Simulation of the impacts of land use/cover and climatic changes on the runoff characteristics at the mesoscale

https://doi.org/10.1016/j.ecolmodel.2006.01.005Get rights and content

Abstract

The analysis of climatic and anthropogenic effects on mesoscale river basins has become one of the main concerns for hydrologists, environmentalists, and planners during the last decade. Many attempts at dealing with this issue have been proposed in the literature. In most studies, however, the main components of the system have been isolated in order to reduce the complexity of the system and its intrinsic uncertainty. In this paper, an attempt to couple two realms of the water system at a mesoscale catchment, namely, the hydrological behaviour of a catchment and the state of the land cover at a given point in time, is presented. Here, instead of using a standard hydrological model, various nonlinear models relating several runoff characteristics with physiographic, land cover, and meteorological factors were linked with a stochastic land use/cover change model. Then, using this integrated model, the magnitude of the effects of the hydrological consequences of land use/cover and climatic changes was assessed in a probabilistic way by a sequential Monte Carlo simulation provided four different scenarios which take into account likely developments of macroclimatic and socioeconomic conditions relevant for a given study area. The proposed methodology was tested in a river basin of approximately 120 km2 located to the south of Stuttgart, Germany.

Introduction

A mesoscale river basin1 is an open, highly complex dynamic system with poorly controlled boundary conditions, whose complexity stems from the fact that it comprises a large number of agents (e.g. people, firms, and government) and a number of tightly coupled subsystems interacting in both space and time at various scales Weidlich and Haag, 1983a, Blöschl and Sivapalan, 1995, Allen and Strathern, 2003, Allen, 2004. These subsystems have very different characteristics but most of them – if not all – exhibit intrinsic nonlinear relationships, delayed responses, feedback loops Wu and Marceau, 2002, Porporato and Ridolfi, 2003 and, in some cases, stochastic behaviours. As a result, the macrostates that emerge from such a system may range from stochastic to chaotic – in the long-term – as was shown by Jayawardena and Lai, 1994, Sivakumar, 2000, and Sivakumar et al. (2002) among others. Moreover, it is recognised that there is a considerable uncertainty in our knowledge of the component processes and the representation of that knowledge when one builds predictive models for such a system Refsgaard, 1997, Beven, 2002, Young and Parkinson, 2002.

Despite the intricacies of this system, remarkable progress in analysing it has been achieved during the past decades. The main approach employed consists of subdividing it into sub-components which are, in turn, modelled as uncoupled subsystems having no endogenous links to any other one. For example, land use models and river basin econometric models were not coupled with hydrological ones [e.g. Berry et al., 1996, Guise and Flinn, 1973, respectively] and, conversely, precipitation-runoff models neither consider models dealing with the land use dynamics nor the economic development of the region [e.g. Leavesley et al., 1983 among others]. The main shortcomings of the lack of integration are: (1) the loss of both the synergy and the feedback effects among subsystems, and (2) the impossibility of estimating the degree to which a variable in a given subsystem may be changed due to the increase or decrease of another one belonging to an exogenous subsystem.

The former is evident when one tries to assess the most likely impacts that land use/cover and climatic changes would induce on the runoff characteristics (e.g. total discharge in winter or frequency of high flow in summer) at the mesoscale (Samaniego, 2003). In order to deal with this problem properly, a holistic formal system (Casti, 1984)– or model – that mimics a number of links (most of them nonlinear) among the relevant processes (e.g. macroclimatic conditions, soil type distribution, hydrological regime) and the driving forces behind land use/cover change should be formulated. There are several studies reported in the literature dealing with the integration of land use/cover change and hydrological models, for instance those of Bronstert et al. (2002) and Niehoff et al. (2002). In those studies, however, the treatment of the land use/cover change model did not follow a stochastic framework.

This paper presents an attempt to develop a simple but robust integrated model to address this problem, as well as the main assumptions therein, and to propose suggestions for future improvements. The proposed approach is illustrated by a case study carried out in a river basin in Germany.

