A Review Study on Stationary and Non-Stationary IDF Models Used in Rainfall Data Analysis around the World from 1951-2020

This article focuses on an overview of the processes of generating rainfall intensity-duration-frequency (IDF) models, the different types and applications. IDF model is an important tool applied in the design of either hydrologic or hydraulic design such as prediction of rainfall intensities to estimate peak runoff volumes for mitigation of flooding. IDF models evolved from stationary – parametric (empirical) and non-parametric (stochastic) models, to non-stationary models in which variables vary with time. Each category controls the ways models predict rainfall intensities, and reveals their strength and weaknesses. IDF models must therefore, be chosen in terms of the project objective, data availability, size of the study, location, output needed, and the desired simplicity. For instance, while the parametric model predicts better for shorter durations and return periods only, the non-parametric models predict better for both shorter and longer durations and return periods. For projects requiring change of input data over time and evaluation of uncertainty bounds, risk assessment, including incorporation of changes in extreme precipitation, the non-stationary model approach must be selected. Also, of importance for catchments without rainfall amount and corresponding duration records but has daily (24-hourly) record of rainfall depth, the Indian Meteorological Department (IMD) method of shorter duration disaggregation can be adopted to generate in-put data for the development of IDF curves for such a location. Therefore, each model type has limitations that may make it unsuitable for some projects. Reviewing input data and output requirements, and simplicity are all necessary to decide on which model type should be selected.