Comparisons between nowcast, spinup, and observed salinity in the Haro Strait

In [1]:

import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
import ACTDR
import netCDF4 as nc
from salishsea_tools import viz_tools
import numpy as np

import comparisons

%matplotlib inline

In [2]:

sns.set_style('darkgrid')

In [3]:

grid_B = nc.Dataset('/data/nsoontie/MEOPAR/NEMO-forcing/grid/bathy_meter_SalishSea2.nc')

Load observations¶

In [4]:

ACTDR.load_dat('SOG_2000.dat')

> open  SOG_2000.dat
> load CTD_DAT
> load STANDARD_KEYS
> close  SOG_2000.dat
> complete

In [5]:

data = pd.DataFrame(ACTDR.CTD_DAT)

lon_min=-123.4; lat_min=48.3;
lon_max=-123; lat_max=48.85;

data_region = comparisons.isolate_region(data, lon_min, lon_max, lat_min, lat_max)

In [6]:

fig,axm = plt.subplots(1,figsize=(5,5))
data_region.plot(x='Longitude',y='Latitude',kind='scatter', marker='o',ax=axm)
viz_tools.plot_coastline(axm,grid_B,coords='map')
axm.set_xlim([lon_min,lon_max])
axm.set_ylim([lat_min, lat_max])

Out[6]:

(48.3, 48.85)

In [7]:

data_region.hist('Month',bins=np.arange(0.5,13.5))

Out[7]:

array([[<matplotlib.axes._subplots.AxesSubplot object at 0x7fccf475a810>]], dtype=object)

In [8]:

data_region.hist('Year', bins= np.arange(data_region['Year'].min()-0.5, data_region['Year'].max() +1.5 ))
ax=plt.gca()
ax.get_xaxis().get_major_formatter().set_useOffset(False)

Data model comparisons¶

Compare model salinity profiles with observed in this region. Apply to both spinup and nowcasts.
Focus on all depths

In [9]:

paths = {'nowcast': '/data/dlatorne/MEOPAR/SalishSea/nowcast/',
        'spinup': '/ocean/dlatorne/MEOPAR/SalishSea/results/spin-up/'}
zmax= 250

November

In [10]:

month=11
fig=comparisons.compare_model_obs(month,2003,'Salinity',data_region,paths['spinup'],zmax=zmax,vmin=24,vmax=33,
                                  xmin=lon_min,xmax=lon_max,ymin=lat_min,ymax=lat_max)
fig=comparisons.compare_model_obs(month,2014,'Salinity',data_region,paths['nowcast'],zmax=zmax, vmin=24,vmax=33,
                                  xmin=lon_min,xmax=lon_max,ymin=lat_min,ymax=lat_max)
data_month=data_region[data_region['Month']==month]
data_month.hist('Year',bins = np.arange(data_month['Year'].min()-0.5,data_month['Year'].max()+1.5 ))
ax=plt.gca()
ax.get_xaxis().get_major_formatter().set_useOffset(False)

December

In [11]:

month=12
fig=comparisons.compare_model_obs(month,2003,'Salinity',data_region,paths['spinup'],zmax=zmax,vmin=24,vmax=33,
                                  xmin=lon_min,xmax=lon_max,ymin=lat_min,ymax=lat_max)
fig=comparisons.compare_model_obs(month,2014,'Salinity',data_region,paths['nowcast'],zmax=zmax, vmin=24,vmax=33,
                                  xmin=lon_min,xmax=lon_max,ymin=lat_min,ymax=lat_max)
data_month=data_region[data_region['Month']==month]
data_month.hist('Year',bins = np.arange(data_month['Year'].min()-0.5,data_month['Year'].max()+1.5 ))
ax=plt.gca()
ax.get_xaxis().get_major_formatter().set_useOffset(False)

Hypothesis is too much mixing here. But some model profiles are stratified more than the observations. Certainly there is a strong freshening trend in the deeper water. But is that because of too much mixing or too much fresh water?

January

In [12]:

month=1
fig=comparisons.compare_model_obs(month,2003,'Salinity',data_region,paths['spinup'],zmax=zmax,vmin=24,vmax=33,
                                  xmin=lon_min,xmax=lon_max,ymin=lat_min,ymax=lat_max)
fig=comparisons.compare_model_obs(month,2015,'Salinity',data_region,paths['nowcast'],zmax=zmax, vmin=24,vmax=33,
                                  xmin=lon_min,xmax=lon_max,ymin=lat_min,ymax=lat_max)
data_month=data_region[data_region['Month']==month]
data_month.hist('Year',bins = np.arange(data_month['Year'].min()-0.5,data_month['Year'].max()+1.5 ))
ax=plt.gca()
ax.get_xaxis().get_major_formatter().set_useOffset(False)

