I am wondering the best way of knowing how to define the area of each raster cell in global data. Specifically, I downloaded all ~400 tiles of MODIS land cover data, mosaiced them into one, and now am summarizing land cover areas at admin levels. The data come downloaded as custom spheroid sinusoidal (~463m resolution). Each subset appears to contain cells all of the same size. My question is after I mosaic the data into one raster, if I changed the projection to WGS84, would I then have to account for the cell sizes changing with latitude? Does it depend on what the projection the original data were in? I can't think of any way to test this, since when looking at the data in ArcMap the cells all measure the same, using the measure tool, which may or may not just be a product of the dataframes coordinate system. Would converting a subset of the raster to shapefile and then calculating the area tell me anything?
Overall my question is a basic one, does every raster in a geographic coordinate system by its nature have cell sizes that change with latitude? Or are there exceptions?