[R-sig-Geo] Error in Universal Kriging with 2nd order trend in geoR

[GEMA FERNANDEZ-AVILES CALDERON]

Dear list,

I have two problems related with UK in geoR package.

Q1) I can't perform UK with a 2nd order trend. The krige.conv() and ksline() functions reports me the same error all the time.

Q2) I can't understand how carrying out an UK with 1st order trend and choosing moving neighborhood of prediction with ksline() function, the smaller the neigborhood, the highest the error prediction.

I can send you the data.

Thanks in advance,
Gema





###****###

### Q1 ###

###****###

> summary(obj.uk.house)
Number of data points: 1000

Coordinates summary
    longitud latitud
min   -3.789   40.34
max   -3.555   40.53

Distance summary
      min       max
6.708e-06 2.511e-01

Data summary
   Min. 1st Qu.  Median    Mean 3rd Qu.    Max.
   7.14    7.94    8.17    8.17    8.40    9.42

##semivariogram
wls.nwind  <- variofit( bin.nwind, cov.model="sph", ini = c(0.05, 0.05), nug = 0.03)


## UK, 2nd order trend
UK.krig.2 <- krige.conv(obj.uk.house, loc = loci.cuadro, borders = bordes.munimadrid,
              krige = krige.control(obj.model=wls.nwind, trend.l ="2nd", trend.d = "2nd"))

krige.conv: results will be returned only for prediction locations inside the borders
krige.conv: model with mean given by a 2nd order polynomial on the coordinates
Error in solve.default(ttivtt, crossprod(ivtt, as.vector(data))) :
  sistema es computacionalmente singular: número de condición recíproco = 7.65008e-19



###****###

### Q2 ###

###****###

uk.2.nwind.250= ksline(obj.uk.house, coords = obj.uk.house$coords, data = obj.uk.house$data,
       loc = loci.cuadro, borders = bordes.munimadrid,
       cov.model = "sph", cov.pars = c(0.059, 0.045), nug = 0.037,
       lambda = 1, m0 = "kt", nwin = 250,
       trend = 1, d = 2 )







	[[alternative HTML version deleted]]