Access to data (without downloading it)

[frafra]

GDAL supports accessing to data remotely without having to download it in advance.

Should we mention it and provide some code snippets on how to do it in Python/R?

Here is an example of using GDAL virtual file system to access a remote NetCDF file without downloading it:

url='NETCDF:/vsicurl/https://thredds.met.no/thredds/fileServer/senorge/seNorge_2018/Archive/seNorge2018_2006.nc'
data2006 = stars::read_stars(url, sub='rr', proxy=TRUE)
[...]
st_crs(data2006)
reg1_st <- reg1 %>% st_as_sf(.,
                             coords = c("longitude", "latitude"), 
                             crs=4326)

reg1_st <- st_transform(reg1_st, crs=st_crs(data2006))
reg1_st$rr <- st_extract(data2006, reg1_st, time_column="date")

This snippet uses R stars library.

Such a URL works with gdalinfo from command line and on a wide variety of software using GDAL.

Maybe Benjamin Cretois can help with that as well?

[BenCretois]

A more reproducible code below:

# Create a dataset - assuming the coordinates are in CRS=4326
data <- data.frame(
  longitude = c(11.95501, 11.95501, 11.95498, 11.95493, 11.95487, 11.95497),
  latitude = c(65.67812, 65.67814, 65.67821, 65.67808, 65.67809, 65.67810),
  date = as.Date(rep("2006-03-26", 6))
)

# fetch the ncdf file for a specific year
  url=paste0('NETCDF:/vsicurl/https://thredds.met.no/thredds/fileServer/senorge/seNorge_2018/Archive/seNorge2018_', year,'.nc')

  # Read the netcdf file from the url
  netcdf_file=stars::read_stars(url, sub='rr', proxy=TRUE) 

  # Transform the dataset into an sf object
  data_st <- data %>% st_as_sf(.,
                               coords = c("longitude", "latitude"), 
                               crs=4326)

  # Reproject the CRS to match the CRS of the netcdf
  data_st <- st_transform(data_st, crs=st_crs(netcdf_file))
  data_st$rr <- st_extract(netcdf_file, data_st, time_column="date")