In [1]:

import sqlalchemy
from sqlalchemy import (create_engine, Column, String, Integer, Float, MetaData, 
                        Table, type_coerce, ForeignKey, case)
from sqlalchemy.orm import mapper, create_session, relationship, aliased, Session
from sqlalchemy.ext.declarative import declarative_base
from sqlalchemy import case
import numpy as np
from sqlalchemy.ext.automap import automap_base
import matplotlib.pyplot as plt
import sqlalchemy.types as types
from sqlalchemy.sql import and_, or_, not_, func
from sqlalchemy.sql import select
import os
from os.path import isfile
import pandas as pd
import netCDF4 as nc

%matplotlib inline

In [169]:

from IPython.display import HTML

HTML('''<script>
code_show=true; 
function code_toggle() {
 if (code_show){
 $('div.input').hide();
 } else {
 $('div.input').show();
 }
 code_show = !code_show
} 
$( document ).ready(code_toggle);
</script>
<form action="javascript:code_toggle()"><input type="submit" value="Click here to toggle on/off the raw code."></form>''')

Out[169]:

In [2]:

# definitions
basepath='/ocean/eolson/MEOPAR/obs/'
basedir=basepath + 'DFOOPDB/'
dbname='DFO_OcProfDB'
# if db does not exist, exit
if not isfile(basedir + dbname + '.sqlite'):
    print('ERROR: ' + dbname + '.sqlite does not exist')
engine = create_engine('sqlite:///' + basedir + dbname + '.sqlite', echo = False)
Base = automap_base()
# reflect the tables in salish.sqlite:
Base.prepare(engine, reflect=True)
# mapped classes have been created

# existing tables:
StationTBL=Base.classes.StationTBL
ObsTBL=Base.classes.ObsTBL
CalcsTBL=Base.classes.CalcsTBL
JDFLocsTBL=Base.classes.JDFLocsTBL

session = create_session(bind = engine, autocommit = False, autoflush = True)

In [3]:

qry0=(session
      .query(StationTBL.StartTimeZone.label('TimeZone'))
      .filter(StationTBL.StartYear>2014)
      .group_by(StationTBL.StartTimeZone))

In [4]:

for row in qry0.all():
    print(row)
    # one value means time zone is the same for all data for 2015 on

('UTC',)

In [5]:

SA=case([(CalcsTBL.Salinity_Bottle_SA!=None, CalcsTBL.Salinity_Bottle_SA)], else_=
         case([(CalcsTBL.Salinity_T0_C0_SA!=None, CalcsTBL.Salinity_T0_C0_SA)], else_=
         case([(CalcsTBL.Salinity_T1_C1_SA!=None, CalcsTBL.Salinity_T1_C1_SA)], else_=
         case([(CalcsTBL.Salinity_SA!=None, CalcsTBL.Salinity_SA)], else_=
         case([(CalcsTBL.Salinity__Unknown_SA!=None, CalcsTBL.Salinity__Unknown_SA)], 
              else_=CalcsTBL.Salinity__Pre1978_SA)
        ))))

In [6]:

Tem=case([(ObsTBL.Temperature!=None, ObsTBL.Temperature)], else_=
         case([(ObsTBL.Temperature_Primary!=None, ObsTBL.Temperature_Primary)], else_=
         case([(ObsTBL.Temperature_Secondary!=None, ObsTBL.Temperature_Secondary)], else_=ObsTBL.Temperature_Reversing)))
TemUnits=case([(ObsTBL.Temperature!=None, ObsTBL.Temperature_units)], else_=
         case([(ObsTBL.Temperature_Primary!=None, ObsTBL.Temperature_Primary_units)], else_=
         case([(ObsTBL.Temperature_Secondary!=None, ObsTBL.Temperature_Secondary_units)], 
              else_=ObsTBL.Temperature_Reversing_units)))
TemFlag=ObsTBL.Quality_Flag_Temp

In [7]:

qry1=session.query(ObsTBL.Nitrate_plus_Nitrite_units).group_by(ObsTBL.Nitrate_plus_Nitrite_units)
for row in qry1.all():
    print(row)

(None,)
('umol/L',)

In [8]:

qry=session.query(StationTBL.StartYear.label('Year'),StationTBL.StartMonth.label('Month'),
                  StationTBL.StartDay.label('Day'),StationTBL.StartHour.label('Hour'),
                  StationTBL.Lat,StationTBL.Lon,
                 ObsTBL.Pressure,ObsTBL.Depth,ObsTBL.Ammonium,ObsTBL.Ammonium_units,ObsTBL.Chlorophyll_Extracted,
                 ObsTBL.Chlorophyll_Extracted_units,ObsTBL.Nitrate_plus_Nitrite.label('N'),
                  ObsTBL.Silicate.label('Si'),ObsTBL.Silicate_units,SA.label('AbsSal'),Tem.label('T'),TemUnits.label('T_units')).\
            select_from(StationTBL).join(ObsTBL,ObsTBL.StationTBLID==StationTBL.ID).\
            join(CalcsTBL,CalcsTBL.ObsID==ObsTBL.ID).filter(and_(StationTBL.StartYear>2014,
                                                                StationTBL.Lat>47-3/2.5*(StationTBL.Lon+123.5),
                                                                StationTBL.Lat<47-3/2.5*(StationTBL.Lon+121)))

