Dear list, I am predicting PM measurements on a spatiotemporal grid with monthly intervals in time. At modeling time, I am looking at the experimental 3D variograms (`wireframes`) but I see that weird decreasing behavior in time (see wireframes_2008-1.eps for January 2008): there is a peak at 0 time lags, then correlation in time is much higher over different days. How can I interpret such variogram? Would it mean that there is a very high spatial variability for values on the same day, whereas temporal variability is significantly lower? Thanks for any hint, (I can provide implementation details in case of need) Piero -------------- next part -------------- An HTML attachment was scrubbed... URL: <https://stat.ethz.ch/pipermail/r-sig-geo/attachments/20130110/a16591a2/attachment.html> -------------- next part -------------- A non-text attachment was scrubbed... Name: wireframes_2008-1.eps Type: application/postscript Size: 19612 bytes Desc: not available URL: <https://stat.ethz.ch/pipermail/r-sig-geo/attachments/20130110/a16591a2/attachment.eps>
Weird experimental 3D spatiotemporal variogram
5 messages · Benedikt Gräler, Piero Campalani, Edzer Pebesma
4 days later
Piero, from the orientation of your graph, I could not see very well what happens at zero-time, zero-space lag. Did you compute a pure-time variogram, i.e. with zero space distance? This one should have a missing zero-value, unless you have duplicate measurements.
On 01/10/2013 03:20 PM, Piero Campalani wrote:
Dear list, I am predicting PM measurements on a spatiotemporal grid with monthly intervals in time. At modeling time, I am looking at the experimental 3D variograms (`wireframes`) but I see that weird decreasing behavior in time (see wireframes_2008-1.eps for January 2008): there is a peak at 0 time lags, then correlation in time is much higher over different days. How can I interpret such variogram? Would it mean that there is a very high spatial variability for values on the same day, whereas temporal variability is significantly lower? Thanks for any hint, (I can provide implementation details in case of need) Piero
_______________________________________________ R-sig-Geo mailing list R-sig-Geo at r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-geo
Edzer Pebesma Institute for Geoinformatics (ifgi), University of M?nster Weseler Stra?e 253, 48151 M?nster, Germany. Phone: +49 251 8333081, Fax: +49 251 8339763 http://ifgi.uni-muenster.de http://www.52north.org/geostatistics e.pebesma at wwu.de
Dear Piero, how did you compute these variograms, using variogramST in gstat (which version?)? Figures 4-7 in gstat's vignette "Spatio-temporal geostatistics using gstat" on CRAN show the missing value for the zero temporal and zero spatial lag class. I could not identify this property in your wireframe plots. To me, the temporal effect looks like being "upside-down". I'll be happy to take a quick look at your script/data in case your problem still remains. Best, Ben
On 15.01.2013 11:51, Edzer Pebesma wrote:
Piero, from the orientation of your graph, I could not see very well what happens at zero-time, zero-space lag. Did you compute a pure-time variogram, i.e. with zero space distance? This one should have a missing zero-value, unless you have duplicate measurements. On 01/10/2013 03:20 PM, Piero Campalani wrote:
Dear list, I am predicting PM measurements on a spatiotemporal grid with monthly intervals in time. At modeling time, I am looking at the experimental 3D variograms (`wireframes`) but I see that weird decreasing behavior in time (see wireframes_2008-1.eps for January 2008): there is a peak at 0 time lags, then correlation in time is much higher over different days. How can I interpret such variogram? Would it mean that there is a very high spatial variability for values on the same day, whereas temporal variability is significantly lower? Thanks for any hint, (I can provide implementation details in case of need) Piero
_______________________________________________ R-sig-Geo mailing list R-sig-Geo at r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-geo
Benedikt Gr?ler ifgi - Institute for Geoinformatics University of Muenster http://ifgi.uni-muenster.de/graeler Phone: +49 251 83-33082 Mail: ben.graeler at uni-muenster.de
Edzer, Ben, (I was missing the vignette on spatio-temporal analysis with gstat.. !!) I'm attaching a clearer figure of the experimental variogram: the /origin/ at 0 spatio-temporal lag is not zero actually. Yes, I am using **variogramST** on a STFDF: the spacetime object looks ok, then I simply call:
variogramST(pm~1, mySTFDF, tlags=0:6)
Ben, out of band I can give you my .Rdata cutout with the STFDF if you're interested. (It seems from the ST vignette that `variogramST` is now merged into `variogram` ?) --------------------------
R.version.string
[1] "R Under development (unstable) (2012-10-03 r60866)" -------------------------- Thanks, Piero
On 15 January 2013 15:15, Benedikt Gr?ler <ben.graeler at uni-muenster.de>wrote:
Dear Piero, how did you compute these variograms, using variogramST in gstat (which version?)? Figures 4-7 in gstat's vignette "Spatio-temporal geostatistics using gstat" on CRAN show the missing value for the zero temporal and zero spatial lag class. I could not identify this property in your wireframe plots. To me, the temporal effect looks like being "upside-down". I'll be happy to take a quick look at your script/data in case your problem still remains. Best, Ben On 15.01.2013 11:51, Edzer Pebesma wrote:
Piero, from the orientation of your graph, I could not see very well what happens at zero-time, zero-space lag. Did you compute a pure-time variogram, i.e. with zero space distance? This one should have a missing zero-value, unless you have duplicate measurements. On 01/10/2013 03:20 PM, Piero Campalani wrote:
Dear list, I am predicting PM measurements on a spatiotemporal grid with monthly intervals in time. At modeling time, I am looking at the experimental 3D variograms (`wireframes`) but I see that weird decreasing behavior in time (see wireframes_2008-1.eps for January 2008): there is a peak at 0 time lags, then correlation in time is much higher over different days. How can I interpret such variogram? Would it mean that there is a very high spatial variability for values on the same day, whereas temporal variability is significantly lower? Thanks for any hint, (I can provide implementation details in case of need) Piero
______________________________**_________________ R-sig-Geo mailing list R-sig-Geo at r-project.org https://stat.ethz.ch/mailman/**listinfo/r-sig-geo<https://stat.ethz.ch/mailman/listinfo/r-sig-geo>
-- Benedikt Gr?ler ifgi - Institute for Geoinformatics University of Muenster http://ifgi.uni-muenster.de/**graeler<http://ifgi.uni-muenster.de/graeler> Phone: +49 251 83-33082 Mail: ben.graeler at uni-muenster.de
______________________________**_________________ R-sig-Geo mailing list R-sig-Geo at r-project.org https://stat.ethz.ch/mailman/**listinfo/r-sig-geo<https://stat.ethz.ch/mailman/listinfo/r-sig-geo>
-------------- next part -------------- An HTML attachment was scrubbed... URL: <https://stat.ethz.ch/pipermail/r-sig-geo/attachments/20130115/67fa614a/attachment.html> -------------- next part -------------- A non-text attachment was scrubbed... Name: wireframes_2008-1.eps Type: application/postscript Size: 20100 bytes Desc: not available URL: <https://stat.ethz.ch/pipermail/r-sig-geo/attachments/20130115/67fa614a/attachment.eps>
3 days later
Thanks Piero; this report revealed a bug in the gstat spatio-temporal variogram computation, which should be fixed in the development version. gstat source code has move to r-forge, development binaries can be downloaded from https://r-forge.r-project.org/projects/gstat/
On 01/15/2013 04:09 PM, Piero Campalani wrote:
Edzer, Ben, (I was missing the vignette on spatio-temporal analysis with gstat.. !!) I'm attaching a clearer figure of the experimental variogram: the /origin/ at 0 spatio-temporal lag is not zero actually. Yes, I am using **variogramST** on a STFDF: the spacetime object looks ok, then I simply call:
variogramST(pm~1, mySTFDF, tlags=0:6)
Ben, out of band I can give you my .Rdata cutout with the STFDF if you're interested. (It seems from the ST vignette that `variogramST` is now merged into `variogram` ?) --------------------------
R.version.string
[1] "R Under development (unstable) (2012-10-03 r60866)" -------------------------- Thanks, Piero On 15 January 2013 15:15, Benedikt Gr?ler <ben.graeler at uni-muenster.de>wrote:
Dear Piero, how did you compute these variograms, using variogramST in gstat (which version?)? Figures 4-7 in gstat's vignette "Spatio-temporal geostatistics using gstat" on CRAN show the missing value for the zero temporal and zero spatial lag class. I could not identify this property in your wireframe plots. To me, the temporal effect looks like being "upside-down". I'll be happy to take a quick look at your script/data in case your problem still remains. Best, Ben On 15.01.2013 11:51, Edzer Pebesma wrote:
Piero, from the orientation of your graph, I could not see very well what happens at zero-time, zero-space lag. Did you compute a pure-time variogram, i.e. with zero space distance? This one should have a missing zero-value, unless you have duplicate measurements. On 01/10/2013 03:20 PM, Piero Campalani wrote:
Dear list, I am predicting PM measurements on a spatiotemporal grid with monthly intervals in time. At modeling time, I am looking at the experimental 3D variograms (`wireframes`) but I see that weird decreasing behavior in time (see wireframes_2008-1.eps for January 2008): there is a peak at 0 time lags, then correlation in time is much higher over different days. How can I interpret such variogram? Would it mean that there is a very high spatial variability for values on the same day, whereas temporal variability is significantly lower? Thanks for any hint, (I can provide implementation details in case of need) Piero
______________________________**_________________ R-sig-Geo mailing list R-sig-Geo at r-project.org https://stat.ethz.ch/mailman/**listinfo/r-sig-geo<https://stat.ethz.ch/mailman/listinfo/r-sig-geo>
-- Benedikt Gr?ler ifgi - Institute for Geoinformatics University of Muenster http://ifgi.uni-muenster.de/**graeler<http://ifgi.uni-muenster.de/graeler> Phone: +49 251 83-33082 Mail: ben.graeler at uni-muenster.de
______________________________**_________________ R-sig-Geo mailing list R-sig-Geo at r-project.org https://stat.ethz.ch/mailman/**listinfo/r-sig-geo<https://stat.ethz.ch/mailman/listinfo/r-sig-geo>
_______________________________________________ R-sig-Geo mailing list R-sig-Geo at r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-geo
Edzer Pebesma Institute for Geoinformatics (ifgi), University of M?nster Weseler Stra?e 253, 48151 M?nster, Germany. Phone: +49 251 8333081, Fax: +49 251 8339763 http://ifgi.uni-muenster.de http://www.52north.org/geostatistics e.pebesma at wwu.de