2

I have a table of data with such structure:

longitude_for_data = c(32,68,89,145,176, -14, -42)   
latitude_for_data  = c(22,98,21,13 , 16,-134,-102)
Z_for_data         = c(30,40,60,70 , 20,  40, 100)
data_to_analise    = cbind(longitude_for_data,latitude_for_data,Z_for_data)

I want to make grid and fill it using linear kriging with variogram(as a variant -simple kriging):

longitude    = seq(-180,180,by=5.0)
latitude     = seq(-180,180,by=2.5)
lat_long_grid= expand.grid(longitude,latitude)

as a result I want to get something like that:

   lon  lat Z-kriging
1 -180 -180 32
2 -175 -180 35
3 -170 -180 38 
4 -165 -180 39
5 -160 -180 40
6 -155 -180 41

How should I use kriging function to recieve that data table?

  • some notes: 1. latitudes outside -90 to +90 degrees are invalid. Are these geographic lat-long coordinates or something else? 2. if they are lat-long degrees then you need to make sure your variogram and kriging is using the great circle distance, otherwise you will end up with a discontinous prediction at the pole and distortions everywhere else. – Spacedman Mar 17 at 10:23
  • I saw this peoblem,thank you. It was my mistake – Владимир Кузовкин Mar 18 at 20:51
3

Package sf also provides sf::st_make_grid

library(sf)
library(gstat)

data <- as.data.frame(data_to_analise)
colnames(data)[1:3] <- c("longitude", "latitude", "z")

st.sf <- st_as_sf(x = data, coords = c("longitude", "latitude"), crs=NA)
colnames(st.sf)[1] <- "z"

vgm1 <- gstat::variogram(z~1, st.sf)
fit1 <- gstat::fit.variogram(vgm1, model = gstat::vgm("Gau")) # fit model
krig <- gstat::krige(z~1, st.sf, gbbx, model=fit1)

  • You can skip the intermediate sf and data.table representation by making a data frame and using sp::coordinates(data)=c("longitude","latitude"). At least until gstat can take sf objects! – Spacedman Mar 17 at 10:25
2

Thank you very much for your help. I solved problem in such way:

library(gstat)
library(sp)
longitude_for_data = c(32,68,89,145,176, -14, -42)   
latitude_for_data  = c(22, 8,21,13 , 16,- 34,-12)
Z_for_data         = c(10,20,30,40 , 50,  60, 70)
data_together      = cbind(longitude_for_data,latitude_for_data,Z_for_data)
data_to_analise    = as.data.frame(data_together)
coordinates(data_to_analise) <- ~longitude_for_data + latitude_for_data
proj4string(data_to_analise) <- CRS("+init=epsg:4326")

longitude    = seq(-180,180,by=5.0)
latitude     = seq(-90 ,90 ,by=2.5)

lat_long_together= expand.grid(longitude,latitude)
lat_long_grid    =as.data.frame(lat_long_together)
colnames(lat_long_grid)[1:2] <- c("longitude", "latitude")
coordinates(lat_long_grid) <- ~longitude + latitude
proj4string(lat_long_grid) <- CRS("+init=epsg:4326")
m <- vgm(.59, "Sph", 874, .04)

result= krige(Z_for_data~1, data_to_analise, lat_long_grid, model = m)

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.