In [9]:

df1=pd.DataFrame(qry.all())
df1.head()

Out[9]:

	Year	Month	Day	Hour	Lat	Lon	Pressure	Depth	Ammonium	Ammonium_units	Chlorophyll_Extracted	Chlorophyll_Extracted_units	N	Si	Silicate_units	AbsSal	T	T_units
0	2015.0	2.0	11.0	11.068611	48.300833	-124.000333	1.9	None	None	None	NaN	mg/m^3	15.31	32.14	umol/L	29.227507	9.7647	'deg_C_(ITS90)'
1	2015.0	2.0	11.0	11.068611	48.300833	-124.000333	6.6	None	None	None	2.57	mg/m^3	17.13	33.90	umol/L	29.484341	9.6880	'deg_C_(ITS90)'
2	2015.0	2.0	11.0	11.068611	48.300833	-124.000333	6.7	None	None	None	NaN	mg/m^3	NaN	NaN	umol/L	29.484839	9.6828	'deg_C_(ITS90)'
3	2015.0	2.0	11.0	11.068611	48.300833	-124.000333	11.0	None	None	None	NaN	mg/m^3	NaN	NaN	umol/L	30.144549	9.3646	'deg_C_(ITS90)'
4	2015.0	2.0	11.0	11.068611	48.300833	-124.000333	11.0	None	None	None	NaN	mg/m^3	20.62	37.65	umol/L	30.157913	9.3586	'deg_C_(ITS90)'

In [10]:

with nc.Dataset('/ocean/eolson/MEOPAR/NEMO-forcing/grid/mesh_mask201702_noLPE.nc') as f:
    tmask=np.copy(f.variables['tmask'])
    lon=np.copy(f.variables['nav_lon'])
    lat=np.copy(f.variables['nav_lat'])
plt.pcolormesh(lon,lat,tmask[0,0,:,:])
plt.plot(df1.Lon,df1.Lat,'k*')
x0=-123.5
y0=47
x1=-126
y1=50
x=np.arange(-127,-123,.5)
y=y0+(y1-y0)/(x1-x0)*(x-x0) # take lat>47-3/2.5*(lon+123.5)
plt.plot(x,y,'m-')
x0=-121
y0=47
x1=-123.5
y1=50
x=np.arange(-124,-120,.5)
y=y0+(y1-y0)/(x1-x0)*(x-x0) # take lat<47-3/2.5*(lon+121)
plt.plot(x,y,'m-')

Out[10]:

[<matplotlib.lines.Line2D at 0x7f62ec4a38d0>]

In [11]:

plt.plot(df1['N'],df1['Si'],'k.')
plt.xlabel('N')
plt.ylabel('Si')

Out[11]:

<matplotlib.text.Text at 0x7f62ec497f28>

In [12]:

from salishsea_tools import gsw_calls, viz_tools, geo_tools, tidetools
import datetime

In [13]:

depths = gsw_calls.generic_gsw_caller('gsw_z_from_p.m', 
                                      [df1.Pressure.values, df1.Lat.values])

In [14]:

depths = depths * -1

In [15]:

df1 = df1.assign(depth = depths)

In [16]:

df1.shape

Out[16]:

(2495, 19)

In [17]:

df1.dropna(subset=['Year', 'Month', 'Hour', 'Lat', 'Lon','Pressure']).shape

Out[17]:

(2495, 19)

In [18]:

df2 = df1.dropna(subset=['Year', 'Month', 'Hour', 'Lat', 'Lon', 'Si', 'N'])

In [19]:

deptht = (nc.Dataset(
'/results/SalishSea/nowcast-green/01jan18/SalishSea_1d_20180101_20180101_dia2_T.nc')
          .variables['deptht'][:])

In [20]:

deptht

Out[20]:

array([  0.5000003,   1.5000031,   2.5000114,   3.5000305,   4.5000706,
         5.5001507,   6.5003104,   7.500623 ,   8.501236 ,   9.502433 ,
        10.5047655,  11.509312 ,  12.518167 ,  13.535412 ,  14.568982 ,
        15.634288 ,  16.761173 ,  18.007135 ,  19.481785 ,  21.389978 ,
        24.100256 ,  28.229916 ,  34.685757 ,  44.517723 ,  58.484333 ,
        76.58559  ,  98.06296  , 121.866516 , 147.08946  , 173.11449  ,
       199.57304  , 226.2603   , 253.06664  , 279.93454  , 306.8342   ,
       333.75018  , 360.67453  , 387.6032   , 414.5341   , 441.4661   ],
      dtype=float32)

