Comparison of models for estimation of groundwater recharge, using data from a deep weighing lysimeter

Most conceptual models of catchment hydrology include algorithms for estimation of recharge to groundwater, which is then routed to stream base flow and/or deep regional groundwater. The majority of models attempt to simulate the process of percolation to groundwater from an overlying soil water store, which may be conceived as a single store or a series combination of stores. However, these models have seldom been tested against field observations, and therefore, the purpose of this study is to compare the performance of 13 single store and two-store models, selected from the literature, using an 11 years daily record from a weighing lysimeter 2.4 m deep at Coshocton, OH, supporting a pasture grass. The percolation data are measured directly, while the volume of water in the soil column is measured by weighing the monolith every 5 min. Use of data from three adjacent nonweighing lysimeters enabled an assessment of the spatial variability of groundwater recharge. The results of the study show that five models perform well in simulating recharge to a level consistent with spatial variability over a scale of a few metres, but did not perform as well in simulating variations of total soil water in the column, particularly during periods of high evapotranspiration. All the models, however, tend to underestimate the sharp peaks in recharge which may occur in extreme events.