Environmental Risk Assessment of Intensive Livestock Systems Using Hybrid AI–GIS Models

Abstract

The rapid intensification of livestock production has increased environmental challenges, including water pollution, soil degradation, greenhouse gas emissions, nutrient loading, and ecosystem deterioration. Conventional environmental risk assessment methods often rely on field surveys and statistical analyses, limiting their ability to process large-scale heterogeneous datasets and complex spatial relationships. To address these limitations, this study proposes a Hybrid Artificial Intelligence–Geographic Information System (AI–GIS) framework for comprehensive environmental risk assessment in intensive livestock systems. The framework integrates environmental monitoring data, remote sensing imagery, climatic variables, land use, soil characteristics, hydrological information, and livestock population records into a unified decision-support platform. GIS is employed for spatial data integration, hotspot detection, and environmental risk mapping, while AI models, including Random Forest, XGBoost, Artificial Neural Networks (ANN), and Long Short-Term Memory (LSTM), predict multiple environmental risks. The predicted outputs are combined with GIS to generate high-resolution risk maps that classify regions into very low, low, moderate, high, and very high-risk categories. Experimental results demonstrate that the proposed framework achieves an overall classification accuracy of 98.86%, with a Precision of 98.54%, Recall of 98.27%, F1-score of 98.40%, and RMSE of 0.081. The framework provides accurate spatial risk visualization and supports evidence-based environmental management, sustainable livestock production, and climate-resilient agricultural planning.