Sie sind auf Seite 1von 10

C. A. Brebbia (ed.), 2000, Risk Analysis II, WIT press, Southampton, Boston: 295-304.

Hydrologic modelling on a catchment scale using GIS and remote sensed land use information
F. De Smedt, L. Yongbo and S. Gebremeskel
Department of Hydrology and Hydraulic Engineering, Free University Brussels, Belgium.

Abstract
A spatially distributed hydrologic model is presented for simulation of runoff in a river basin. The generation of runoff is triggered by the rain intensity and soil moisture status, and is calculated as the net precipitation times a runoff coefficient, which depends upon slope, land use and soil type. The runoff is routed through the basin along flow paths determined by the topography using a diffusive wave transfer model, that enables to calculate response functions between any start and end point, depending upon slope, flow velocity and dissipation characteristics along the flow lines. All these calculations can be performed with standard GIS tools. The model is tested on a catchment in Belgium for which topography and soil data are available in GIS form, while the land-use and soil cover is obtained from remote sensed images. The resulting calculated discharges compare favourably with measurements, without any need to model optimisation. Next a 100 year series of hourly precipitation data is processed with the model and the resulting hydrograph is analysed statistically to determine the characteristics of extreme floods.

Introduction

The problem of estimating runoff from a storm event is one of the key points in hydrologic modelling. Classical techniques as the rational method or the Soil Conservation Service curve number approach are still widely used in practice. More accurate is to estimate runoff on the basis of the soil moisture conditions

296

Risk Analyses II

and geophysical characteristics, as for instance presented by Willmott et al. [1] and Dickinson et al. [2]. Due to the complexity of the hydrological processes and the basin characteristics, physically based distributed models using GIS techniques are becoming popular. For instance, Naden [3] and Troch et al. [4] present spatially distributed hydrologic rainfall-runoff models including hillslope overland flow and channel flow routing. Muzik [5] presents a method for surface runoff routing by using a GIS-based distributed unit hydrograph. Schumann et al [6] apply a two-dimensional instantaneous unit hydrograph within a GIS framework. Maidment et al. [7] present an elaborate grid-based model, in which the spatially distributed unit hydrograph is derived from a time-invariant flow velocity field. Olivera et al. [8] propose a method for routing spatially distributed excess precipitation over a watershed using response functions derived from a digital terrain model. Some of these physically based distributed models have obtained a world wide recognition, as for instance Topmodel [9,10] and SHE [11,12]. In this paper we present a physically based distributed hydrological model that uses detailed basin characteristics to predict hydrological processes. We will focus especially on the simulation of runoff. The model is validated for a small watershed in Belgium by comparing calculated and observed hourly discharges for a 6 months period. Next, the utility of the model is demonstrated by forecasting peak discharges resulting from an observed 100 year precipitation series.

Theory

The present model is an extension of WetSpa developed by Wang et al. [13], in which the hydrological processes are simulated in a grid based schematisation of a river basin. The processes considered in the model are precipitation, interception, evaporation, surface runoff, infiltration, evapotranspiration, soil moisture storage, interflow, percolation, groundwater storage and discharge. The main outputs of the model are river flow hydrographs and spatially distributed hydrological characteristics as soil moisture, infiltration rates, groundwater recharge, surface water retention or runoff, etc. The model is especially useful to simulate the effects of topography, soil type, and land-use or soil cover on the hydrologic behaviour of a river basin. In this study we will only focus on the runoff and flood routing components of the model. The calculation of the runoff is rather classical:
V = CP ( s ) ,

(1)

where V is the amount of surface runoff [L], P the net precipitation [L] (rainfall minus interception), the soil moisture content, s the saturated soil moisture content, and C the runoff coefficient, which is assumed to depend upon slope, soil type and soil cover. Runoff coefficients were collected from the literature (Kirkby [14], Chow et al. [15], Browne [16], Mallants & Feyen [17], and Pilgrim

