Simulation of snowpack and discharge in an alpine karst basin

Doctoral Thesis , 2002

Severe floods regularly occur at different places all over the world including the European Alps and Prealps. They are responsible for important damages and a large number of fatalities. Consequently, research activities in physical and engineering sciences have long focused on understanding the underlying physical processes and on developing methods to reproduce and predict floods. Despite considerable progress in both fields over the last few years, a number of questions still remain open. The frequent over- or underestimation of both structural and non-structural measures for flood mitigation are, however, only partially caused by this fact. Recent and more sophisticated methods of flood estimation do not often penetrate to everyday hydrological practice in engineering companies and federal agencies. These methods are abandoned in favour of simple empirical methods, mainly due to their extended demand in input data. The reliability of the results obtained by these simple methods is, however, often unsatisfactory and there is a strong demand for better performing practice-oriented methods. This work intends to add a step towards the solution of this problem. The overall goal consists in developing / adapting a methodology for the estimation of overland flow suited to the Swiss alpine and prealpine context. The methodology should further exclusively rely on widely available geographic information. It was decided in this respect to work with the geographic database GEOSTAT. GEOSTAT is operated by the Swiss Federal Statistical Office (Bundesamt für Statistik) and covers the whole country. The highest spatial resolution of the maps is 1 hectare. For overland flow estimation the Curve Number method is used. The Curve Number method was developed by the US Natural Resources Conservation Service (formerly known as Soil Conservation Service) and allows for the estimation of direct runoff for a given rainfall input based only on qualitative information on soil type and land use. The goal then consists in adapting the Curve Number method in order to allow for a correct reproduction of overland flow for all types of hydrological different reacting areas which can be differentiated based on information from the GEOSTAT maps. This goal is addressed by reproducing hydrographs of annual flood events observed in four alpine and prealpine catchments by means of a spatially distributed Rainfall-Runoff model. The spatial resolution of the model corresponds to the spatial resolution of the information from the GEOSTAT maps (1ha). For each single grid cell, the overland flow is further calculated separately by the Curve Number equation. The simulations then allow for indications on the performance and the parameterisation of the Curve Number equation for single hydrological different reacting areas. These indications are however exposed to larger uncertainties. The simulated hydrograph is influenced by different hydrological areas which react similarly and different modules of the model (e.g. subsurface flow module). Therefore the single hydrological different reacting areas cannot be analysed independently. In order to reduce the uncertainty, the performance of the Curve Number equation was also analysed based on data from a number of small scale 6800$5< rainfall experiments. These experiments are of interest in the context of the present study, as the runoff processes and the soil characteristics at the plot sites were observed and measured in detail. Consequently, the performance of the Curve Number equation can be investigated separately for single soil types. The results from the analysis at this plot scale show that the equation mainly performs well for rainfall plots in which Hortonian runoff processes dominate and which are located on Cambisols and Gleysols. The equation shows a lower performance for plots on soils such as Rendzina, Ranker and Podsol. In terms of reparameterisation, only preliminary indications are gained from the analysis mainly due to the limited number of experiments. For a number of plot sites the runoff behaviour is furthermore dominated by local structures such as macropores. At basin scale the Rainfall-Runoff model in general and the Curve Number equation in particular show a satisfactory performance in reproducing the flood runoff observed in the prealpine catchments. In the alpine catchment considered a reproduction of the flood events was however not successful due to the large uncertainty of the spatial rainfall distribution and the influence of snow on the flood events. The simulations generally result in rough indications on the performance and the parameterisation of the Curve Number equation for a number of hydrological different reacting areas. For storm events characterised by rainfall peaks interrupted by periods of rainfall of low intensity, modification of the Curve Number method is further required and proposed. A comparison of the resulting parameterisation with event characteristics then shows that the relation of parameterisation and antecedent wetness conditions presented in the original Curve Number method are not valid in the alpine and prealpine context. The found Curve Number values further decline with increasing storm rainfall volume. The indications on the parameterisation of the Curve Number equation are however exposed to larger uncertainties. The findings from the analysis at plot scale are partially helpful in reducing these uncertainties. This study represents a first important step towards the modification of the Curve Number method to a method for the estimation of the overland flow in the Swiss alpine and prealpine environment. Further steps mainly consist in extending the analysis / adaptation of the Curve Number method to a large number of additional mesoscale catchments of the Swiss Plateau and the Prealps. When selecting the catchments, similar hydrological different reacting areas should be present in several catchments. This permits a more precise idea on the performance of the Curve Number equation for single hydrological different reacting areas. The simulations of flood events from additional catchments may also serve to refine the proposed modification of the Curve Number equation and gain insight into how the parameterisation relates with the antecedent wetness conditions.

