Integrated Combination of the Multi Hydrological Models by Applying the Least Square Method

Muhammad Tayyab,Xiaofan Zeng,Na Zhao,Rana Adnan,Jianzhong Zhou

doi:10.19026/rjaset.10.2561

Abstract

Different hydrological models show different outputs for specific catchment, thus combining all the models in suitable way is very important to improve the forecast. To solve the issue, researchers have applied different techniques which ranges from simple inter-comparison of different hydrological models to extended combination of hydrological models. The aim of this research is to find a suitable and applicable combination technique, by applying least square method to get more valuable flood forecasting results for the Jinshajiang River basin. The combination forecast has been compared with the results of the three models individually, based on the comparison of the simulation outputs and the Nash-Sutcliffe efficiency and Correlation coefficient. The result showed that the performance of combine system of three conceptual hydrological models including Xin’anjiang model, Antecedent Precipitation Index (API) model and Tank model is much more reliable as compared to their individual performance.

Highlights

For precise evaluation or prediction of floods, numerous hydrological models have been developed by different researchers and scientists according to need
These models includes empirical Black-box models, which are generally based on hydrograph theory of Sherman (1932), to further complex, sophisticated conceptual models, like the Stand ford watershed model (Crawford and Linsley, 1966) and the Sacramento model (Burnash et al, 1973) etc. and extra physically based models like, the ‘‘SHE’’ model (Abbott et al, 1986a, b)
The use of Least Square Method (LSM) in a current statistical outline can be sketched to Galton (1886) who used it in his study on the heritability of size which laid down the foundations of correlation and regression analysis

Summary

Introduction

For precise evaluation or prediction of floods, numerous hydrological models have been developed by different researchers and scientists according to need. Even though a verity of hydrological rainfall-runoff models exists, yet no one can claim any of them as an “Ideal model” which has all properties to perform well other models in all conditions, regarding data need, catchment area and all other circumstances. To overcome these problems researchers has found another very useful but inadequate way by inter-comparison studies of models. In its place of trusting on one distinct hydrological model, or changing model one from another, a substitute methodology could be applied to produce discharges all together from a numeral of changed hydrological models and to chain the predictions in a best mode (Shamseldin et al, 1997)

Objectives

Methods

Results

Conclusion