297

Risk Analyses II

& Cordery [18]) and a table was generated, linking values of the runoff coefficient to slope, soil type and land-use classes. Next, the generated runoff needs to be routed through the basin such that river discharge can be computed. In this study, the diffusive wave approximation is used to simulate both overland flow and channel flow:
Q 2Q Q , = D 2 c t x x

(2)

where Q is the discharge [L/T], D the wave damping coefficient [L2/T], c the wave celerity [L/T], x the distance along the flow path [L] and t the time [T]. The wave transport parameters c and D depend upon the flow velocity, flow depth and terrain characteristics. In case the flow velocity, v [L/T], is determined by the Manning equation

v=

1 23 R i, n

(3)

where R is the hydraulic radius [L], n the roughness coefficient [L-1/3T], and i the slope [-], c and D are given as 3 c= v, (4) 2 and vR . (5) D= 2i If it is assumed that the hydraulic radius is a static terrain characteristic that does not change during a flood event, it follows from eqns (3) and (4) that c and D only depend upon position. In such case, we propose an approximate solution of eqn (2) in the form of a response function, relating the discharge at the end of a flow path to the available runoff at the start of the flow path:
Q =V ( 1 + t0 t ) ( t t0 )2 exp , 2 2 2 2

(6)

where t0 is the average flow time [T] and the deviation of the flow time [T], i.e.

t0 =
and

dx , c

(7)

2D 2D dt 3 dx . c2 c

(8)

298

Risk Analyses II

Hence, the flow routing consists of tracking the runoff along its topographic determined flow path, such that a response function is obtained for every grid cell to the outlet of the catchment, or any other downstream convergence point This routing response serves as an instantaneous unit hydrograph and the total discharge is obtained by convolution of the flow response from all grid cells. Very convenient is the fact that the response functions can be obtained using standard GIS techniques. First maps are produced of c and D with eqns (4) and (5). Next, the contributing area is determined from topographic data for a particular downstream convergence point, and for each contributing grid cell the value of t0 and are calculated by integration along each flow path, using eqns (7) and (8). Finally, the total river discharge at the downstream convergence point is obtained by superimposing all contributions from every grid cell.

Application

The model was tested on the 67 km2 Barebeek catchment, situated north of Brussels, Belgium. The study area was divided into grid cells of 50m by 50m. The topography was digitised from 1/10.000 maps and the soil types were obtained from the physical system map of Flanders (Vlaamse Landmaatschappij). Figure 1 presents the topography of the area, with the basin boundary, main river courses, and measuring stations. The soil cover was obtained from the digital land use map of Flanders, which is based on remote sensed data of 1995 (Ondersteunend Centrum GIS-Vlaanderen). About thirty land use classes are shown on this map. But because not all of these are relevant from hydrological point of view, or the knowledge is lacking about their hydrologic behaviour, the map was resampled to 5 hydrologic land use classes, i.e.: urban areas, agricultural crops, pasture, forest, and open water. Figure 2 shows the resulting land use map. One can notice a mosaic of land uses. Forrest (16.8%) is predominant in the river valleys, while the higher terrain consists of agricultural areas, with pasture (24.7%) or crops (36.9%), strongly intermixed with urban areas (16.2%), as villages, roads and Brussels airport in the south. Figure 3 shows the runoff coefficients that results from the different slope, soil type and land use class combinations. The influence of the urban areas is self-evident. Notice, that due to the grid size, urban areas were assumed to be impervious for 30% and behaving as pasture for the remaining percentage. This results in runoff coefficients around 40% in the urban areas, while other areas have much smaller values, up to 5% for forests in the valleys with practically zero slopes. Next the overland flow velocity was calculated with eqn (3). Also, here the urban areas have a marked influence, due to the artificial drainage facilities as sewer systems, that results in flow velocities of order 1 m/s or more. On the other hand, in the valleys the flow velocity is very small, due to the high resistance of the soil cover, being mostly forest, and the very faint slopes. With this information the response functions were calculated using the diffusive wave approximation. Figure 4 shows the resulting average flow time from each grid cell to the outlet of the basin. Again, marked differences result from the land uses; especially urbanisation has a large impact.

