TR-42 : Monthly stream flow simulation for Mahanadi basin using HEC-4 model

Abstract

HEC-4 is a monthly streamflow simulation program developed by the Hydrologic Engineering Centre, , U.S. Army Corps of Engineers. The program analyses monthly streamflows at a number of inter-related stations to determine their statistical characteristics and generate a sequence of hypothetical streamflows.

In this study, HEC-4 model has been used to simulate monthly streamflows at some sites of Mahanadi basin based on single station analysis, two station analysis, and three station analysis to identify the best combinations of stations for flow generations so that the historical mean, standard deviation and serial correlation coefficients are reproduced by the generated series. The monthly flow series analysed belong to the sites of Hirakud and Naraj on Mahanadi river, Salebhata on Ong river and Kantamal on Tel river, both tributaries to Mahanadi river.

The analysis of Hirakud flow series was carried out in four different ways, viz. by considering the entire available flow data at Hirakud for the period of 1946-82, by considering only the flow data prior to the construction of the Hirakud dam 1946-1957) by considering only the flow data after the construction of the dam (i.e. 1958-1982) and by considering the flow data corresponding to the period of flow available at sites on Ong and Tel rivers, i.e. the data corresponding to the period of 1972-82. However, in considering the two stations and three stations analysis, the flow data of entire period of observation for 37. years (1946-82) were used alongwith the available data on other sites.

The best combinations of stations were identified from the analysis for each site, and with that combination the monthly flows were generated at each site for 37 years, the maximum length of flow series available at Hirakud and for 100 years.

it is inferred from the study that there is no difference between the simulated results of two station and three station models, and two station analysis based models are better able to reproduce the statistical characteristics of the historical series than the single station analysis based models.