I'm rasterizing some wind data from oos.soest.hawaii.edu/erddap/griddap/NCEP_Global_Best to use as an overlay of a final image, but I'm having issues calculating the bounding box of the final image.

Can I safely assume, giving the data is continuous, that each latitude/longitude corresponds to the center of my final pixel? If not, which metadata keys should I have been looking?

In the case that the position is indeed referencing the center of the pixel, then can I say the following?

bounding_south = smallest_latitude - (latitude_resolution / 2)
bounding_north = largest_latitude + (latitude_resolution / 2)

bounding_west = smallest_longitude - (latitude_resolution / 2)
bounding_east = largest_longitude + (latitude_resolution / 2)

Unfortunately I couldn't access the data using gdal_translate to test it.


Doing an ncdump on the data from one of the soest files gives this chunk of metadata for the precipitation flux variable:

    double pratesfc(time, latitude, longitude) ;
            pratesfc:_FillValue = -999000000. ;
            pratesfc:colorBarMaximum = 0.005f ;
            pratesfc:colorBarMinimum = 0.f ;
            pratesfc:coordinates = "time_run time latitude longitude" ;
            pratesfc:ioos_category = "Meteorology" ;
            pratesfc:long_name = "rainfall rate" ;
            pratesfc:missing_value = -999000000. ;
            pratesfc:standard_name = "precipitation_flux" ;
            pratesfc:units = "kg m-2 s-1" ;

The data in netcdf files should correspond exactly to the coordinate points in the file. If you are representing pixels as little squares, (contrary to A Pixel Is Not A Little Square) then each of the data points is a 1-pixel area in the raster, then it would be center-pixel data, and the pixel edges would be +/- resolution/2. wider than the coordinates in the netcdf file.

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.