299

Risk Analyses II

Figure 1: Topography of the study area.

Figure 2: Land use map.

300

Risk Analyses II

Figure 3: Distribution of the runoff coefficients.

Figure 4: Average flow time of surface runoff.

301

Risk Analyses II

The model was verified and the different components of the hydrological cycle were calculated in each cell on a hourly basis for a period of about 6 months. A comparison of calculated and observed discharges in the most downstream measuring station is presented in Figure 5. Results for the other measuring stations are similar. One can notice a reasonable agreement between the model results and the observed data. Peaks in the hydrograph are rather well simulated, as well for size as for time of occurrence. Because the model is clearly overparameterised, we do not intend to optimise its performance by calibration, also in view of the fact that hydrological observations are never completely free from errors. Moreover, due to the natural variability of hydrologic processes and the complexity of basin characteristics, it is evident that mathematical models will never reach perfection. Hence, we accept the results as they are.
6 5 4 3 0 3

observed model

6 9 12 15 18

Q (m /s)

2 1 0
1/9/98 16/9/98 1/10/98 14/1/99 16/10/98 31/10/98 15/11/98 30/11/98 15/12/98 30/12/98 29/1/99

Figure 5: Observed and simulated discharges.

To demonstrate the usefulness and performance of the model, a historical 100 year series of hourly precipitation data was processed and the resulting hydrograph for the total catchment was analysed statistically to determine the characteristics of peak discharges. The largest flood is given in figure 6. It shows a typical pattern that is present in all of the precipitation events that lead to flood discharges. In all cases the actual storm is preceded by another storm, which generally does not lead to flooding, but which suffices to saturate the soils to such an extend that the second storm as a maximal impact leading to large runoff volumes and flooding in the river system. Figure 7 shows the resulting peak discharges of the fifty highest floods versus return period. Such information will

P (mm/h)

302

Risk Analyses II

enable to determine the influence of changes in land use or soil cover on the hydrological behaviour of the river basin.
14 12 10
0 5 10 15 20 25 30 35
2/09/96 4/09/96 6/09/96 23/08/96 25/08/96 27/08/96 29/08/96 31/08/96 8/09/96

Q (m/s)

8 6 4 2 0

Figure 6: Simulated peak flood in the 100 years series.

12 10

Q (m/s)

8 6 4 2 0 1 10 100

Return period (year)


Figure 7: Simulated peak discharge versus return period.

P (mm/h)

303

Risk Analyses II

Conclusions

A physically based distributed hydrological model was presented for simulating the hydrologic behaviour and especially runoff in a river basin. The generation of runoff depends upon rain intensity and soil moisture status and is calculated as the net precipitation times a runoff coefficient, which depends upon slope, land use and soil type. The runoff is subsequently routed through the basin along flow paths determined by the topography using a diffusive wave transfer model, that leads to response functions between any start and end point, depending upon slope, flow velocity and dissipation characteristics along the flow lines. The model uses detailed basin characteristics and calculations are for the most part performed by standard GIS tools, such that the model is especially useful to analyse the effects of topography, soil type, and land-use or soil cover on the hydrologic behaviour of a river basin. The model was validated on a small watershed in Belgium for which topography and soil data are available in GIS form, while the land-use and soil cover was obtained from remote sensed images. The resulting calculated hydrograph compared favourably with measurements, without any need to model optimisation. The usefulness and utility of the model was subsequently demonstrated by forecasting peak discharges resulting from an observed 100 year precipitation series. The resulting discharges were analysed statistically to determine the characteristics of extreme flood events. We believe that the model has great potentiality to determine the influence of changes in land use or soil cover on the hydrological behaviour of the river basin.

