Generating Monthly Stream Flow Using Nearest River Data: Assessing Different Trees Models

doi:10.1007/s11269-019-02299-4

Return

Generating Monthly Stream Flow Using Nearest River Data: Assessing Different Trees Models

2019-06-07

PRE

Anas Mahmood Al-Juboori *

DOI：10.1007/s11269-019-02299-4

Original

Original request for help

Save

Abstract

En 中文

The data-driven techniques have gained more attention in stream flow prediction in recent years. In the current study, three different trees models (random forest, TreeBoost, and decision tree) were applied to predict the monthly stream flow for a river using the nearest river monthly stream flow data as external predictor variables. The cross-correlation function was used to select the optimum input predictor variables for the proposed models. A different scenario for selecting the optimal input predictor variables combination was studied. The performances of the models were evaluated by using root mean squared error and Nash and Sutcliffe coefficient indices. The Greater Zab River and the Lesser Zab River in Iraq were chosen as a case study to apply the proposed models. The monthly stream flow data for the Greater Zab River were generated using the monthly stream flow data for the Lesser Zab River, and the monthly stream flow data for the Lesser Zab River were generated using the monthly stream flow data for the Greater Zab River. The results showed a high performance of the random forest model to generate the monthly stream flow in comparing with the TreeBoost and decision tree models. The Nash and Sutcliffe coefficient is 0.84 and 0.89 in validating periods to generate monthly stream flow data using the random forest model for the Greater Zab River and the Lesser Zab River, respectively.

Keywords:

Stream flow modeling

Data-driven techniques

Decision tree

Random forest

And TreeBoost

AI Summary

Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.