Digital twin-enabled intelligent irrigation-drainage system for precision water-salt management in saline agroecosystems
Mengting Qin, Chuansong Zhang, Guorong Ma, Yongcheng Ma, Feng Xiong, Peijie Li, Kepeng Feng
Abstract
Soil salinization and freshwater scarcity are widely recognized as persistent constraints on sustainable agricultural intensification, especially in arid and semi-arid regions. This study presents a targeted approach by developing an integrated intelligent irrigation-drainage platform that combines Internet of Things (IoT)-based multi-layer environmental sensing, Long Short-Term Memory (LSTM) deep learning for predictive modeling, and digital twin-assisted management. A closed-loop irrigation-drainage infrastructure was concurrently implemented, incorporating multi-source water reuse pipelines, subsurface drainage networks, and photovoltaic-powered units for brackish water purification. Field deployment in saline-affected agricultural plots in northwestern China enabled high-frequency monitoring of soil moisture, temperature, and salinity dynamics under complex soil stratification conditions. The Long Short-Term Memory (LSTM) model demonstrated high predictive accuracy (R² = 0.97 for soil electrical conductivity and R² = 0.92 for water content), while feature importance analysis identified soil moisture, temperature, and groundwater depth were the dominant factors driving water-salt interactions. Coordinated irrigation and drainage scheduling significantly reduced soil salinity across permeable soil profiles, while closed-loop water reuse strategies enhanced water-use efficiency without increasing salinity risks. This study highlights the potential of intelligent irrigation-drainage systems integrating real-time monitoring, predictive modeling, and closed-loop water reuse to enhance salt leaching efficiency, optimize water utilization, and stabilize crop production across heterogeneous saline agroecosystems. • A digital twin-based irrigation-drainage system was developed for saline farmlands. • Real-time monitoring and Long Short-Term Memory (LSTM) prediction enabled adaptive water-salt control. • A closed-loop reuse system improved leaching efficiency and reduced salinity. • Field experiments demonstrated significant salt reduction across soil profiles.