A wide variety of models at the catchment level is used to describe and design torrents. They differ in detail regarding spatial subdivisions and regarding the physical processes resolved. The least sophisticated models are based on simple empirical approaches using basic statistics, the most complex models are based on mathematical descriptions of those physical processes that are sufficiently understood to frame them in equations. Models at the lower end of the sophistication scale make do with little and basic input, those at the upper end need considerable input that generally requires very much effort to generate and in many cases is not available due to lack of field data. By choosing the most sophisticated model regarding processes and spatial subdivisions for that matching high quality input data are available the best quality of results can be achieved for a specific catchment in general. However, a "physical" model that describes surface runoff and water flow in the soil matrix, but not along preferential flow paths will produce worse results in a catchment area in which these flows are relevant, than a conceptual model that explicitly takes account of the preferential flow paths, even if using a simple approach. Thus the selection of the appropriate model is essential to achieve good results. Independent of the sophistication of a model and the quality of the data, every model needs validation, before results can be assigned any reliability. Transfer of validated models to different catchments is theoretically easy for the most sophisticated models, however it necessitates the same large effort in acquiring input data. Simple models need to be newly calibrated when transferred to a new site. In both cases, they need new validation. Applying models to conditions of future changed climate requires a) knowledge of the new climate in the detail required of meteorological input, b) knowledge of the impact of changed climate on all other input variables of the model and c) knowledge of possible changes in basic processes that would require changes in the physics of the model. This last type of changes is not expected to occur within the range of climate scenarios presently considered, thus can be neglected at present. The impact of climate change on the environment in any catchment area can encompass as divers aspects as changes in spatial and temporal snow cover patterns, glaciated area and vegetation cover, changes in frequency distributions and seasonality of soil moisture or enhancement of erodible debris potentials. In order to describe and quantify these impacts, details of the expected meteorological changes are also needed. However, as stated in Chapter 2: "For applications in small catchments even the RCMs (regional climate models) with the highest spatial resolution (10 km) are far too coarse to resolve all relevant processes for extreme precipitation events in this scale. To resolve the small catchment scale a different type of RCMs is needed, that directly resolve convective cloud systems. This is not only a question of computer resources, but also of how the processes are represented within the models." These models are not yet available. Thus, the meteorological input necessary for to run hydrological and runoff models under climate change conditions is more readily derived from Global Climate Models (GCM) than for the mesoscale, and qualitative information can be derived for more situations than quantitative information. There are considerable gaps, as e.g. regarding area parameters of precipitation. Nevertheless, as major changes in precipitation extremes have occurred and had an impact over the last 40 years, and as further major changes are projected for the coming decades, these changes need to be taken account of in hydrological precipitation/runoff models, as well as in designing torrents. Efforts must therefore be continued to significantly improve the modelling of precipitation in GCMs, in RCMs and in subsequent statistical treatments (see Chapter 2). A further development of some aspects of hydrological precipitation/runoff models and methodologies of torrent designe would be helpful, as uncertainties play a significant role here as well. The lack of meteorological input to determine the climate change induced impacts on the environment and thus on other input parameters for the designing of torrents is partially of less consequence, as realistic worst conditions are assumed with no reference to frequency of occurrence. In many cases these input data will therefore not change with climate change.
111.83 (Klimaänderungen. Paläoklimatologie) 116.21 (Einfluß meteorologischer Faktoren) 116.3 (Untersuchungen über Wasserführung in Gewässern und Ufererosion [Unterteilung wenn nötig wie 116.2]) 111.781 (Regen, Sprühregen)
