def get_all_z_agl(self):
"""get all STEM Z cell heights above ground level (from surface to
top of domain).
"""
topo_fname = os.path.join(os.getenv('SARIKA_INPUT'), 'TOPO-124x124.nc')
wrfheight_fname = os.path.join(os.getenv('SARIKA_INPUT'),
'wrfheight-124x124-22levs.nc')
dom = domain.STEM_Domain(fname_topo=topo_fname)
dom.get_STEMZ_height(wrfheight_fname)
if self.z_obs_mean is None:
self.get_z_lev_mean()
n_zlevs = dom.agl.shape[0]
z_all_agl = pd.DataFrame({'z_stem': np.arange(n_zlevs) + 1,
'z_agl': dom.agl[:, self.x_stem,
self.y_stem]})
self.z_obs_mean = pd.merge(z_all_agl, self.z_obs_mean,
how='outer', sort=True)
self.z_obs_mean['ocs_interp'] = self.z_obs_mean['analysis_value']
idx_min = self.z_obs_mean.sample_altitude.idxmin()
idx_max = self.z_obs_mean.sample_altitude.idxmax()
ocs_interp = np.interp(
self.z_obs_mean.z_agl,
rm_nan(self.z_obs_mean.sample_altitude.values),
rm_nan(self.z_obs_mean.analysis_value.values),
left=self.z_obs_mean.analysis_value[idx_min],
right=self.z_obs_mean.analysis_value[idx_max])
nan_idx = np.where(np.isnan(self.z_obs_mean['analysis_value']))
self.z_obs_mean['ocs_interp'].iloc[nan_idx] = ocs_interp[nan_idx]
def get_z_lev_mean(self):
d = domain.STEM_Domain()
topo_dir = os.path.join(os.getenv('PROJ'), 'Data',
'STEM_124x124_NA_inputs')
self.noaa_site.get_stem_xy(d.get_lon(), d.get_lat())
self.noaa_site.get_stem_z(
topo_fname=os.path.join(topo_dir, 'TOPO-124x124.nc'),
wrfheight_fname=os.path.join(topo_dir,
'wrfheight-124x124-22levs.nc'))
bin_min = 0 # bottom of bottom altitude bin, meters
bin_max = 16000 # top of top altitude bin, meters
bin_width = 1000 # size of altitude bins, meters
bin_edges = np.arange(bin_min, bin_max, bin_width)
self.noaa_site.obs['altitude_bin'] = np.digitize(
self.noaa_site.obs.sample_altitude, bin_edges)
groups = self.noaa_site.obs.groupby('altitude_bin')
self.z_obs_mean = groups.mean()
self.z_obs_mean = self.z_obs_mean[['x_stem', 'y_stem',
'sample_altitude',
'analysis_value']]
# # change the index (which is the z level after the groupby to
# # a column)
self.z_obs_mean.reset_index(drop=True, inplace=True)
# a handful of obs are in adjacent STEM cells, resulting in
# non-integral mean x or y cell locations after the mean is
# taken. I think that rounding will pick the the "mode" x and
# y and convert to an integer in one step
self.x_stem = np.int(np.unique(np.round(self.z_obs_mean['x_stem']))[0])
self.y_stem = np.int(np.unique(np.round(self.z_obs_mean['y_stem']))[0])
self.z_obs_mean['x_stem'][:] = self.x_stem
self.z_obs_mean['y_stem'][:] = self.y_stem
self.z_obs_mean['z_stem'] = domain.get_stem_z_from_altitude(
self.z_obs_mean.sample_altitude.values,
stem_x=self.z_obs_mean.x_stem.values,
stem_y=self.z_obs_mean.y_stem.values)
self.z_obs_mean = self.z_obs_mean[['x_stem', 'y_stem',
'z_stem', 'sample_altitude',
'analysis_value']]
