Application of a new soil data aggregation method to project irrigation water requirements at a watershed scale

The uncertainties in soil data can affect gridded crop growth simulations for the development of climate change adaptation strategies in a region. Soil data aggregation (SDA) methods, such as major soil type (MS) or all soil types (AW-All) approaches, involve a trade-off between computational efficiency and simulation reliability. Here we propose the clustered area-weighted average (CAW) method to balance efficiency and reliability. The key soil property, which was identified using a random forest analysis, was used to perform a cluster analysis of soil types for the CAW method. Gridded simulations were conducted to determine the irrigation water requirements (IR) in South Korea in the 2000s and the 2040s using the CROPGRO model. Soil data at the spatial resolution of 30 m and 1 km were used to compare the uncertainties caused by CAW, MS, and AW-All methods. Absolute difference (AD) between simulation outputs using soil data at fine and coarse resolutions was determined to represent the uncertainties of SDA methods by watershed. Uncertainty measures from the CAW method were comparable to the AW-All method (p > 0.05) when using four or more soil groups classified by topsoil drained upper limit. The proportion of site-years in which the AD value exceeded 10 mm was 7 % for the CAW method, compared to 20 % for the MS method across all periods and cultivars. Our results suggest the CAW method offers a scalable and reliable alternative for regional-scale simulations, improving the utility of crop models in irrigation planning and climate change adaptation strategies.