Assimilation of Groundwater Measurement Data at High Resolution into the Community Land Model 4.0

This study provides the development and verification of a methodology for assimilating groundwater depth measurements from multiple wells into a high spatial resolution Land Surface Model (LSM) using kriging based interpolation techniques. The interpolated observation data is assimilated into the Community Land Model 4.0 at 1km resolution at a test region in the Sierra Nevada Foothills in Northern California. The results show improved simulation of the monthly water table depth variation in CLM4.0 at high spatial resolution. The linear Pearson correlation coefficient between observed well data and the assimilated model is 0.810 and with the non-assimilated model it is 0.107. This improvement is most significant where the water table depth is greater than 5m. The assimilation also improves the soil moisture content, especially in the dry season when the water table is at its lowest level. Other variables including sensible heat flux, ground evaporation, infiltration, and runoff are analyzed to indicate the effect of the assimilation methodology. These variables are improved with a maximum 5% change in soil water content during summers and a difference of 25 W/m² in sensible heat. These changes may be extremely important in coupled models, and this improved simulation of groundwater with assimilation will lead to improved coupled model performance. This assimilation technique can be used where observations are sparsely distributed and can be easily adapted to accommodate other groundwater observation values, such as the GRACE. It can also be easily adapted to other LSMs that have a functional groundwater component.