Second attempt, of 2017-09-11

The COSMO-DE namelist can provide a more precise analogy. Its name is ‘lmrun_uc’. The orders to the CLM (Climate Limited-Area Modeling Community) are expressed with it, and the test files were created with it (by Annika Schomburg).

Table of contents

• Second attempt, of 2017-09-11
• Cropping indices
• Extreme detail
• Footnotes

Some historical research came up with:

dlon=0.00357143
dlat=0.00357143

This works out to be exactly (6 significant digits) one seventh of the former 0.025. In this spirit, ie_tot and je_tot are set to 421 * 7 and 461 * 7, respectively. The latter exceeds 3000, so it is set to 3000 instead, at the expense of the Northern third of Sjoelland, from Kobenhavn northwards.
The full specs are:

origin lon              =10.0
origin lat              =50.0
ie_tot                  =2947
je_tot                  =3000
startlon                =-5.0
startlat                =-5.0
dlon                    =0.00357143
dlat                    =0.00357143
Orography               =ASTER
Orographic Filtering    =No
Land use                =GLC2000
Soil                    =FAO-DSMW
Aerosols                =NASA/GISS
Surface Albedo          =MODIS dry & sat
TERRA-URB               =Fine

Fig. 1: the coarse external data which accompanied the coarse active data
Fig. 2: the new fine external data from CLM (Climate Limited-Area Modeling Community).

Cropping indices

The simple divide-by-7 approach should make cropping easy. I use NCL coordinates, which start at 0. The desired domain in the coarse fields is 160 points in both directions:

1st direction (longitude):  40...199
2nd direction  (latitude): 151...310

The two fields share the South-Western corner. A further advantage of this 7-approach is that the coordinates of a point in the coarse grid reappear for a point in the fine grid, though not the other way around - perfect match. What other points from the fine mesh should usefully map to what point from the coarse grid is less intuitive. This project goes for a square of fine points with the coarse point at the center[11], which incorporates the perfect match.
The coarse coordinate 0 therefore belongs to the fine coordinates –3…+3. The following 7 fine point at 4…10 map to the coarse point 1. Simply adding 7 to the previous interval boundaries allows the increment of the coarse coordinate by 1. Adding n*7 to the boundaries allows the increment of the coarse coordinate by n.
In order to associate 49 fine coordinates with one coarse one, there needs to be 7*160 fine points in either direction.

1st direction (longitude): -3+ 40*7...+3+199*7 =  277...1396
2nd direction  (latitude): -3+151*7...+3+310*7 = 1054...2173

Though any cropping method should do, CDO (Climate data operators) does a fine job of this:

cdo selindexbox,278,1397,1055,2174 domain2017091116126.nc fineexternal.nc

Note that CDO (Climate data operators) complies with conventional counting rather than NCL counting.

Fig. 3: the crop from the coarse external data.

Fig. 4: the crop from the fine data.

Extreme detail

And just to be sure that the values are not uniform, fig. 5 is a detail of the easternmost southernmost part of the cropped area, which should be close the the mountain Schwaebisch-Fraenkischer Wald:

Fig. 5: a detail of the easternmost southernmost part of the cropped area. There are 7 x 7 pixels in this plot.

Footnotes

fn1^. ‘Center’ is an uncontroversial when talking about squares: the centroid, circumcenter and incenter are all the same point.