Satellite images or aerial photos are often the most appropriate method of gaining a first general view of large areas. In order to detect different spatial patterns (e.g. vegetation patterns, soil patterns), aerial or satellite images can assist ecological studies. The revealed patterns could provide the basis for the selection of sampling points. Until recently, there has been much debate on how to optimally design sampling programs at large spatial scales. The problem of locating the sampling points in the area of interest, and how many sampling points should be taken, remains largely unresolved. Random sampling is too laborious for comprehensive surveys, whereas traditional subjective sampling, which has been commonly used in e.g. vegetation ecology, violates basic scientific principles for quantitative assessments (e.g. reproducibility, comparability, statistical analysis). One compromise is stratified random sampling. This process of stratification divides the space into subunits (strata) based on factors like precipitation, land management, or habitat constants, e.g. habitat type or elevation. Modern tools for computer-assisted data handling, especially Geographical Information System (GIS) and programs for image processing, have greatly simplified the selection of strata. We present a case study which aims to describe meadow vegetation in a 150 Km2 area of the Prealpine region of Lower Austria based on stratified random sampling to minimize field work, and to maximize the reliability of the result, i.e. the description of the variability, the character of the vegetation in the whole research area based on a vegetation distribution model. Main emphasis was to establish monitoring system for detection of land use change effects in this unique meadow vegetation in the mountain belt. The procedure of stratification was based on satellite data (LandsatTM with a pixel size of 28.5 m × 28.5m) and the use of a digital elevation model (DEM) with a grid size of 250 m × 250 m. The classification of all the input data-sets resulted in a total 20 strata. Each stratum consists of several disconnected sub areas so called sampling regions. Within each Stratum five sampling regions were randomly selected for further analysis. Within these sampling regions, several relevs were subjectively selected resulting in 30 plant communities. These group were found to closely resemble community types revealed from a previous vegetation survey of the meadow-land of that area. The applied satellite images, as well as the coarse grained DEM-data (commercially available data sets), provide sufficient information to delineate sampling regions but not to fix relev locations in the regions. Whether the application of satellite images with a pixel size of less than 10 meters will lead a new approach, which allow a point centered sampling design in contrast to the sampling regions concept, will be shown in the near future. Based on the relevs and the derived plant community types with high degree of representation, a monitoring system which covers the different ecological situations in a particular area, was established.
Weitere Kapitel dieses Buchs durch Wischen aufrufen
- Semi-Objective Sampling Strategies as One Basis for a Vegetation Survey
- Springer Netherlands