Comparative Evaluation of Three Penman-Monteith Variant Models for Groundwater Level Estimation in Jeju Island

Abstract

Accurate estimation of groundwater levels is crucial for effective water resource management, particularly in regions like Jeju Island, where groundwater is the primary water source. This study emphasizes the importance of preprocessing meteorological observations to address the temporal disconnect between rapid meteorological fluctuations and the slower responses of groundwater systems. Key factors such as cumulative effects, delayed reactions, and seasonal variations were considered during preprocessing to improve the reliability of evapotranspiration estimates. Three Penman-Monteith evapotranspiration models (PM, FAO-24, FAO-56) were evaluated using pre-processed data, including accumulated precipitation (100-day), temperature (20-day), and other meteorological parameters. Validation against 2023 groundwater level data demonstrated that the FAO-56 model preprocessing achieved the best performance, with the highest correlation (r = 0.89), lowest mean squared error (MSE = 0.10), and smallest error range (overestimation: +1.1m, underestimation: -1.4m). These results highlight that accurate groundwater level estimation relies on proper preprocessing of observations rather than solely optimizing model operations. The findings provide valuable insights for enhancing groundwater monitoring and sustainable water management in areas with complex geological and hydrological conditions.