References
[1] Willmott, C.J., Rowe, C.M., & Mintz, Y. Climatology of the terrestrial seasonal water cycle, J. Climatology., 5, pp. 589-606, 1985. [2] Dickinson, R.E., Henderson-Sellers, A., & Kennedy, P.J. BiosphereAtmosphere Transfer Scheme (BATS) Version 1e as coupled to the NCAR community climate model, NCAR Technical Note, National Center for Atmospheric Research, Boulder, Colorado, pp. 40-41, 1993. [3] Naden, P.S. Spatial variability in flood estimation for large catchment: The exploitation of channel network structure. J. Hydrol. Sci. (37), pp. 53-71, 1992. [4] Troch, P.A., Smith, J.A., Wood, E.F., & de Troch, F.P. Hydrologic controls of large floods in a small basin, J. Hydrology (156), pp. 285-309, 1994. [5] Muzik, A. GIS-derived distributed unit hydrograph. HydroGIS 96, Vienna, ed. K. Kovar & H.P. Nachtnebel, IAHS Publ. No. 235, pp. 453-460,1996. [6] Schumann, A.H., & Funke, R. GIS-based components for rainfall runoff models. HydroGIS96, Vienna, ed. K. Kovar & H.P. Nachtnebel, IAHS Publ. No. 235, pp. 477-484,1996

304

Risk Analyses II

[7] Maidment, D. R., Olivera, J.F., Calver, A., Eatherall, A. & Fraczek, W. A unit hydrograph derived from a spatially distributed velocity field. Hydrologic Processes, 10, pp. 831-844, 1996. [8] Olivera, F., & Maidment D.R. Geographic information systems (GIS)-based spatially distributed model for runoff routing. Water Res. Research., 35(4), pp. 1155-1164, 1999. [9] Beven, K.J., &Kirkby, M.J. A physically-based variable contributing area model of basin hydrology. Hydrol. Sci. Bull., 24(1), pp. 43-69, 1979. [10] Beven, K.J. Spatially distributed modeling: Conceptual approach to runoff prediction. Recent Advances in the Modeling of Hydrologic Systems, ed. D.S. Bowles and P.E. OConnell, Kluwer Academic, pp. 373-387, 1991. [11] Abbott, M.B., Bathurst, J.C., Cunge, P.E., OConnell, P.E., & Rasmussen, J. An introduction to the European Hydrological System Systme Hydrologique Europen, SHE, 2. Structure of a Physically-based, Distributed Modelling System. J. Hydrology, 87, pp. 61-77, 1986. [12] Refsgaard, J.C. & Storm, B. MIKE SHE. Computer Models of Watershed Hydrology, ed. V.P. Singh, Water Resources Publications, Colorado, USA, pp. 809-846, 1995. [13] Wang Z., Batelaan, O. & De Smedt, F. A distributed model for Water and Energy Transfer between Soil, Plants and Atmosphere (WetSpa). Phys. Chem. Earth, 21(3), pp. 189-193, 1996. [14] Kirkby, M.J. (ed.), Hill Slope Hydrology, John Wiley & Sons, Ltd., pp. 235, 1978. [15] Chow, V.T, D.R. Maidment and L.W. Mays, Applied Hydrology, McGrawHill Inc., New York, pp. 498, 1988. [16] Browne. F.X. Stormwater management, Standard Handbook of Environmental Engineering, edited by R.A. Corbitt, McGraw-Hill, New York, pp. 7.1-7.135, 1990. [17] Mallants D. and Feyen J., Kwantitatieve en kwalitatieve aspecten van oppervlakte en grondwaterstroming (in Dutch), Volume 2., KUL, pp96 , 1990 [18] Pilgrim, D.H., Travel times and non-linearity of flood runoff from tracer measurements on a small watershed, Water Resour. Res., 12, 487-496, 1976. [19] Miller, W.A. & Cunge, J.A. Simplified equations of unsteady flow. Unsteady Flow in Open Channels, ed. K. Mahmood & V. Yevjevich, Vol. 1, Chap. 5, Water Resour. Publ., Fort Collins, Colorado, 1975. [20] Jansen, P.Ph., van Bendegom, L., van den Berg, J., de Vries, M., & Zanen, A. Principles of River Engineering, Pitman Publ. Ltd.: London, pp. 319, 1979.

Das könnte Ihnen auch gefallen