Two possible minor bugs to report:
- Possible error on my part, or not really a bug since it can be changed by setting Hstart = 0 (but it’s 12 by default), but I noticed that when generating Mercator OBCs with pytools 2.0.3, the first month generated is OK, in the sense that the first day of the OBCs is <= the first day of the simulation (—> interpolation ok). The following months start at DstHrt-start (12h by default) on the first day —> CROCO crashe. One solution would be to set Hstart=0, but we would expect to find the same file at the beginning of the loop or somewhere else.
Example, it will surely be simpler:
1858 is 01/02/2005
file generated in the first month in the loop (so Mstart=2):
bry_time = 1857.5, 1858.5, 1859.5, 1860.5, 1861.5, 1862.5, 1863.5, 1864.5,
1865.5, 1866.5, 1867.5, 1868.5, 1869.5, 1870.5, 1871.5, 1872.5, 1873.5,
1874.5, 1875.5, 1876.5, 1877.5, 1878.5, 1879.5, 1880.5, 1881.5, 1882.5,
1883.5, 1884.5, 1885.5, 1886.5, 1887.5;
February file generated in a loop in the second month (therefore Mstart=1):
bry_time = 1858.5, 1859.5, 1860.5, 1861.5, 1862.5, 1863.5, 1864.5, 1865.5,
1866.5, 1867.5, 1868.5, 1869.5, 1870.5, 1871.5, 1872.5, 1873.5, 1874.5,
1875.5, 1876.5, 1877.5, 1878.5, 1879.5, 1880.5, 1881.5, 1882.5, 1883.5,
1884.5, 1885.5, 1886.5, 1887.5;
- CROCO2CROCO: I’ve been told:
"By creating boundary conditions from the low-resolution model outputs, I obtain one file per month: croco_bry_zoom_offline_YxxxxMxx.nc. The problem is that this file only contains the 31 days of the month. So when I run the medium-resolution model, it fails on the last day because it searches for the boundary conditions for the following day in order to interpolate. Is there a way to prevent it from searching? Or do you simply recommend that I group all the croco_bry_YxxxxMxx.nc and croco_blk_YxxxxMxx.nc files into a single giant netcdf file, such as croco_bry.nc and croco_blk.nc? —>
Basically, a date is missing > the final date —> CROCO stops. An easy fix to to change MONTHLY to YEARLY or FULL, but if we want small monthly files, this causes bugs, even though we can always work around it manually by changing the end date… So, to sum up, there must be a “+1” missing somewhere in the date selection (I don’t know pytools well enough yet to say where).
cheers