Introduction:

A grib message contains the grid point values. If you are a C programmer, the grid would naturally be in "C" order. A fortran programmer would have the grid in fortran order. Some people would start from the NW corner and others would start from the SW corner. In grib1, all these scan orders are possible. The grib1 scan orders are,

1. WE:SN
2. WE:NS
3. EW:SN
4. EW:NS
5. SN:WE
6. NS:WE
7. SN:EW
8. NS:EW

AB:CD
   do i = C to D
     do j = A to B
       data
     enddo
   enddo

1. WE|EW:SN
2. WE|EW:NS
3. EW|WE:SN
4. EW|WE:NS

AB|BA:CD
   odd_row=true
   do i = C to D
     if (odd_row) {
       do j = A to B
         data
       enddo
     }
     else {
       do j = B to A
         data
       enddo
     }
     odd_row = not odd_row
   enddo

This "plow the field" order reduces the file size when using regional grids and saving the increments.

Grib2 has 16 scan orders of which 3 are common: WE:SN, WE:NS and WE|EW:SN. Wgrib2 simplfies life by internally converting fields to WE:SN by default.

Usage

-scan

Example

-$ wgrib2 ds_ens.grb -scan
1:0:scan=5 input=WE|EW:SN output=WE:SN
2:218709:scan=5 input=WE|EW:SN output=WE:SN
3:434276:scan=5 input=WE|EW:SN output=WE:SN
...
input=WE|EW:SN   the grib file is in WE|EW:SN scan order
output=WE:SN     the internal registers and output files except for grib
                 will be in WE:SN order, use the -order option to change
                 the output scan order