I am using google earth engine to access Landsat 8 NDVI data (30m resolution). I have population data at 100m res so I have first reduced the resolution to match that of the population data set:
var ndvi_100 = ndvi_clipped
.reproject({crs: 'EPSG:4326',scale: 100})
.reduceResolution({reducer: ee.Reducer.mean()});
Because many of the epidemiology studies use radius around a person's residence to determine exposure to green space, I'd like to change each pixel's value to be an average of the surrounding area. I've done this using the .reduceNeighborhood function and selecting each of the surrounding pixels with the kernel.square option.
var ndvi_100_blur = ndvi_100
.reduceNeighborhood({reducer: ee.Reducer.mean(),kernel: ee.Kernel.square({radius:1})});
I see that there is an option to use a kernel.circle and select meters as the radius distance rather than pixels (which is ideally what i want, i.e. to test radii of 100, 250, and 500m), but I'm unclear how this works. How is gee calculating this average using partial pixels? Just want to make sure I understand what's going on before using.
Thank you in advance for your help.