Intercomparison of data transformation methods for the assessment of extreme rainfall

Abstract

Assessment of extreme rainfall for a given return period is considered as one of the important aspects while planning, design and management of hydraulic and civil structures, which can be estimated through deterministic, probabilistic and transformation methods. This paper presents a study on intercomparison of five data transformation methods such as Square Root, Cube Root, Logarithmic, Box-Cox and modified SMEMAX (Small, MEdian, MAXimum) applied in assessing the extreme (i.e., 1-day maximum) rainfall of Gaganbawada, Lanja and Radhanagari sites. The adequacy of fitting transformation methods to the observed data was evaluated through a non-parametric Goodness-of-Fit test, Kolmogorov-Smirnov (KS) while the Relative Mean Squared Error (RMSE) was used for the selection of most suitable method for estimation of rainfall. The outcomes of the study with KS test results and RMSE values showed that the Square Root transformation is better suited method for the assessment of extreme rainfall for all three sites.