Quite often, one wants to generate some data at high resolution (say process some image or images) and then calculate some relevant spatial statistics at some other resolution. For example, you might want to process Landsat TM data at 30m resolution, and might want to aggregate it to a resolution ...

Supervised classification of EO data uses a set of samples from known patterns (usually reflectance spectra) to decide whether a given pattern belongs to one class or another. In landcover applications, one goes to the field, and observes that a given location is indeed class $\omega_c$. These ground observations usually ...

Usually, we remote sensing types ignore country boundaries: they don't really make much sense, as they are not aligned with MODIS pixels ;-) However, I was asked what's the easiest way to produce a global mask of countries, so that all th grid cells (say at 0.5 degree ...

In a previous post, I demonstrated how to stitch and put together a number of MODIS data files. This is useful and interesting, but in the end, we are interested in analysing the data we get out of the satellite. One first way around this might be to extract time ...

MODIS has been around for some 12 years. There are many products that have large time series over the whole globe which one can use to study things. So I might want to produce a a timeseries of the 8-day Gross Primary Productivity (GPP) product over say the UK. Moreover ...