Sensitivity of an ecological model to soil moisture simulations from two different hydrological models

Abstract

Although advanced land surface schemes have beendeveloped in the past decade, many biosphere models stilluse the simple bucket model, partly due to its efficiencywhen it is coupled with an CGCM model. In this paper,we use a sophisticated land surface model, the Simulatorfor Hydrology and Energy Exchange at the Land Surface(SHEELS), including an explicit vegetation canopy and itsphysiological control on evapotranspiration and multiple,interactive subsurface soil layers. It is found that this modelhas potential for improving the carbon cycling descriptionof a widely used biosphere model, the Carnegie-Ames-Stanford approach (CASA), especially for multiple seasonalintegrations.Verifying with observations from Oklahoma AtmosphericSurface-layer Instrumentation System (OASIS) stations, weshow that a bucket model as implemented in the CASAproduces good simulations of the seasonal cycle of soil moisturecontent, but only for the upper 15-cm soil depth, nomatter how it is initialized. This is partly due to its inabilityto include vegetation characteristics other than a fixed wiltingpoint. Although only approximate, the soil depth towhich CASA simulates storage of below-ground carbon isassumed to be about 30 cm depth, significantly deeper thanthe 15 cm depth. The bucket model cannot utilize the soilprofile measurements that have recently been made widelyavailable.A major finding of this study is that carbon fluxes aresensitive to the soil moisture simulations, especially the soil water content of the upper 30cm. The SHEELS exhibitspotential for simulating soil moisture, and hence the totalsoil water amount, accurately at every level. For the NetPrimary Production (NPP) parameter, the differences betweentwo hydrological schemes occur primarily duringthe growing seasons, when the land surface processes aremore important for climate. However, soil microbial respirationis found to be sensitive all year round to soil moisturesimulations at our 7 selected Oklahoma Mesonet stations.These suggest that for future implementing of interactiverepresentation of soil carbon in CGCMs, the accompanyinghydrological scheme should not be over-simplified.