Selon Knut-Frode Dagestad <[email protected]>: > Hi, > > From a time series of colocated images, I construct a VRT with one band > for each time step using gdalbuildvrt: > > > gdalbuildvrt -separate mtsat.vrt mtsat/*_IR*00.tif > > Are there any tools that can be used to calculate some statistics (min, > max, mean, etc) versus time for each pixel of such a 3D Dataset?
Not that I'm aware of > The rasters are large and timeseries long, so reading everything into a > Python NumPy cube is not a good solution. You don't need to have them all simultaneously opened. An idea for an algorithm would be : 1) - allocate one matrix for storing the min of each pixel - allocate one matrix for storing the max of each pixel - allocate one matrix for storing the sum of each pixel - allocate one matrix for storing the sum of the square of each pixel 2) Open sequentially each band and update the above matrixes 3) Then you can use directly the min and max matrices, and deduce the mean and standard deviation from the other matrices. > > Thanks, > Knut-Frode > > _______________________________________________ > gdal-dev mailing list > [email protected] > http://lists.osgeo.org/mailman/listinfo/gdal-dev > _______________________________________________ gdal-dev mailing list [email protected] http://lists.osgeo.org/mailman/listinfo/gdal-dev
