Geostatistical Downscaling and Simulation of Fine-Resolution NDVI Data from Satellite Images

This paper describes a method to generate higher-resolution information based on geostatistical simulation, using low-resolution information as secondary data. The simulation is carried out as a collocated co-simulation using a sequential Gaussian simulation scheme. Application of the geostatistical scaling laws on the two data sets adds information about their true spatial resolution, which is important for the further combination of the different data. The scaling laws can be used to derive the correlation structures during up- and downscaling for the parameter analysed, in this case the Normalised Difference Vegetation Index – NDVI, derived from remote sensing data. The analysis of scale relations and the true spatial resolution of different data types is often an overlooked subject in petroleum related investigations.