CS(AR)-3/2000-2001 : Rainfall-Runoff Analysis using flood analysis and protection systems (FLAPS) Model

Abstract

In this report a conceptual distributed hydrological catchment model, FLAPS is compared with HEC-1 in simulating the rainfall-runoff processes for selected events. The study areas are selected as three basins of North eastern region of India; Myntdu leska basin (Meghalaya), Krishnai basin (Assam) and Dudhnai basin (Assam). Through the established rainfall-runoff relationships and calibrated model parameters, the results are meant to be a design tool where historical design storm with a certain recurrence can be analyzed in terms of runoff. The primary objective of the study is to ; (i) Calibrate and validate the model for Krishnai, Dudhnai and Myntdu leska basins (ii) Study the sensitivity of the model parameters with respect to different land use conditions, incipient soil moisture and transmissivity on direct runoff and total recharge (iii) Comparison of the results with HEC-1.

Methods applied in FLAPS model is based on principles and experience of an earlier model (Bengtsson, 1994) . The model has the flexibility to treat the study area either as a linked system of a number of subbasins or a lumped system, which have parameters corresponding to their physiological conditions. In linked system subbasins are connected separately either by river reaches, channels or lakes represented by corresponding routines. The estimated runoff is used with the actual runoffs far calibrating model parameters. The calibrated model parameter is used on some test events to check the validity of the parameters for the basin. HEC-1 is also a distributed rainfall -runoff modelling based on Clark's method, also known as time-area histogram. The model involves the transformation of spatially varying rainfall excess hydrograph and losses into flood hydrograph. It assumes that the rainfall excess first undergoes pure translation that is estimated by a travel time-area histogram and the attenuation is routed through a linear reservoir at the catchment outlet. The time of concentration and storage constant of the basin are optimized using HEC-1 and the parameters are validated for storm. The parameters thus obtained are used in the model to arrive at unit hydrograph for the basin.

In this report, the main effort of the calibration work was concentrated on the flood peaks and their time of occurrence. However, a few numbers of computations were also done for checking the flood volume. The verification of the FLAPS and HEC model was done for the selected events. The broad conclusions arrived after the analysis for three test catchment is that HEC model tends to overestimate the peak for low rainfall events and overestimate for high rainfall events. However, the variations are within 15 % for maximum events analysed in the study. The performances of the two models for these selected events on basis of bias, RMSE, mean absolute error, efficiency and R². As observed both the models performs well on the analysed data, however, HEC model tend to be less biased than FLAPS. An attempt has also been made in this report to check the non-linearity in rainfall-runoff response using simple storage-runoff dynamic model for selected events. The results showed degree of regression coefficient (R²) ranging from 62 to 91 % . Since only three basins are considered for the analysis the relationship of parameter 'B' (that represents the degree of non-linearity in catchment response) with the catchment area could not be established. However, It can be observed that Krishnai basin with an area of 950 Sq.km has a low B-parameter compared to Myntdu-Leska basin, a smaller catchment