using regression models

Gauging stations of meteorological networks generally record rainfall on a daily basis. However, sub-daily rainfall observations are required for modelling flood control structures, or urban drainage systems. In this respect, determination of temporal distribution of daily rainfall, and estimation of standard duration of rainfall are significant in hydrological studies. Although sub-daily rainfall gauges are present at meteorological networks, especially in the developing countries, their number is very low compared to the gauges that record daily rainfall. This study aims at developing a method for estimating temporal distribution of maximum daily rainfall, and hence for generating maximum rainfall envelope curves. For this purpose, the standard duration of rainfall was examined. Among various regression methods, it was determined that the temporal distribution of 24-hour rainfall successfully fits the logarithmic model. The logarithmic model’s regression coefficients (named a and b) were then linked to the geographic and meteorological characteristics of the gauging stations. The developed model was applied to 47 stations located at two distinct geographical regions: the Marmara Sea Region and Eastern Black Sea Region, Turkey. Various statistical criteria were used to test the method's accuracy, and the proposed model provided successful results. For instance, the RMSE values of the regression coefficients a and b in Marmara Regions are 0.004 and 0.027. On the other hand, RMSE values are 0.007 and 0.02 for Eastern Black Sea Region.