Validation of a soil water balance model using soil water content and pressure head data

Abstract

Abstract:The validation of soil water balance models and the evaluation of the quality of the model predictions at field-scale require time-series of in situ measured model outputs. In our study, we have validated such a model using a 6-year period with time-series of automatically recorded, daily volumetric soil water contents measured with the time-domain reflectometry with intelligent microelements (TRIME) method and daily pressure heads measured with tensiometers. The comparisons of simulated with measured soil water contents and pressure heads were analysed using the modelling efficiency index (IA) and the square root of the mean square error (RMSE) in order to evaluate the prediction quality of the model. In our study, IA and RMSE, obtained either from the comparison of simulated with measured soil water contents or the comparison of calculated with observed pressure heads, in some cases lead to different results regarding the evaluation of the simulation quality of the soil water balance model. For example, a good fit between simulated and observed soil water contents does not necessarily result in a comparably good fit between the corresponding calculated and measured pressure heads. Therefore, a combined use of both measurement techniques, which takes into account their respective advantages and disadvantages, gives a more complete overview on the simulation quality of the soil water balance model than the single use of one of those techniques.

Biomass Production with Willow and Poplar Short Rotation Coppices on Sensitive Areas—the Impact on Nitrate Leaching and Groundwater Recharge in a Drinking Water Catchment near Hanover, Germany

“…Data quality assessment of water tension time series was done following the protocol described in Wegehenkel (2005) [33]. Average soil water tensions were excluded from the model performance evaluation for periods where the values of one or more tensiometers had to be rejected, i.e., because dry soil conditions beneath the measuring limit (−850 hPa).…”

“…Here, the point in time of water flow breakthrough in 100 cm could not be monitored, due to measuring errors caused by the dry subsoil. However, coefficients of determination (Table 6) are relatively good for pressure head data [33] and the absolute deviation from measurements is low. The fact that measured values of soil water storage capacities (retention curves) were used in the simulations and throughfall was only slightly overestimated (ME; 1.4 mm, Table 6, possibly due to wetting and evaporation losses during measurements), strengthens our belief that the estimations of the water balance components are reliable.…”

Modelling hourly evapotranspiration and soil water content at the grass-covered boundary-layer field site Falkenberg, Germany

“…This comparison resulted in a RMSD of 0.025 cm 3 cm -3 between soil water contents measured by TDR and those estimated by gravimetry (Beyrich and Mengelkamp 2006). In other studies, corresponding RMSD values ranged from 0.01 to 0.05 cm 3 cm -3 (Wegehenkel 2005, Paige and Keefer 2008, Abbas et al 2011, Evett et al 2012. Therefore, it is unlikely that significant measurement errors by TDR probes are a reason for these discrepancies between simulated and measured soil water contents at 45, 60 and 90 cm depth.…”

“…These findings indicate appropriate conditions for the application of the approach using compensatory root-water uptake, (Figs 4-7). In other studies, the comparison of simulated and measured daily soil water contents gave IA values of 0.70-0.98 and NS of between 0.21 and 0.93 (Herbst et al 2005, Wegehenkel 2005, Giertz et al 2006, Luo and Sophocleous 2010. At an international workshop, an evaluation of the simulation quality of 18 different complex agro-ecosystem models obtained from the comparison of daily simulated volumetric soil water contents with those measured by TDR resulted in an IA value of 0.50-0.93 (Kersebaum et al 2007).…”

“…The first consists of a physiologically-based growth model for grass cover combined with a simple soil water balance model obtained from the modelling system Theseus (Wegehenkel 2005). This model was only applied for calculation of the time series of leaf area index (LAI) that is used as an input to the second model used in our study, the well-known Hydrus-1D model (Šimůnek et al 2008).…”