# Get regional averages
da_81reg = metrics.agg_by_domain(da_grid=da_81, ds_region=ds_region)
# In[7]:
# Get date 30 days ago
ctime = np.datetime64(datetime.datetime.now())
lag_time_30days = ctime - np.timedelta64(30, 'D')
lag_time_90days = ctime - np.timedelta64(90, 'D')
last_sept = metrics.get_season_start_date(ctime)
# Select recent period
da_81_30 = da_81.where(da_81.time >= lag_time_30days, drop=True)
# Aggregate over domain
da_81_30_avg = metrics.calc_extent(da_81_30, ds_region, fill_pole_hole=True)
# Select recent period
da_81_3m = da_81.where(da_81.time >= lag_time_90days, drop=True)
# Aggregate over domain
da_81_3m_avg = metrics.calc_extent(da_81_3m, ds_region, fill_pole_hole=True)
# In[8]:
# # Test regional sums
# da_test_nolake = ((da_81>=0.15).astype('int') * ds_region.area).sum(dim='x').sum(dim='y')/(10**6)
# da_test = esio.calc_extent(da_81, ds_region, fill_pole_hole=True)
# %matplotlib inline
# da_81reg.sum(dim='nregions').plot(color='r', label='Sum of Regional Extents', linewidth=4)
# da_test_nolake.plot(color='b', label='Sum panArctic, Including Lakes')
# da_test.plot(color='k', label='Sum panArctic, Excluding Lakes')
def Update_Model_Aggs():
'''Calculates pan-arctic and regional extents from different forecast models'''
E = ed.EsioData.load()
model_dir = E.model_dir
# Directories
# Define models to plot
all_models = list(E.model.keys())
all_models = [
x for x in all_models if x not in ['piomas', 'MME', 'MME_NEW']
] # remove some models
# all_models = ['uclsipn']
runType = 'forecast'
updateall = False
ds_region = xr.open_mfdataset(
os.path.join(E.grid_dir, 'sio_2016_mask_Update.nc')).load()
for model in all_models:
print(model)
data_dir = E.model[model][runType]['sipn_nc']
data_out = os.path.join(model_dir, model, runType, 'sipn_nc_agg')
if not os.path.exists(data_out):
os.makedirs(data_out)
all_files = glob.glob(os.path.join(data_dir, '*.nc'))
print("Found ", len(all_files), " files.")
if updateall:
print("Updating all files...")
else:
print("Only updating new files")
# Remove any "empty" files (sometimes happends with ecmwf downloads)
all_files_new = []
for cf in all_files:
if os.stat(cf).st_size > 0:
all_files_new.append(cf)
else:
print("Found empty file: ", cf,
". Consider deleting or redownloading.")
all_files = sorted(all_files_new) # Replace and sort
# For each file
for cf in all_files:
# Check if already imported and skip (unless updateall flag is True)
# Always import the most recent two months of files (because they get updated)
f_out = os.path.join(data_out,
os.path.basename(cf)) # netcdf file out
if not updateall:
if (os.path.isfile(f_out)) & (cf not in all_files[-2:]):
print("Skipping ", os.path.basename(cf),
" already imported.")
continue # Skip, file already imported
ds = xr.open_mfdataset(
cf,
chunks={
'fore_time': 10,
'ensemble': 5,
'init_time': 10,
'nj': 304,
'ni': 448
},
parallel=True
) # Works but is not eiffecent 5-15 mins wall time
ds.rename({'nj': 'x', 'ni': 'y'}, inplace=True)
# Calc panArctic extent
da_panE = metrics.calc_extent(da=ds.sic, region=ds_region)
da_panE['nregions'] = 99
da_panE['region_names'] = 'panArctic'
# Calc Regional extents
da_RegE = metrics.agg_by_domain(da_grid=ds.sic,
ds_region=ds_region)
# Merge
ds_out = xr.concat([da_panE, da_RegE], dim='nregions')
ds_out.name = 'Extent'
ds_out.load(
) # This prevents many errors in the dask graph (I don't know why)
# # Save regridded to netcdf file
ds_out = None # Memory clean up
da_panE = None
da_RegE = None
ds = None
print('Saved ', f_out)
print("Finished...")
# Add lat and lon dimensions
ds_sic.coords['lat'] = ds_lat_lon.lat
ds_sic.coords['lon'] = ds_lat_lon.lon
# Stereo projected units (m)
dx = dy = 25000
xm = np.arange(-3850000, +3750000, +dx)
ym = np.arange(+5850000, -5350000, -dy)
ds_sic.coords['xm'] = xr.DataArray(xm, dims=('x'))
ds_sic.coords['ym'] = xr.DataArray(ym, dims=('y'))
# Calculate extent and area
# ds_sic['extent'] = ((ds_sic.sic>=0.15).astype('int') * ds_region.area).sum(dim='x').sum(dim='y')/(10**6)
ds_sic['extent'] = metrics.calc_extent(ds_sic.sic,
ds_region,
fill_pole_hole=True)
# ds_sic['extent'] = ds_sic['extent'] + (ds_sic.hole_mask.astype('int') * ds_region.area).sum(dim='x').sum(dim='y')/(10**6) # Add hole
ds_sic['area'] = (ds_sic.sic * ds_region.area).sum(dim='x').sum(
dim='y') / (10**6) # No pole hole
# Save to netcdf file
ds_sic.to_netcdf(os.path.join(nc_dir, nf.split('.b')[0] + '.nc'))
ds_sic = None
# # Calculate extent and area (saved to separte file)
# if len(new_files) > 0 : # There were some new files
# print('Calculating extent and area...')
# ds_all = xr.open_mfdataset(os.path.join(nc_dir,'*.nc'), concat_dim='time',
# autoclose=True, compat='no_conflicts',
# data_vars=['sic'])