Infilling of Rainfall Information using Genetic Programming

The study suggests the use of Genetic Programming (GP) based monthly model for infilling of missing rainfall records in the rainfall time series for 3 rain gauge stations in the Yarra River Basin in Australia from the available rainfall information from the nearby stations. This study compares simple linear model, polynomial model, logarithmic model and a complex model based on GP to infill the missing monthly rainfalls. The RMSE and CC values of the validation data indicate the potential of the suggested model. Further, it is also interesting to note that GP evolved mathematical models are able to predict the subtle inherent nonlinearity in the apparently predominantly linear behavior of the process. (C) 2015 The Authors. Published by Elsevier B.V.