Winiger Matthias

Zusammenhang zwischen Vegetation und Relief in alpinen Einzugsgebieten des Wallis (Schweiz)

Project Number: Parcs Data Center 28466 / 4D CH-4401
Project Type: Dissertation
Project Duration: 02/01/1998 - 11/30/2001 project completed
Funding Source: other ,
Project Leader: Prof. Matthias Winiger
Climatology & Landscape Ecology
Dept. of Geography
University of Bonn
Meckenheimer Allee 166
DE-53115 Bonn
Phone: +49 228 73 2101
FAX: +49 228 73 7506
e-Mail: winiger(at)

related to this project.
for which the project has a relevance.

Research Areas:

environmental sciences

Distribution of the vegetation, relief, anthropogenic influence

The vegetation distribution of high mountain cultural landscapes is characterised by an extreme spatial heterogeneity concerning natural and anthropogenic site factors. In a complex system of direct and indirect factors influencing vegetation distribution landform is the major natural, but indirect ecological factor: landform highly varies the spatial distribution of precipitation, irradiation, air and soil temperature as well as geomorphological processes (avalanches, rockfall etc.). It thus produces a patchwork-like pattern of diverse vegetation habitats. Besides these natural factors the human dimension in the sense of historic and recent land use was and is important for the development of vegetation. Thus present-day vegetation in alpine cultural landscapes has to be regarded largely as quasi-natural replacement vegetation, resulting from the interaction of human impact and natural environmental conditions.
In this study present-day vegetation types were therefore defined as the „status quo“ in the sense of „habitat-units“ on the basis of plant sociological vegetation alliances after DELARZE et al. (1999). As spatial information on climatic and other environmental conditions are commonly lacking in mountain areas, the main aim was to analyse to what extent vegetation alliances were dependent on landform i.e. on the spatial distribution of landform parameters. Additionally it had to be investigated to what extent vegetation habitats could be indicated by the exclusive use of landform parameters represented by digital elevation models.
For this purpose field campaigns were conducted in a 40km by 20km transect from the Lötschental to the Turtmanntal in the Swiss canton Wallis. Integrating the data, supported by semi-automatic satellite- and aerial photograph-based classification procedures as well as GIS-methodology, resulted in the production of 2 vegetation maps of different resolution (25m, 5m). The potential of the different spatial resolutions of aerial and satellite based remote sensing data was analysed for differentiating geobotanically defined vegetation types: while satellite based classifications could only distinguish between vegetation alliances with habitat areas large enough and were mainly capable of defining (sub-)formations, it was possible to extract 52 vegetation classes at vegetation alliance level using CIR-aerial photographs.
Additionally, a hemerobiotic map was derived, showing the degree of anthropogenic influence on the vegetation.
Based on the high (spatial and thematic) resolution vegetation map the relation between vegetation and landform was analysed, after a set of appropriate landform parameters had been derived. Furthermore the landscape represented by the digital elevation model was decomposed into area-units of homogeneous geomorphometric characteristics. These area-units in the sense of „geotopes“ were created by applying different landform classification schemes, in order to analyse the spatial concurrence of these “geomorphometric topes” with vegetation alliances.
Using qualitative-descriptive distribution measures as well as analytic-statistical methods like contingency tables, principal components analysis and a students t-test, it was possible to prove significant correlations between certain vegetation alliances and landform parameters. In contrast to this the landform classification schemes were hardly suitable for the indication of habitats.
In a last step the spatial distributions of different vegetation alliances were simulated using 2 different model approaches (classification and regression trees CART, parallel-epiped box-classificators PPD) and relevant landform parameters. Besides a better performance of the PPD as compared to the CART model it appeared that the spatial distribution of large, zonally arranged and intensively cultivated vegetation classes could be simulated with high accuracy (> 95%) by the exclusive use of landform parameters. In contrast to that respective accuracies for small, azonal and extensively used classes only reached 50-70%. Furthermore it was possible to reconstruct potential distributions of vegetation alliances from the false positive rate of the simulations, which could be verified by the use of historic sources.
Besides the lack of information on (historic) land use, insufficient parameterisation of geomorphological processes as well as topological relations, the major source of error within the correlation calculations and in the model simulations mainly of smaller vegetation classes was the poor spatial resolution of the digital elevation model as compared to the vegetation map, not being capable of representing microscale landform characteristics.
Especially in the view of a low information density concerning ecological data within high mountain areas, the methodology used here and the models developed aim at improving global vegetation investigations. This research is closely related to the analysis, the monitoring and the protection of biodiversity. Generally, with regard to a potential climate and therefore habitat change, the present study aims at contributing both thematically and methodologically to the research fields of vegetation and high mountain ecology.

Leading questions:
  • To what extent can the integration of multi-temporal and multi-scale remote sensing data and terrain elevations meet the needs of a vegetation mapping based on geobotanic and phytosociology in high mountains? What are the spatial and thematic boundaries of differentiability?

  • Can correlations between vegetation and relief a) be identified using descriptive statistical mass and b) analyzed qualitatively. If so, what methods are suitable?

  • Are vegetation associations tied to specific formal elements or relief units, ie can a relief classification for the vegetation habitat structure in the high mountains be used wisely? Which approach gives the best results?

  • To what extent can relief parameters alone parameterize the natural site factors and anthropogenic modifications of individual vegetation associations and index their habitats in the modeling? Can the residuals between model and reality for the identification of conflict areas between natural ecological habitats and human influence be used to build scenarios of landscape development trends?

  • What is the impact of the spatial resolution and related scale of the data sets on the analysis of the relationships between habitat and relief structure in the high mountains?

    Hörsch, Bianca. 2001. Zusammenhang zwischen Vegetation und Relief in alpinen Einzugsgebieten des Wallis (Schweiz). Ein multiskaliger GIS- und Fernerkundungsansatz. Dissertation, Universität Bonn.
    pdf Dissertation

    PDF Dissertation

    Last update: 4/5/22
    Source of data: ProClim- Research InfoSystem (1993-2024)
    Update the data of project: CH-4401

    Go Back