within the project
Geographic survey of nitrogen status and the risk of nitrate leaching of forests in Bavaria, Germany.
Show forest map
Description
The goal of this project is to provide an overview of nitrogen status in the forests of Bavaria. In undisturbed ecosystems, nitrogen availability is a natural limiting factor to forest growth. Due to fossil fuel combustion and intensification in agriculture, however, nitrogen has become a major environmental pollutant. Thus, nitrogen saturation of forest ecosystems and subsequent leaching of nitrate to groundwater is a problem of increasing concern in Central Europe. Nitrogen status of forest ecosystems depends upon a set of variables, such as soil conditions, nitrogen deposition and state and type of forest cover.
Within this ongoing project, a statistical model will be developed for the different spatial scales to estimate nitrogen status with the help of soil data, deposition data and forest type and quality information.
In order to get an overview of nitrogen status, we will be working at three
spatial scales: provincial (scale 1:50000), regional (1:200000) and local
(1:10000). Currently, there is no up-to-date map of forest cover type available in Bavaria. While
the CORINE land-cover map can be used at provincial scale and forest management maps at local scale, more detailed information is needed at regional scale.
This sub-project therefore aims at the production of a medium-resolution forest map for the Munich region using Landsat-TM 7 images.
This map differentiates coniferous, mixed and deciduous forests as well as two age classes.
Methods
We acquired two Landsat scenes from the same overpass on 13th of May, 2001. The images were mosaiced together
and corrected for athmospheric influences using dark point subtraction. From the image mosaic, a subset of the study was extracted and georeferenced using a set of 30 Ground control points from topographic maps.
A forest mask at the scale 1:50 000 was used to mask out non-forest areas.
Several image processing techniques were employed to achieve optimal results:
Results were visually interpreted with respect to the desired classification output (Forest type and age class) in two well known test area, the "Ebersberger Forst" and the "Höhenkirchener Forst" near Munich and compared to digitized forest management maps at a scale of 1:5000 and ground experience.
Artificial Bands were then selected for an unsupervised classification with the
ISODATA method.
With this provisional classification stored in Geographic Information System (GIS) on a portable computer, we performed further ground checks and extracted training areas for a supervised classification.
The supervised classification was then performed with maximum likelyhood classification.
Results
A detailed, up-to-date
map at regional scale (1:200 000) of forest type and age group was produced
for the larger area of Munich. Although the desired accuracy was not achieved for all classes, the quality of the result map at scale 1: 200 000 is sufficient as an input in the nitrogen leaching model.
Landsat data
LANDSAT ETM+ Scene
193/26 13 May, 2001
193/27 13 May, 2001