This study is focused on the use of Moderate Resolution Imaging Spectroradiometer (MODIS) reflectance data along
with high and very high resolution satellite images. The use of these data requires special attention and the elaboration of
novel preprocessing methods, which is presented through two case studies. In the first one, the aim is to assess subpixel-scale
changes associated to agricultural practice in an agricultural landscape. This requires the integration of temporal
information from MODIS daily time series and spatial information from high resolution satellite images by means of
subpixel unmixing. Our second case study is concentrated on the use of these images in direct radiometric rectification of
high-resolution optical imagery. Our results show that the "standard" preprocessing is not sufficient for carrying out
accurate subpixel analysis in a MODIS time series or for reliable radiometric rectification. The main reasons include
raster-based processing not taking into account the changes in observation dimensions throughout the scene, and pixel
mixing caused by the triangular point spread function (PSF). To resolve this, we propose a novel method based on the
vector data model, in which each MODIS pixel is replaced by a polygon with its real size and orientation. Our results
show this method yields a significant improvement in the radiometric fit of high-resolution and MODIS data.
Remote sensing methods make it possible to analyze and describe landscape changes. However, one can hardly acquire sufficient data for direct long-term analysis. Multiple sensors, geometric distortions, phenological phase differences, atmospheric conditions, different solar angles and many other effects cause inter-scene variability. Furthermore, the temporal distribution of available data sets is often inhomogeneous, which tends to amplify the above-mentioned problems. In our work, we propose a methodology to cope with these difficulties for long-term environmental monitoring and quantitative change detection. A complex approach was chosen with the objective of integrating different methods and disciplines (radiometric and geometric correction, classification, image segmentation and GIS analysis, among others) to extract the maximum of information from the available data. This methodology is presented
and tested on an interesting case study that deals the environmental effects of a barrage system in the northwestern part of Hungary.