Evaluating Daily Soil Moisture and Water Balances in a Semi-arid Catchment in Southeastern Australia Using the IBIS Land-surface-scheme

Soil moisture is an important variable in the climate system, and is intrinsically linked with the climate and vegetation. Both soil properties and vegetation exert great influence on the hydrological cycle via infiltration and evapotranspiration, especially in arid and semi-arid areas with thin soils and no groundwater recharge. In this study, we analyzed a 3-year (2005~2007) dataset of daily soil moisture at five closely spaced monitoring sites in the Stanley Catchment using a bio-hydrological land surface model (IBIS), driven with daily meteorological data. Both IBIS's static and dynamic vegetation modes were examined in order to investigate the complexity of vegetation representation required to adequately simulate soil moisture and the water cycle for the surface 30cm and full soil profile (site-specific from 30 to 90cm deep). In the first set of simulations the static vegetation model was used with constant leaf area index (LAI). IBIS could not reproduce the soil moisture time series using field observed soil texture (i.e. percent sand, silt and clay). This was due to the poor performance of the built-in pedotransfer function at our site. In the second set of simulations we estimated the six soil hydraulic parameters in IBIS using the GLUE optimization algorithm, and the modeled soil moisture agreed well with the observations (R=0.96, RMSE=0.03v/v, NSE=0.89). For each of the five sites the calibrated parameter sets were different, implying significant spatial heterogeneity in the soils (including soil type, soil cracking and macropores) for sites only several hundred meters apart. In a third set of simulations the dynamic vegetation mode (i.e. time varying seasonal plant phenology and interannual vegetation dynamics) was used. The vegetation type and characteristics evolved with time in response to the climate variability during the simulation period (including a 150-yr spin-up prior to 2005). Sensitivity analysis showed that the soil functional properties (especially saturated hydraulic conductivity) strongly influenced the evolution of the vegetation to an equilibrium state. In the final set of simulations, we used the calibrated soil parameters from the static vegetation runs with a simple parameterization of disturbances (the original forest has been cleared at our site), in order to evaluate the capability of the model to reproduce the vegetation structure given the observed level of vegetation disturbance. This will further allow us to explore if dynamic vegetation actually improves our predictions of soil moisture and hydrology.