Flood Peak Equations Based on Partial Duration Series (PDS) Approach of Rainfall Data Selection in Lack-Data Agricultural Catchment

Abstract

The present work aims to develop a flood peak equation because of data limitations due to the impact of damage to the streamflow measuring instrument as a result of the 2015 floods at an agricultural catchment in Sulawesi, Indonesia. Hydrologic data for the period 2002-2014 obtained from two hydrologic stations and one hydrometric station were applied to establish the research variables. Three variables were determined using the frequency analysis approach: design rainfall generated from partial duration series (RDP) and annual maximum series (RDA) of daily rainfall data, and design streamflow predicted from annual maximum series of daily streamflow data (QDA). Four types of frequency distributions are tested to determine those variables, consisting of Normal, Log Normal, Log Pearson Type III and Gumbel distributions. The third distribution was selected for determining all the variables with the largest difference in χ2 and Δ values based on Chi-squared and Kolmogorov–Smirnov tests respectively. The design streamflow equation that represents the peak of the flood was formulated by substituting the QDA with an equation generated from the regression analysis of those three variables. A streamflow peak equation was successfully developed as a function of RDP in the form of a power equation with excellent performance measured using Mean Absolute Error (MAE) and Correlation Coefficient (r). This equation could be applied in all of the catchments by accommodating the area's weighting factor of the sub-catchments