The sensitivity of global climate model simulations to the representation of soil moisture heterogeneity

Abstract

Improving the treatment of subgrid-scale soil moisture variations is recognized as a priority for the next generation of land surface schemes. Here, the impact of an improved representation of subgrid-scale soil moisture heterogeneity on global climate model (GCM) simulations of current and future climates is carried out using Version three of the Hadley Centre Atmospheric Climate Model (HadAM3) coupled to the Met Office Surface Exchange Scheme (MOSES). MOSES was adapted to make use of the rainfall runoff model TOPMODEL algorithms, which relate the local water table depth to the grid box mean water table depth, assuming that subgrid-scale topography is the primary cause of soil moisture heterogeneity. This approach was also applied to produce a novel model for wetland area, which can ultimately be used to interactively model methane emissions from wetlands. The modified scheme was validated offline by forcing with near-surface Global Soil Wetness Project (GSWP) data, and online within the HadAM3 global climate model. In both cases it was found to improve the present-day simulation of runoff and produce realistic distributions of global wetland area. (Precipitation was also improved in the online simulation.) The new scheme results in substantial differences in the modeled sensitivity of runoff to climate change, with implications for the modeling of hydrological impacts.