This paper summarizes the major findings on variations of the individual components of the water balance in the head watershed of the Linth River, situated in the high alpine region of Glarus, north-eastern Switzerland, where direct measurements have been available since the beginning of this century. In the current discussion of possible climate changes, the natural variability of the water balance components may help to put today's research results in perspective. In a first step, the individual components of basin precipitation, snow accumulation, glacier mass balance and discharge are assessed. Special emphasis is placed on a comprehensive interpretation of the results. In a second step, a conceptual precipitation-runoff model running on a daily time step is applied. The required data input are daily values of precipitation and air temperature as measured at standard meteorological stations, which facilitates the transfer of this modelling approach to other mountainous regions where data are scarce. The results obtained using the various methods compare favourably with one another. It can be concluded that the high alpine Linth-Limmern region is well suited for the assessment of the water balance components thanks to the long-term series of available data. It is recommended to continue the various measurement programmes being conducted by federal agencies, hydro-power companies and private individuals, and to engage in further detailed investigations concerning runoff processes and pathways.

Hydrological Sciences Bulletin, 1981

This paper describes and compares various approaches to modelling a high mountain basin with dominant snow yields. Three different conceptual models were thus applied to the same test basin over the same test periods, and with identical calibration: (a) one specially developed for the given basin, using a refined description of the physiographic features and including a snowmelt routine based on energy budgets at 12 h intervals; (b) a general purpose hydrological model (HSP model), partially standardized and applied to the basin considered following the users' manual; (c) an intermediate model, much like the HSP model except for the snowmelt routine. Conclusions have been drawn about the structure of models such as the usefulness of introducing some routines far more sophisticated than those of the average model, but mostly about estimations of missing input data required by the model. Some variables such as thermometric gradients or spatial distribution of precipitation are much more crucial than the possible choices between different approaches for modelling évapotranspiration and even snowmelt.

2017

MULTI-VARIABLE CALIBRATION OF A CONCEPTUAL RAINFALLRUNOFF MODEL IN THE SELECTED ALPINE CATCHMENT PATRIK SLEZIAK, JÁN SZOLGAY, KAMILA HLAVČOVÁ, JURAJ PARAJKA Department of Land and Water Resources Management, Faculty of Civil Engineering, Slovak University of Technology, Radlinského 11, 810 05, Bratislava, Slovakia; Institute of Hydraulic Engineering and Water Resource Management, Vienna University of Technology, Karlsplatz 13/222, Vienna, Austria

Forest

An application of the Precipitation Runoff Modelling System (PRMS) based on the concept of Hydrological Response Units (HRUs) is presented for hydrological modelling of an alpine catchment. This is the Aare River catchment upstream of the Lake Thun, in the Bernese Oberland Region, Switzerland, which is characterised by large glacierised areas. Accounting for these areas required to develop further the original PRMS, which was rarely used in alpine regions. Particular attention was devoted to the analysis of the temporal and spatial distribution of temperature and rainfall within the catchment. The derivation of distributed model's parameters was based on an extensive database of catchment characteristics available for the region, thereby including a 25 m resolution Digital Elevation Model (DEM), and digital maps of geotechnical properties, soil and landuse. The encouraging results in spite of the highly complex catchment morphology underline the importance of the availability of spatially distributed data to be used for HRUs identification and parameterisation. Such availability allowed transferring the parameter set from one subcatchment to another without significant loss of model efficiency. However, as expected, the model was strongly sensitive to the parameters describing the runoff generation processes (retention capacity of the unsaturated storage, snowmelt infiltration capacity) and the routing of water in subsurface and groundwater reservoirs. This is due to the intrinsic variability of these parameters, but may be enhanced by the general lack of specific distributed data that could be used to improve calibration. Accordingly, the study concludes about the evident need for enlarging data availability in relation to subsurface and groundwater processes, or, alternatively, in fostering the development of robust parameter calibration methods, which rely on data that are generally available.

