Application of ANN-Based Streamflow Forecasting Model for Agricultural Water Management in the Awash River Basin, Ethiopia

Abstract

This paper presents the application of a long-term streamflow forecasting model developed using artificial neural networks at a stream gauging station in the Awash River Basin, Ethiopia. The gauging station is located above the headworks of a large irrigation scheme called the Middle Awash Agricultural Development Enterprise (MAADE). Based on the forecasted streamflow time series and water requirements for irrigation and environmental purposes, appropriate agricultural water management strategies have been proposed for the irrigation scheme (MAADE). The water management strategies which were evaluated in this study are based on different scenarios of abstraction demands. These demands were formulated based on a range of options for agricultural development and change in MAADE. The scenarios evaluated were based on such factors as the existing planting patterns, changing planting dates, changing crop varieties and reducing the area under cultivation. An appropriate scenario of agricultural development was decided on the basis of the modified flows in the river vis-a-vis the trigger/threshold value established at the Melka Sedi stream gauging station. Considering all the scenarios, it is suggested that a 1–24% reduction in the area currently irrigated in the scheme will ensure a reliable supply of water to the scheme throughout the growing season and will provide sustainable environmental flow in the river.