Relating Alpine Treeline Position to Mountain Topography

Laboratory of Geo-Information Science and Remote Sensing
  Education
  Research
  Publications
  Models
  News & Calendar
  News
  Calendar
  Archive
  News
  Calendar
  2011
  2010
  2009
  2008
  2007
  2006
  2005
  2012
  Staff
  Equipment
  Contact details
  Workshops

17 May 2006 14:30 - 17 May 2006 15:00
Unit: Laboratory of Geo-Information Science and Remote Sensing
Location: GAIA 1
Organisation: Wageningen University

 By Johan Ruijten  

Abstract:     
The alpine treeline is the altitudinal transition zone from montane forest to low alpine vegetation. It is an intensively studied landscape boundary but there still remains much uncertainty about the (interacting) factors influencing tree growth at high altitude. Most previous research has focused on mid- and high-latitude mountain areas. Data from tropical regions are generally lacking.
The aims of this study were to provide insight in the relationship between forest distribution at alpine treelines and mountain topography; to use this description to obtain a better insight in the processes and conditions influencing forest distribution alpine treelines; and the design of a consistent method which could be used to compare forest distribution at alpine treelines between different areas.
A Landsat ETM+ image and an SRTM DEM of Sangay National Park, Ecuador, were used to derive forest distribution at treeline and several topographical and environmental indices. These variables were related to forest distribution by means of a logistic regression model. Several models were made in order to investigate the role of several variables showing multicollinearity, but the model used for prediction included all variables. The model was cross-validated in the training area, and applied in a different study area nearby.
Forest distribution was mainly explained by altitude, wetness index (CTI), eastness and erosion index (STCI). Predictive accuracy of the model ranged from 74.2 % to 84.0 % in the test and the training area. The test area has more human impact, which probably explains the over-estimation of forest there.
The logistic regression approach is suitable for discriminating the relative importance of the variables, but because of dependence between the data points, only the most important variables can be assumed to really affect forest distribution. The ecological meaning of some variables is hard to assess, because they affect several biophysical factors. The method developed during this research allows a quick investigation of factors with potential influence on alpine forest distribution, by using inexpensive and easy to obtain Landsat and DEM imagery. It also allows for a rapid comparison of forest distribution between different areas and the localization of potential disturbances.
Print this activity