# after binning some z levels end up with multiple
# observations; average them together.
self.z_obs_mean = self.z_obs_mean.groupby('z_stem').mean()
self.z_obs_mean.reset_index(inplace=True)
def assemble_bar_plot_data(cpickle_fname=os.path.join(
os.getenv('HOME'), 'STEM_all_runs.cpickle')):
noaa_dir = sau.get_noaa_COS_data_path()
noaa_ocs_dd, ocs_daily = sau.get_JA_site_mean_drawdown(noaa_dir)
d = domain.STEM_Domain()
stem_lon = d.get_lon()
stem_lat = d.get_lat()
(noaa_ocs_dd['stem_x'],
noaa_ocs_dd['stem_y']) = domain.find_nearest_stem_xy(
noaa_ocs_dd.sample_longitude, noaa_ocs_dd.sample_latitude, stem_lon,
stem_lat)
stem_ocs_dd = get_STEM_cos_conc(cpickle_fname)
# place model drawdowns into the data frame
for k, v in stem_ocs_dd.items():
noaa_ocs_dd[k] = stem_ocs_dd[k][noaa_ocs_dd['stem_x'],
noaa_ocs_dd['stem_y']]
return (noaa_ocs_dd)
norm=cos_norm,
dd_map=map_objs[3, i])
fname = '/tmp/maps_basc.pdf'
print("saving {}".format(fname))
fig.savefig(fname)
if draw_site_drawdown_timeseries:
# FIX THIS: where does dd come from? -TWH
plot_site_drawdown_timeseries(dd)
if draw_site_locations_map:
# draw observation sites map
data = noaa_ocs.get_all_NOAA_airborne_data(get_noaa_COS_data_path())
location_map = data.plot_obs_site_locations()
n_amer_domain = domain.STEM_Domain()
n_amer_domain.get_STEM_perimeter_latlon()
location_map.map.plot(n_amer_domain.bnd_lon,
n_amer_domain.bnd_lat,
latlon=True)
location_map.fig.savefig(
os.path.join(os.getenv('HOME'), 'plots',
'noaa_obs_sites_STEMdomain.pdf'))
plt.close(location_map.fig)
if draw_observation_altitude_histograms:
# FIX THIS: where does data come from? -TWH
draw_observation_altitude_histograms(data)
if False:
def top_bounds_QC(top_fname):
"""plot a map of top boundary file to make sure orientation is correct"""
cos = parse_STEM_var('upbound_124x124-climatological_124x124.nc',
varname='CO2_TRACER1')
m = STEM_mapper.Mapper124x124(cos['data'].squeeze())
m.draw_map(fast_or_pretty='pretty',
t_str='Climatological top bounds I/O API',
cmap=plt.get_cmap('Blues'))
m.map.fig.savefig('./top_bounds_ioapi_map.png')
plt.close(m.map.fig)
if __name__ == "__main__":
# --
# get STEM domain parameteres
d = domain.STEM_Domain(Consts().topo_fname)
# --
# read NOAA [COS] observations data
sites = noaa_ocs.get_all_NOAA_airborne_data(Consts().noaa_dir)
sites_list = list(sites.obs.sample_site_code.unique())
# drop WGC because there are no Jul/Aug observations
if 'WGC' in sites_list:
sites_list.remove('WGC')
# drop OIL because of weird-looking column profile from only two
# days of data, with three nearby sites (AAO, WBI, HIL) with much
# more data and very different column means.
if 'OIL' in sites_list:
sites_list.remove('OIL')
lateral_bounds_sites_list = ['THD', 'PFA', 'ESP',
'TGC', 'NHA', 'SCA', 'CMA']
if __name__ == "__main__":
# plot the boundary ring cell number vs. vertical cell number
nc = netCDF4.Dataset(
os.path.join(os.getcwd(), 'ClimatologicalBounds',
'climatological_COS_bdy_22levs_124x124.nc'))
ppb_2_ppt = 1e3 # unit conversion factor
cos = nc.variables['CO2_TRACER1'][:].squeeze() * ppb_2_ppt
cmap, norm = colormap_nlevs.setup_colormap(vmin=cos.min() - 1,
vmax=cos.max() + 1,
nlevs=20,
cmap=plt.get_cmap('Blues'),
extend='neither')
d = domain.STEM_Domain()
d.get_STEMZ_height(wrfheight_fname=os.path.join(
os.getenv('SARIKA_INPUT'), 'wrfheight-124x124-22levs.nc'))
agl_perim = np.array(
[domain.get_2d_perimeter(d.agl[z, ...]).mean() for z in range(22)])
fontsz = 14
matplotlib.rcParams.update({'font.size': fontsz})
fig, ax = plt.subplots(nrows=1, ncols=1, figsize=(7, 6))
cm = ax.pcolormesh(np.arange(500),
agl_perim,
cos,
cmap=cmap,
norm=norm,
linewidth=0,
rasterized=True)