Projection of Future Agricultural Water Demand in South Korea according to SSP Climate Change Scenarios; [SSP 기후변화 시나리오에 따른 우리나라 미래 농업용수 수요량 전망]

초록

Agricultural water accounts for 63% of total water use in South Korea; however, practical methods for estimating agricultural water demand rely on empirical formulas based on historical data and therefore have limitations in representing non-stationary future conditions. This study developed municipal-level prediction models for agricultural water demand under Shared Socioeconomic Pathway (SSP) scenarios and projected future agricultural water demand for 160 municipalities in South Korea. For this purpose, municipal-level data on agricultural water use, climatic variables, and socioeconomic variables were compiled for the period 2000–2022, and five prediction models—Polynomial Ordinary Least Squares Regression (Poly-OLS), Partial Least Squares Regression (PLSR), Random Forest (RF), Gradient Boosting (GB), and the State Space Model (SSM)—were evaluated to identify the optimal model for individual cities. The results showed that the SSM was selected as the optimal model for 48.75% of municipalities, followed by PLSR (25%), Poly-OLS (21.88%), GB (3.75%), and RF (0.63%). These findings suggest that, when the length of the input data is limited, statistical models may provide more robust predictive performance than machine learning models. Future projections under SSP scenarios indicated that the spatial distribution of agricultural water demand remained largely unchanged throughout the projection period, whereas the SSP5 scenario projected a demand increase of up to 528,767×10³m³/year compared to the SSP1 scenario, underscoring the potential for significantly higher water requirements under a high-growth and high-carbon pathway. © 2026, Korean Wetlands Society. All rights reserved.

키워드