Comprehensive framework for agricultural water management in data-scarce regions: Integration of hydrological models and remotely sensed crop type data
Year: 2026
Type: article
Abstract: Water resources are essential for human activities, with no substitutes. Agriculture consumes a significant portion of water to sustain food production. This study proposes a framework for agricultural water management by integrating hydrological modeling and remotely sensed crop type data in a data-scarce region. It focuses on the Amu River Basin (ARB), the largest watershed in northeastern Afghanistan, accounting for 57% of the region’s surface water. The WRF-Hydro stand-alone model was used to simulate daily discharge for three rivers from 2014 to 2019 and was calibrated and validated with the Global Land Data Assimilation System version 2. Statistical indicators were used to assess the model performance. The overall performance results for three rivers show a correlation coefficient (R) of 0.85–0.42, Nash-Sutcliffe Efficiency (NSE) 0.52 to −8.64, Killing-Gupta Efficiency (KGE) 0.74 to −0.56, and coefficient of determination (R 2 ) of 0.73–0.17. Three machine learning (ML) algorithms, namely, random forest, support vector machine, and gradient boosting, were ensembled via a maximum voting classifier to predict crop type maps for the study period. The model was trained using Moderate Resolution Imaging Spectroradiometer (MODIS) normalized difference vegetation index (NDVI) data from the Aqua and Terra satellites. A high-resolution 2020 crop type map from United Nation Food and Agriculture Organization (UNFAO) was used for training. The model predicted the five major crop types from 2014 to 2019 across eight elevation zones. Then irrigation water requirements (IWR) were estimated for major crops via the UNFAO Penman‒Monteith method. The estimated IWRs were combined with the spatially and temporally explicit crop maps derived from the multi-model ensemble to quantify irrigation water demand and assess its balance with available water under different management scenarios. This study addresses water management challenges in data-scarce regions, improving the performance of the WRF-Hydro model in snowmelt-influenced watersheds and enhancing ML model accuracy through ensemble techniques. The findings of the current study could provide a deeper understanding of the challenges and possible solutions in arid and semiarid climates in developing countries.
Source: Journal of Hydrology
Institution: Hiroshima University
Language: English
Subfield: Water Science and Technology
Field: Environmental Science
Domain: Physical Sciences
Open Access status: hybrid
Estimated APC paid: $4,030
OpenAlex ID: W7127319415