eguil / Density_bining

Density bining code

2 stars 5 forks source link

Bug in interpolation at date line or other longitudes #53

Open eguil opened 7 years ago

eguil commented 7 years ago

Some models have undefined values around the date line (i-179 & 180, eg CanESM2) or at given longitudes (eg CCSM4). Need to check interpolation option in binDensity.py

eguil commented 7 years ago

Models with issues (so far I only checked those that have histNat as well as hist runs):

CanESM2 (i=179/180, all j)
CCSM4 (i=139,140 & j=1->145)
CSIRO-Mk3-6-0 (i=178/179, all j)
GISS-E2-H has another issue in a patch in the west Atlantic where nothing is defined !
GISS-E2-R (i=179/180, all j)
HadGEM2-ES (i=179, all j)
MIROC-ESM (i=179/180, all j)
MIROC-ESM-CHEM (i=179/180, all j)

durack1 commented 7 years ago

@eguil we may need to add the additional argument(s) to binDensity.py:

regridObj = CdmsRegrid(ingrid,outgrid,depthBini.dtype,missing=valmask,regridMethod='distwgt',regridTool='esmf')

becomes:

regridObj = CdmsRegrid(ingrid, outgrid, depthBini.dtype, missing=valmask, regridMethod='distwgt', regridTool='esmf', coordSys='deg', diag = {},periodicity=1)

In particular the diag = {} will provide a bunch of diagnostics that may be useful to interrogate if the issue persists..

There is some discussion of this at https://github.com/UV-CDAT/uvcdat/issues/147

durack1 commented 7 years ago

@eguil we may need to add the additional argument(s) to binDensity.py:

regridObj = CdmsRegrid(ingrid, outgrid, depthBini.dtype, missing=valmask, regridMethod='distwgt',
regridTool='esmf')

becomes:

regridObj = CdmsRegrid(ingrid, outgrid, depthBini.dtype, missing=valmask, regridMethod='distwgt',
regridTool='esmf', coordSys='deg', diag = {}, periodicity=1)

In particular the diag = {} will provide a bunch of diagnostics that may be useful to interrogate if the issue persists..

I am still trying to locate the comprehensive documentation for this functionality (see https://github.com/UV-CDAT/cdms/issues/19) there is some discussion at https://github.com/UV-CDAT/uvcdat/issues/147

durack1 commented 7 years ago

We will need to update code with UVCDAT 2.10 now uses ESMPy rather than ESMF

durack1 commented 5 years ago

35 has a graphic of the issue, which appears associated with the tripolar grid of ORCA about 20N in the North Atlantic and Arctic, where the grid becomes distorted