Calibration of a soil-water balance model with a combined objective function for the optimization of the water retention curve

Measured and modelled soil-water balances as well as analyses of drought stress in forest stands provide important criteria for the planning and scheduling of adaptation options to climate change in forestry. The main goal of this study was the estimation of water-deficiency situations in forest stands in Baden-Wurttemberg with data from the nationwide nutrient-flux monitoring network using the forest hydrological model LWF-BROOK90 (HAMMEL & KENNEL 2001). This requires a calibrated model that can reproduce the natural characteristics of the water-retention curve especially during drought conditions. Accordingly, the objective of this research is to produce plot-calibrated soil-water balance models which consider the soil-water contents and soil-water potentials in different depths. The new method will be demonstrated with a beech stand in Ochsenhausen where the soil-water contents and the soil-water potentials are continuously measured in different depths and under various canopy conditions. A sensitivity analysis identified model parameters which have strong influences on model fit, i.e. the agreement between simulated and observed soil-water contents and soil-water potentials. Subsequently, model parameters identified as sensitive were optimized using the Shuffled-Complex-Evolution-Metropolis-Algorithm (SCEM-UA, VRUGT et al. 2003). As a means to achieve a good fit between modelled and observed water-content and water-potential values, information about field-retention curves was incorporated into the objective function. The calibrated model shows a good fit between observed and simulated water contents and potentials for most depths and crown situations. Good model calibration results were achieved in the evaluated time period, thus providing a good basis for further research into the estimation of past and future drought-deficit risks in forest stands.