Section snippets

Method

Building an integrated model for an environmental system requires trade-offs between the level of complexity and the degree of coupling among the various subsystems. The former is determined by the number of state variables and parameters as well as their spatio-temporal scales, whereas the latter is given by the number and complexity of the coupling functions and/or interface variables (Walter and Liossatos, 1979). The model, however, should (1) provide a sufficiently good resemblance of the

The study area

The proposed simulation model was applied in the catchment located upstream of the gauging station Denkendorf-Sgewerk in the river Körsch, Germany (see Fig. 2). Its area is 126.3 km2 and because of its vicinity to Stuttgart it has endured a rapid land use/cover change in the past four decades – mainly from agricultural areas to settlement – reaching a gross density of about 720 in./km2 by the end of 2001 (Samaniego, 2003).

Calibration and validation of the LUCC model

Considering the data availability for the study area, three (Q=3)

Results and discussion

The behaviour of the system, roughly represented by the model M, was assessed by 2500 realisations under each set of development scenario conditions. In all these simulations, 1994 was used as the reference year. The simulation period was set from 1994 to 2025 with yearly time steps. As a summary of these simulations, the average growth rate in percent per decade for each simulated variable as well as the probability that the long-term mean of a given variable will be exceeded were estimated,

Conclusions

Considering the basic objective proposed at the beginning of this paper, it is possible to draw the following conclusions:

  • (1)

    Coupling parsimonious hydrological models of several runoff characteristics at the mesoscale with a simple stochastic land use/cover change model has proved to be feasible and enlightening. The approach followed here not only allows one to relate a number of variables otherwise exogenous to the system but also to produce results that highlight the uncertainty of each

Acknowledgement

The authors appreciated the helpful comments from two anonymous reviewers.

References (65)

  • P.M. Allen et al.

    Evolution, emergence, and learning in complex systems

    Emergence

    (2003)
  • W.L. Baker

    A review of models of landscape change

    Landscape Ecol.

    (1989)
  • S. Bankes

    Exploratory modeling for policy analysis

    Oper. Res.

    (1993)
  • O. Barreteau et al.

    Shadoc: a multi-agent model to tackle viability of irrigated systems

    Ann. Oper. Res.

    (2000)
  • M.W. Berry et al.

    Lucas: a system for modeling land-use change

    IEEE Comput. Sci. Eng.

    (1996)
  • K. Beven

    How far can we go in distributed hydrological modelling?

    Hydrol. Earth Syst. Sci.

    (2001)
  • Beven, K.J., 2002. Unceratinty and the detection of structural change in models of environmental systems. In: Beck,...
  • G. Blöschl

    Scaling in hydrology

    Hydrol. Process.

    (2001)
  • G. Blöschl et al.

    Scale issues in hydrological modelling: A review

    Hydrol. Process.

    (1995)
  • G.J. Boer et al.

    A transient climate change simulation with greenhouse gas and aerosol forcing: experimental design and comparison with the instrumental record for the 20th century

    Clim. Dyn.

    (2000)
  • Bárdossy, A., Duckstein, L., 1995. Fuzzy rule based modeling with applications to geophysical, biological and...
  • A. Bronstert et al.

    Effects of climate and land-use change on storm runoff generation: present knowledge and modelling capabilities

    Hydrol. Process.

    (2002)
  • D.G. Brown et al.

    Stochastic simulation of land-cover change using geostatistics and generalized additive models

    Photogrammetric Eng. Remote Sensing

    (2002)
  • G. Cada et al.

    Cumulative impacts of hydropower development: Reaching a watershed in impact assessment

    Environ. Professional

    (1990)
  • Cain, J., 2001. Guidelines for Using Bayesian Networks to Support the Planning and Management of Development Programmes...
  • J. Casti

    On the theory of models and modelling natural phenomenon

  • J. Dooge

    Hydrology in perspective

    Hydrol. Sci. J.

    (1988)
  • S. Geman et al.

    Stochastic relaxation, Gibbs distributions, and the Bayesian restoration of images

    IEEE Trans. Pattern Anal. Mach. Intell.

    (1984)
  • J.M. Gregory et al.

    Summer drought in northern midlatitudes in a time-dependent CO2 climate experiment

    J. Clim.

    (1997)
  • J.W.B. Guise et al.

    Allocation and pricing of water resources

  • J.M. Hammersley et al.

    Monte Carlo Methods

    (1964)
  • A.L. Hammond

    Which World? Scenarios for the 21st Century

    (1996)
  • Cited by (0)

    View full text