...
- Reciprocal: the mean of the values, weighted by the inverse of their distance from the target point. If one of the geopoints lies exactly on the target point then its value is used directly and the rest of the values discarded.
Exponential Mean: computes the mean of the values weighted (multiplied) by the following:
if Tolerance is not zero:
Mathinline e^{-distance/tolerance^2}- if Tolerance is zero: 1 if the point is on the target point, 0 otherwise
- note that this method, combined with setting Tolerance to zero computes the proportion of points which lie exactly on the target point
- Exponential Sum: performs the same computation as Exponential Mean, but does not finally divide by the total weight. With a Tolerance of zero, this method will compute the number of input points that lie exactly on each target point.
- Nearest Gridpoint Mean: for each target grid point, computes the unweighted mean value of the geopoints for whom this is the closest grid point; any grid point which is not the closest to any geopoints will be given a missing value
- Nearest Gridpoint Sum: for each target grid point, computes the unweighted sum of the values of the geopoints for whom this is the closest grid point; any grid point which is not the closest to any geopoints will be given a missing value
- Nearest Gridpoint Count: for each target grid point, computes the number of geopoints for whom this is the closest grid point. Note that for a regular target grid, this essentially produces a 'heat map', where the value of a grid point will be the number of geopoints within its grid box. This is not necessarily true for quasi-regular grids, e.g. reduced Gaussian, reduced lat/lon or octahedral (which is just a specific type of reduced Gaussian).
...