def bias_matrix(domain, interpolate, save):
if interpolate == True:
if domain == 'pcse':
fname = 'pcse-domain_bias'
lon_new, lat_new = lonp, latp
elif domain == 'rio':
fname = 'rio-domain_bias'
lon_new, lat_new = lonr, latr
NEinterp = griddata(lonc.flatten(), latc.flatten(), NEbias.flatten(), lon_new, lat_new)
SEinterp = griddata(lonc.flatten(), latc.flatten(), SEbias.flatten(), lon_new, lat_new)
SWinterp = griddata(lonc.flatten(), latc.flatten(), SWbias.flatten(), lon_new, lat_new)
NWinterp = griddata(lonc.flatten(), latc.flatten(), NWbias.flatten(), lon_new, lat_new)
ds = xray.Dataset( variables={'NE_bias': (['y','x'], NEinterp),
'SE_bias': (['y','x'], SEinterp),
'SW_bias': (['y','x'], SWinterp),
'NW_bias': (['y','x'], NWinterp)},
coords={'lat': (['y','x'], lat_new),
'lon': (['y','x'], lon_new)},
attrs={'Name' : 'CFSR - WRF regional averaged-bias for January 2013',
'Description' : 'Bias values were calculated by\
subtratcting WRF regional model from CFSR product.\
WRF model results used here has a spatial resolution\
of 9km while CFSR is ~25km. The computation regards\
the MEDIAN over January 2013'} )
if save == True:
ds.to_netcdf( ROOTDIR + fname + '.nc' )
elif interpolate == False:
fname = 'cfs-domain_bias'
ds = xray.Dataset( variables={'NE_bias': (['y','x'], NEbias),
'SE_bias': (['y','x'], SEbias),
'SW_bias': (['y','x'], SWbias),
'NW_bias': (['y','x'], NWbias)},
coords={'lat': (['y','x'], latc),
'lon': (['y','x'], lonc)},
attrs={'Name' : 'CFSR - WRF regional averaged-bias for January 2013',
'Description' : 'Bias values were calculated by\
subtratcting WRF regional model from CFSR product.\
WRF model results used here has a spatial resolution\
of 9km while CFSR is ~25km. The computation regards\
the MEDIAN over January 2013'} )
if save == True:
ds.to_netcdf( ROOTDIR + fname + '.nc' )
wrf['west_east'].values = wrf['XLONG'][0,0,:].values
u = wrf['U10'].sel(south_north=slice(Glims[2],Glims[3]),
west_east=slice(Glims[0], Glims[1]))
v = wrf['V10'].sel(south_north=slice(Glims[2],Glims[3]),
west_east=slice(Glims[0], Glims[1]))
lonw = u.coords['XLONG'][0,...].values
latw = u.coords['XLAT'][0,...].values
U = np.ones( (u.shape[0], latc.shape[0], lonc.shape[1]) )
V = U.copy()
for k in range(u.shape[0]):
U[k] = griddata(lonw.flatten(), latw.flatten(), u[k].values.flatten(), lonc, latc)
V[k] = griddata(lonw.flatten(), latw.flatten(), v[k].values.flatten(), lonc, latc)
WRFint, WRFdir = uv2intdir(-U, -V)
NEwrf, SEwrf, SWwrf, NWwrf = wind_quadrant(WRFdir, WRFint, type='median')
############################################################################
# Computing BIAS (CFSv2 - WRF)
############################################################################
NEbias = NEcfs - NEwrf
SEbias = SEcfs - SEwrf
SWbias = SWcfs - SWwrf
NWbias = NWcfs - NWwrf
