I'm trying to figure out the 'established' methodology for the following: a colleague of mine has some population data for Russia. The data are supplied in WGS84 with each cell equal to 0.0083333333 of a unit, where the angular unit is degrees. I believe that fraction is a 0.5 x 1/60th of a degree, so half a minute, or 30 arc seconds.
We want to create a projected grid around each point to do some analysis on the data. Ideally we want a 1km grid. That means we have a mismatch between a projected grid and the population data in geographic. How best to proceed?
What projection should I use for Russia? It strikes me as too big to use a single projection, so perhaps we need to reduce our study area extent, and then reproject the original data to the new grid. Since our data represent population, do I need to worry about what happens behind the scenes during the reprojection? Ie, if two old GCS cells fall within one new projected cell, should I take the sum or the mean? Presumably if we want the total population between the unprojected and projected datasets to be the same, we need to take the sums of values during reprojection, thus preserving the total.
But I'm not 100% sure. If anyone can help me figure out what normally happens when transforming between GCS and PCS that would be extremely helpful. Thanks