February

In [13]:

month=2
fig=comparisons.compare_model_obs(month,2003,'Salinity',data_region,paths['spinup'],zmax=zmax,vmin=24,vmax=33,
                                  xmin=lon_min,xmax=lon_max,ymin=lat_min,ymax=lat_max)
fig=comparisons.compare_model_obs(month,2015,'Salinity',data_region,paths['nowcast'],zmax=zmax, vmin=24,vmax=33,
                                  xmin=lon_min,xmax=lon_max,ymin=lat_min,ymax=lat_max)
data_month=data_region[data_region['Month']==month]
data_month.hist('Year',bins = np.arange(data_month['Year'].min()-0.5,data_month['Year'].max()+1.5 ))
ax=plt.gca()
ax.get_xaxis().get_major_formatter().set_useOffset(False)

March

In [14]:

month=3
fig=comparisons.compare_model_obs(month,2003,'Salinity',data_region,paths['spinup'],zmax=zmax,vmin=24,vmax=33,
                                  xmin=lon_min,xmax=lon_max,ymin=lat_min,ymax=lat_max)
fig=comparisons.compare_model_obs(month,2015,'Salinity',data_region,paths['nowcast'],zmax=zmax, vmin=24,vmax=33,
                                  xmin=lon_min,xmax=lon_max,ymin=lat_min,ymax=lat_max)
data_month=data_region[data_region['Month']==month]
data_month.hist('Year',bins = np.arange(data_month['Year'].min()-0.5,data_month['Year'].max()+1.5 ))
ax=plt.gca()
ax.get_xaxis().get_major_formatter().set_useOffset(False)

April

In [15]:

month=4
fig=comparisons.compare_model_obs(month,2003,'Salinity',data_region,paths['spinup'],zmax=zmax,vmin=24,vmax=33,
                                  xmin=lon_min,xmax=lon_max,ymin=lat_min,ymax=lat_max)
fig=comparisons.compare_model_obs(month,2015,'Salinity',data_region,paths['nowcast'],zmax=zmax, vmin=24,vmax=33,
                                  xmin=lon_min,xmax=lon_max,ymin=lat_min,ymax=lat_max)
data_month=data_region[data_region['Month']==month]
data_month.hist('Year',bins = np.arange(data_month['Year'].min()-0.5,data_month['Year'].max()+1.5 ))
ax=plt.gca()
ax.get_xaxis().get_major_formatter().set_useOffset(False)

Model surface waters are super fresh here. Again, we see the model freshening in the deep. Compared to spinup, the stratified layer is thicker, but the stratification still seems stronger than mode of the observations.

A lot of data. Maybe look at stations separately because observations are getting covered up.

May - no data

June

In [16]:

month=6
fig=comparisons.compare_model_obs(month,2003,'Salinity',data_region,paths['spinup'],zmax=zmax,vmin=24,vmax=33,
                                  xmin=lon_min,xmax=lon_max,ymin=lat_min,ymax=lat_max)
fig=comparisons.compare_model_obs(month,2015,'Salinity',data_region,paths['nowcast'],zmax=zmax, vmin=24,vmax=33,
                                  xmin=lon_min,xmax=lon_max,ymin=lat_min,ymax=lat_max)
data_month=data_region[data_region['Month']==month]
data_month.hist('Year',bins = np.arange(data_month['Year'].min()-0.5,data_month['Year'].max()+1.5 ))
ax=plt.gca()
ax.get_xaxis().get_major_formatter().set_useOffset(False)

Surface layer fresh in observations. Make sense because of the freshet.

Again, tonnes of data and some observations are covered.

July

In [17]:

month=7
fig=comparisons.compare_model_obs(month,2003,'Salinity',data_region,paths['spinup'],zmax=zmax,vmin=24,vmax=33,
                                  xmin=lon_min,xmax=lon_max,ymin=lat_min,ymax=lat_max)
fig=comparisons.compare_model_obs(month,2015,'Salinity',data_region,paths['nowcast'],zmax=zmax, vmin=24,vmax=33,
                                  xmin=lon_min,xmax=lon_max,ymin=lat_min,ymax=lat_max)
data_month=data_region[data_region['Month']==month]
data_month.hist('Year',bins = np.arange(data_month['Year'].min()-0.5,data_month['Year'].max()+1.5 ))
ax=plt.gca()
ax.get_xaxis().get_major_formatter().set_useOffset(False)

Consider comparing locations individually.

In Haro Strait, it appears we have freshening at all depths. Is that most consistent with

Too much mixing

or

Too much fresh water

2014¶

In [18]:

data_2014 = data_region[data_region['Year']==2014]
data_2014.hist('Month')

Out[18]:

array([[<matplotlib.axes._subplots.AxesSubplot object at 0x7fccef723850>]], dtype=object)