Skip to content

units for spatial range in stvariogram gstat

3 messages · Edzer Pebesma, sara osama

sara osama
# 
Dear All,
I have question concerning the unit of measurement for the range parameter
in the fit.stvariogram in gstat, i have fitted the following spatio
temporal variogram for data on monthly averages pf pm10 and the longitude
and latitude are in decimel format
SpatialPoints:
     longitude latitude
1     30.92500 29.95300
61    31.36000 30.26000
121   31.33300 29.84720
 and this is the commands for fitting the variogram
variogramST(lnpm10only~1,data=lnpm10_stfdf,cutoff=0.8,assumeRegular=T,na.omit=T)
nugget=0),time = vgm(psill=500,"Mat", range=50, nugget=0,kappa=0.3), joint
= vgm(psill=10,"Mat", range=100, nugget=1), stAni=200)
method="L-BFGS-B",lower=pars.l,upper=pars.u)
sill.s      range.s     nugget.s       sill.t      range.t
nugget.t
  1.73598444  50.63845035   0.03212857   0.01570379  49.99811448
0.00000000
     sill.st     range.st    nugget.st         anis
  0.00000000 100.00004052   0.09151787 200.00000000
Edzer Pebesma
# 
Your email is not nearly enough for a reader to figure out what you did.
However, if your data are long/lat, and you instructed gstat about that
(i.e., proj4string(pm10_sp) indicates coordinates are degrees) then
gstat computes and returns distances in km.
On 17/05/17 20:56, sara osama wrote:

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      sara osama
    

    

      
      #
    

  


  

      
Dear Prof. Pebesma,

Thanks for reply I will try to make it  more clear

the coordinates are converted to decimal degree format using the equation
DD = d + (min/60) + (sec/3600)

and then projected using

            

      
      
      
    

        
SpatialPoints:
     longitude latitude
1     30.92500 29.95300
61    31.36000 30.26000
 and the variogram is fitted using

            

      
      
      
    

        
assumeRegular=T,na.omit=T)

            

      
      
      
    

        
nugget=0),time = vgm(psill=500,"Mat", range=50, nugget=0,kappa=0.3), joint
= vgm(psill=10,"Mat", range=100, nugget=1), stAni=200)

            

      
      
      
    

        
method="L-BFGS-B",lower=pars.l,upper=pars.u)

*the s.range parameter extracted was 50.6 but if you take a look on  the
figure attached the largest distance on the x-axis is 0.6 ??????*

*so i think the unit on the figure is not the unit of the extracted
parameter. *

by the way the largest ditance between any points in my data set is
approximately 190 km


Thanks again




On Wed, May 17, 2017 at 10:08 PM, Edzer Pebesma <

        
edzer.pebesma at uni-muenster.de> wrote: