Closed b-barton closed 1 year ago
It turns out that the trim_domain_size()
method does sometimes need to use unCOAsTlike nav_lat
and nav_lon
. This is a legacy issue that I haven't found a clean solution for. Instead I propose a try/except solution (below).
This will get resolved with a PR #607
# Find the corners of the cut out domain.
try:
[j0, i0] = self.find_j_i_domain(
lat=self.dataset.latitude[0, 0], lon=self.dataset.longitude[0, 0], dataset_domain=dataset_domain
)
[j1, i1] = self.find_j_i_domain(
lat=self.dataset.latitude[-1, -1], lon=self.dataset.longitude[-1, -1], dataset_domain=dataset_domain
)
debug(f"trim_domain_size(): USED dataset.longitude")
except: # if called before variables are re-mapped. Not very pretty...
[j0, i0] = self.find_j_i_domain(
lat=self.dataset.nav_lat[0, 0], lon=self.dataset.nav_lon[0, 0], dataset_domain=dataset_domain
)
[j1, i1] = self.find_j_i_domain(
lat=self.dataset.nav_lat[-1, -1], lon=self.dataset.nav_lon[-1, -1], dataset_domain=dataset_domain
)
debug(f"trim_domain_size(): USED dataset.nav_lon")
fixed
trim_domain_size() in gridded.py needs self.dataset.nav_lat and self.dataset.nav_lon but they are not required in the dataset json.