Is double the resolution twice as good? [ADAR System 1000 multi-spectral data sets]

With the increased pressure of better management of our dwindling resources, image fusion of data originating from different sensors and/or at different scales offer great potential. The methodology for scaling the individual data sets up or down is of major importance for image fusion of multiple resolution data. In this study the relationship between two multi-spectral data sets acquired at different resolutions (50 cm and 1 m nominal resolution) over the same area using the ADAR System 1000 is examined. Researchers often simulate coarser resolution data by averaging resolution cells over a given distance from a centre point. This is assumed to be an approximation to the actual point spread function of a sensor element, which is dependent on the sensor characteristics. To examine the spatial relationship between the two data sets the point spread function of the sensor is investigated by extracting a number of individual coarse scale elements and an array of fine scale resolution elements of varying size, centred at the position of the coarse scale element. A non-linear regression fitted to these data indicates the relative contribution of the fine scale resolution elements to the coarse scale data value. The results demonstrate that upscaling for this sensor must be based on weighted averaging over a much larger area than the coarse resolution cell indicates.