mccauley at ecn.purdue.edu
Thu Jun 11 09:03:13 EDT 1992
One more touchdown on moon...
s.medp.shar.Z - performs a median polish ("another statistical dance") on
sites data. A vector grid (possibly rotated) is input and
sites find their nearest node. Then, medians are removed
from rows, then columns, then rows, etc, until
convergence. The residuals (and optionally row, column,
and all "effects") are stored in a vector output file.
This will most likely be replaced with v.medp.shar.Z in the near future...
I just thought that I would make this available in case the few people
who know of this algorithm would like to test it out. A man page (with
a reference) is included for those who would like to find out more about
Basically, it detrends data so that you minimize problems with
non-stationarity when attempting to use kriging. The row, column, and
all "effects" *could* be used to create a trend surface (though this
has not been implemented).
More information about the grass-user