Diploma Thesis
Application of remote sensing methods for micrometeorological site evaluation
Lucia Seebach (Reithmaier) (10/2002-06/2003)
Support: Thomas Foken
Die Arbeit wurde teilweiseam Max-Planck-Institut für Biogeochemie in Jena angefertigt, Betreuerin: Prof. Dr. N. Buchmann
Several long-term flux measurement sites are located in complex landscapes characterised by a broad variety of land use classes. The complexity of this terrain disagrees with the fundamental assumptions of eddy-covariance measurements such as horizontal homogeneity, steady state and advection free conditions. A procedure by which complex flux measurement sites could be evaluated was developed in order to assess the influence of the spatial heterogeneity on the quality of the fluxes measured on site (Göckede et al., 2003). This procedure combines existing quality assessment tools for flux measurements with footprint modelling. The evaluation method requires micrometeorological flux data as well as further input data defining the structure of the surrounding terrain: a land use map and a roughness length (z0) matrix of the region of interest. In evaluations to date, topographical maps were used as a basis for the required maps. This method was time-consuming and resulted in matrices of low resolution of up to 150 meters. The aim of the study was to test the enhancement of the evaluation method with satellite remote sensing methods, which were applied to produce land use maps of the same area. For this purpose, the optimal remote sensing method had to be determined. Landsat ETM+ and IKONOS satellite data were used to perform a supervised land use classification. Additionally, the utility of a pre-processing step, the correction for atmospheric effects, was investigated. Thereafter, the land use maps derived from remote sensing methods were applied to the evaluation method. Particular interest was placed on the influence of higher spatial resolutions of the land use maps on the evaluation output. Furthermore, roughness length values taken from existing z0 classifications were assigned to the relevant land use classes. Here, the optimal roughness length classification for footprint modelling results was investigated. The most suitable remote sensing application for the micrometeorological site evaluation method developed by Göckede et al. (2003) proved to be a multi-temporal supervised land use classification using Landsat ETM+ data. This standard approach delivered sufficient accuracy for the present study. The classification using IKONOS data did not deliver a satisfying result, because it was not applicable to the image processing software used, ENVI. The use of atmospheric correction as a pre-processing step was found to be of no benefit in this study. The approach using land use maps derived from remote sensing data for the site evaluation method did not deliver significantly different evaluation results from those produced by the existing approach using topographic maps. The application of higher spatial resolution maps, with more detailed land cover information, did not provide a different output. Due to the shortcomings of the footprint model integrated in the evaluation method, the optimal z0 classification could not be found. Generally, the flux measurement site used in this study (Hainich, Germany) obtained high quality results regarding the site evaluation. Similar results from the existing evaluation based on coarse topographic maps as well as from the updated evaluations based on remote sensing methods, prove that the Hainich site fulfils the assumption of homogeneity. Further investigations within a more heterogeneous terrain are required in order to test the influence of remote sensing data.