rfsim problem

[john walter <jfwalter@vims.edu>]

I have a problem with the SPLUS spatial stat function rfsim where
when I create data sets larger than 10x10 or 100 data points, the
function induces a trend in the simulated data. Simulated data values in
the lower left of the created data set slope down to low values,
resulting in a situation where the data set no longer possess a variogram
resembling the one that created it. Does anyone know if this is a problem
within rfsim or whether this is a numerical problem with the LU
decomposition of the covariance matrix? Either way is there a better way
to create a spatially autocorrelated dataset larger than 10x10?
Thanks

John

John Walter
Virginia Institute of Marine Science
P.O. Box 1346
Gloucester Point, VA 23062
(804) 684-7316
jfwalter@vims.edu