Monthly prediction of streamflow using data-driven models

Journal of earth system science

Seyed Abbas Hosseini^2¹  Behrouz Yaghoubi^1²  Sara Nazif^3³ 

[1] Department of Civil Engineering, Science and Research Branch, Islamic Azad University, Tehran, Iran.^2;Department of Water Engineering, Kermanshah Branch, Islamic Azad University, Kermanshah, Iran.^1;School of Civil Engineering, College of Engineering, University of Tehran, Tehran, Iran.^3

关键词: Run-off prediction;    GA-SVR;    GA-ANN;    GMDH;    HYMOD;   

DOI  :  

学科分类：天文学(综合)

来源: Indian Academy of Sciences

【 摘 要 】

The estimation of river run-off is a complex process, but it is of vital importance to the proper operation of reservoirs, the design of hydraulic structures, flood control, drought management and the supply of water and electricity. The high uncertainty in rainfall–run-off modelling and lack of data has made the development of rainfall–run-off models with acceptable levels of accuracy and precision challenging. Furthermore, the rainfall–run-off models commonly do not provide an explicit relationship between run-off and other variables to be used for run-off-related investigations. To overcome the knowledge and information shortage in rainfall–run-off modelling, data-driven models have been used instead of conceptual models for the development of rainfall–run-off models. In this paper, three data-driven models, the genetic algorithm-support vector regression (GA-SVR), genetic algorithm-artificial neural network (GA-ANN) and the group method of data handling (GMDH) have been used to predict the monthly run-off of the Gavehroud basin. Their performances are compared with a conceptual hydrological model (HYMOD) whose parameters are calibrated using the GA. To this end, the monthly data on precipitation, temperature and run-off at the Gavehroud basin over 49 yr (1960–2009) were analysed. Evaluation of the results using performance evaluation indicators showed that the hybrid model of GA-SVR provided better accuracy in predicting the nonlinear behaviour of flow data than the GA-ANN, GMDH and HYMOD.

【 授权许可】

CC BY