Landscape Optimization: Applications of a Spatial Ecosystem Model

Ralf Seppelt, Alexey Voinov

Contents

  1. Overview
  2. Study Area
  3. Results
  4. Technical Documentation
  5. References
  6. Acknowledgements


Go Back

Overview

Spatially explicit ecosystem models allow the calculation of water and matter dynamics in a landscape as functions of spatial localization of habitat structures and matter input. For a mainly agricultural region we studied the nutrient balance as a function of different management schemes. For this purpose we formulated optimization tasks. This required the definition of performance criteria, which compare economic aspects, such like farmer’s income from harvest, with ecologic aspects, such like nutrient loss out of the watershed. The task was to calculate optimum land use maps and fertilizer application maps maximizing the performance criterion. We developed a framework of procedures for numerical optimization in spatially explicit dynamic ecosystem simulation models. The results were tested using Monte Carlos simulation, which based on different stochastic generators for the independent control variables. Gradient free optimization procedures (Genetic Algorithms) were used to verify the simplifying assumptions. Parts of the framework offer tools for optimization with the computation effort independent of the size of the study area. As a result, important areas with high retention capabilities were identified and fertilizer maps were set up depending on soil properties. This shows that optimization methods even in complex simulation models can be a useful tool for a systematic analysis of management strategies of ecosystem use.

This research has been supported by a grant from the U.S. Environmental Protection Agency’s Science to Achieve Results (STAR) program (R827169) and the German Science Foundation (Deutsche Forschungsgemeinschaft, DFG) project SE-796/1-1. The Hunting Creek modeling project was also supported by a grant from the Calvert County Board of Commissioners. 
Go Back

Study Area

For the development of an appropriate methodology for optimization procedures for spatial landscape models we focus on the problem of optimum land use patterns and optimum fertilization in a mainly agricultural region in Southern Maryland, USA. We have focused our studies on the Hunting Creek Watershed which is located entirely within Calvert County in Maryland, USA. The 22,5 km² large study area belongs to the drainage basin of the Patuxent River (2356 km²) which is one of the major tributaries of the Chesapeak Bay. Soil types are well drained, mostly severly eroded soils that have a dominantly sandy clay loam to fine sandy loam subsoil (Soil Survey. Calvert County, MD. USDA, July 1971. 76p.). The annual rainfall varies between 400 and 600mm. Main landuse of the whatershed are forest and agricultural habitats. Rapid population growth, development and change in land use and land cover have become obvious features of the landscape. This Figure displays the location of the study areas as well as the spatial hirarchical concept of embedding different watersheds.
Go Back
 

Results

The Results of the methodological framework (compare Figure) can be study without any computational effort, for instance within the ArcView front end (see Figure). Different optimized landscape patterns can be studied. For the Hunting Creek watershed some precalculated solutions can be studied within the web front end.
Go Back
 

Technical Documentation

The developed methodological framework (compare Figure) makes use of the Spatial Modelling Environmenta SME. The Technical Documentation gives an overview of the steps to be undertaken to use a spatial explicit model within the landscape optimization framework.

Go Back
 

References

Go Back
 

Acknowledgements

We are grateful to Dave Brownlee from Calvert County Planning and Zoning for much needed advice and help with some data. We thank Thomas Maxwell for several upgrades in SME that were required for linkages to the optimization tools. Additional thanks are due to Dagmar Söndgerath for her valuable assistance concerning the statistical analysis. Results in this contribution were obtained using the GAlib genetic algorithm package, written be Matthew Wall at the Massachusetts Institute of Technolgy.

Go Back