Browsing by Author "Bárdossy, András (Prof. Dr.-Ing. habil. Dr. rer. nat. )"

Now showing 1 - 1 of 1

Open Access
Investigation of evapotranspiration concepts in hydrological modelling for climate change impact assessment
(2007) Hartmann, Gabriele Maria; Bárdossy, András (Prof. Dr.-Ing. habil. Dr. rer. nat. )
Climate change (CC) will impact water resources. Assessing the extent of these impacts in due time is an important task, as it forms the basis for decision making. Unfortunately, the extent of this forecasted impact depends very much on data and tools used for this task. Although such methods might work well with present climatic conditions, it has to be doubted whether their results can still be relied upon in a changed climate. The uncertainties in the forecasts are partly of meteorological and partly of hydrological origin. Whereas the uncertainties of GCMs are well known and often discussed, the problems of hydrological models in this context are seldom investigated. In particular the uncertainty in process representation within the hydrological models must be revised. This dissertation focuses on the representation of the evapotranspiration (ET) process, because this process will be strongly influenced by CC. For this purpose, the suitability of nine different ET models was investigated. In a theoretical investigation, the sensitivity of the ET models to only a small change in temperature was found to be very different. Thus the question had to be raised as to how the resulting ET from these models will change with the entire predicted CC. Therefore a spatially distributed hydrological model based on the HBV concept was set up and the results of the different ET models were used consecutively as input to the hydrological model. The modelling was applied on the Upper Neckar catchment, a mesoscale river in southwestern Germany with a basin size of about 4,000 km2. This catchment was divided into 13 subcatchments with different subcatchment characteristics. The suitability of the different ET approaches was checked by calibrating the hydrological model on different climatic periods and then applying the model on other climatic periods. Thus, different 10-year periods with different climatic conditions were compiled: 10 cold, 10 warm, 10 wet and 10 dry years from the time series 1961–1990 were collected. The first step was to adapt the model to the same period it was calibrated to. Then the model was applied to other 10 years, i.e. the model calibrated on for example, the cold years was used on the warm years. The transferability was also checked by applying the models on the period 1991–2000. For the investigation of the impact of CC, the calibration of the model must meet special requirements. Apart from the selection of proper periods for calibration and validation, this also concerns the establishment of a suitable objective function. Such a function is the Nash Sutcliffe efficiency. Usually it is calculated comparing observed and modelled daily values. In this study it is shown that problems in the transfer from one climatic condition to the other cannot be detected on the base of daily values. Therefore parameter sets were optimized by an automatic calibration procedure based on Simulated Annealing, which considered the model performance on different time scales simultaneously (days up to years). As the results show, some of the ET models, which work well under stationary conditions, are not able to reproduce changes in a realistic manner. The results also show that calibrating a hydrological model that is supposed to handle short as well as long term signals becomes an important task; the objective function especially has to be chosen very carefully.