This paper focuses on the comparison of runoff simulations at the plot scale (-30 m2), the hillslope scale (-100 m2) and the small catchment scale (85 000 m2) by means of physically-based and conceptual models using observation data from the above project. To estimate the total lateral runoff from the whole soil profile, a two-dimensional numerical model for saturated and unsaturated matric flow was applied to the plot scale and hillslope domain. A conceptual model based on TOPMODEL was applied to test its ability for process representation from the plot scale to small catchment scale. The simulations were compared to in situ measurements, which were part of an interdisciplinary project between hydrologists and forest ecologists.

Journal of Hydrology, 2003

A temperature index approach including incoming solar radiation was used as a sub-model in the gridded hydrological catchment model WaSiM-ETH to simulate the melt rate of glacierized areas. Melt water and rainfall are transformed into glacier discharge by using linear reservoir approaches. The complex WaSiM model was applied to three Swiss high-alpine river catchments with different portions of glacierized areas to simulate the discharges of the whole catchments. Gridded data sets of elevation, soil type, and land-use were used including meteorological input data from the network of MeteoSwiss. These data were spatially and temporally interpolated and modified according to exposition, slope and topographic shading. Continuous discharge simulations for the catchment areas were performed in a spatial resolution of 100 m and a temporal resolution of 1 h for the period 1981-2000 and compared with hourly discharge observations measured at the catchment outlets. To improve the calculation of glacier runoff, a seasonal varying radiation factor has been implemented in the glacier melt equation. The pronounced diurnal and seasonal fluctuations in discharge, which are typical of partly glacierized catchment areas, were simulated in a good agreement with the observed values.

Karst aquifers are important for freshwater supply, but difficult to manage, due to highly variable water levels and spring discharge rates. Conduits are crucial for groundwater flow in karst aquifers, but their location is often unknown, thus limiting the applicability and validity of numerical models. We have applied a conduit model (SWMM) to simulate highly variable flow in a folded alpine karst aquifer system, where the underground drainage pattern is comparatively well-known from previous tracer studies. The conduit model was coupled with a reservoir model representing recharge, storage and transfer of water in the epikarst and unsaturated zone. The global optimization approach (GA) was applied to achieve an efficient model calibration. It was possible to simultaneously simulate the highly variable discharge characteristics of an estavelle, and overflow spring and a permanent spring draining the conduit system. The model allowed for the collection of spatially differentiated information on recharge, rapid flow and slow flow in four individual sub-catchments. The formation of backwater upgradient from conduit restrictions turned out to be a key process in activating overflow springs. The proposed modeling approach appears to be transferrable to other karst systems with predominant conduit drainage, but requires previous knowledge of the configuration of the conduit system.

2000

CITATIONS 7 READS 48 7 authors, including:

The prediction of peak flow and the simulation of flood hydrographs in a stream or river is a very complex process. An application of a spatially distributed hydrologic model WetSpa working on a daily time scale is presented in this paper. The model combines elevation, soil and landuse data within a GIS framework, and predicts flood hydrograph and the spatial distribution of hydrologic characteristics through a watershed. The hydrological processes considered in the model are precipitation, interception, depression, surface runoff, infiltration, evapotranspiration, percolation, interflow, groundwater flow, and water balance in the root zone and the saturated zone. This version of the WetSpa model uses a modified rational method to calculate runoff and an energy balance approach to estimate the snowmelt runoff based on temperature data. The main focus of the paper is on discussing the simulation of a flood hydrograph as consequence of rainfall and snowmelt. The watershed is represented as a grid cell mesh, and routing of runoff from each cell to the basin outlet is accomplished using the first passage time response function based on the mean and variance of the flow time distribution, which is derived from the advection-dispersion transport equation. The model is applied to the Hornad river basin located in Slovakia. Daily hydrometeorological data from 1993 to 2000, including precipitation, temperature, evaporation, and windspeed, are used as input data to the model. For the simulation of hydrographs at the basin outlet and at the flow monitoring stations inside the catchment, the basin was divided into 223 subcatchments, corresponding to the threshold value of 1000 cells when delineating the stream network based on topographic flow accumulation. The calibration process is mainly performed for the global model parameters, whereas the spatial model parameters are kept as default values. The initial global model parameters are specifically chosen according to the basin characteristics as discussed in the documentation and user manual of the model. Results of the simulations show a good agreement between calculated and measured hydrographs at the outlet of the basin (Nash-Sutcliffe efficiency is equal to 74%). An interesting period, with snow accumulation followed by snowmelt producing a flood is discussed in detail.