In [21]:

grid = nc.Dataset('/data/vdo/MEOPAR/NEMO-forcing/grid/bathymetry_201702.nc')
bathy, X, Y = tidetools.get_bathy_data(grid)
mesh = nc.Dataset('/data/vdo/MEOPAR/NEMO-forcing/grid/mesh_mask201702.nc')
HINDCAST_PATH= '/results/SalishSea/nowcast-green/'

In [22]:

df2.shape

Out[22]:

(2329, 19)

In [23]:

list_of_lons = np.array([])
list_of_lats = np.array([])
list_of_datetimes = np.array([])
list_of_cs_ni = np.array([])
list_of_cs_si = np.array([])
list_of_model_ni = np.array([])
list_of_model_si = np.array([])
list_of_depths = np.array([])
for n in df2.index:
    Yind, Xind = geo_tools.find_closest_model_point(df2.Lon[n], df2.Lat[n], 
                                                    X, Y, land_mask = bathy.mask)
    depth = np.argmin(np.abs(deptht - df2.depth[n]))
    if mesh.variables['tmask'][0,depth,Yind, Xind] == 1:
        date = datetime.datetime(year = int(df2.Year[n]), month = int(df2.Month[n]),
                                day = int(df2.Day[n]), hour = int(df2.Hour[n]),
                                 minute = int((df2.Hour[n] - int(df2.Hour[n]))*60))
        sub_dir = date.strftime('%d%b%y').lower()
        datestr = date.strftime('%Y%m%d')
        fname = 'SalishSea_1h_{}_{}_ptrc_T.nc'.format(datestr, datestr)
        nuts = nc.Dataset(os.path.join(HINDCAST_PATH, sub_dir, fname))
        if date.minute < 30:
            before = datetime.datetime(year = date.year, month = date.month, day = date.day, 
                               hour = (date.hour), minute = 30) - datetime.timedelta(hours=1)
            after = before + datetime.timedelta(hours=1)
            sub_dir2 = after.strftime('%d%b%y').lower()
            datestr2 = after.strftime('%Y%m%d')
            fname2 = 'SalishSea_1h_{}_{}_ptrc_T.nc'.format(datestr2, datestr2)
            nuts2 = nc.Dataset(os.path.join(HINDCAST_PATH, sub_dir2, fname2))
            delta = (date.minute + 30) / 60
            ni_val = ((1-delta)*(nuts.variables['nitrate'][before.hour, depth, Yind, Xind] ) + 
                       (delta)*(nuts2.variables['nitrate'][after.hour, depth, Yind, Xind] ))
            si_val = ((1-delta)*(nuts.variables['silicon'][before.hour, depth, Yind, Xind] ) + 
                       (delta)*(nuts2.variables['silicon'][after.hour, depth, Yind, Xind] ))
        if date.minute >= 30:
            before = datetime.datetime(year = date.year, month = date.month, day = date.day, 
                                   hour = (date.hour), minute = 30)
            after = before + datetime.timedelta(hours=1)
            sub_dir2 = after.strftime('%d%b%y').lower()
            datestr2 = after.strftime('%Y%m%d')
            fname2 = 'SalishSea_1h_{}_{}_ptrc_T.nc'.format(datestr2, datestr2)
            nuts2 = nc.Dataset(os.path.join(HINDCAST_PATH, sub_dir2, fname2))
            delta = (date.minute) / 60
            ni_val = (delta*(nuts.variables['nitrate'][before.hour, depth, Yind, Xind] ) + 
                       (1- delta)*(nuts2.variables['nitrate'][after.hour, depth, Yind, Xind] ))
            si_val = (delta*(nuts.variables['silicon'][before.hour, depth, Yind, Xind] ) + 
                       (1- delta)*(nuts2.variables['silicon'][after.hour, depth, Yind, Xind] ))

        list_of_lons = np.append(list_of_lons, df2.Lon[n])
        list_of_lats = np.append(list_of_lats, df2.Lat[n])
        list_of_datetimes = np.append(list_of_datetimes, date)
        list_of_cs_ni = np.append(list_of_cs_ni, float(df2['N'][n]))
        list_of_cs_si = np.append(list_of_cs_si, float(df2['Si'][n]))
        list_of_model_ni = np.append(list_of_model_ni, ni_val)
        list_of_model_si = np.append(list_of_model_si, si_val)
        list_of_depths = np.append(list_of_depths, depth)

In [24]:

list_of_cs_ni.shape

Out[24]:

(2279,)

In [30]:

print('shallow nutrient samples', list_of_cs_ni[list_of_depths < 15].shape)

shallow nutrient samples (644,)

In [34]:

print('intermediate nutrient samples',  list_of_cs_ni[(22 > list_of_depths) & (list_of_depths >= 15)].shape)

intermediate nutrient samples (355,)

In [36]:

print('deep nutrient samples',  list_of_cs_ni[list_of_depths >= 22].shape)

deep nutrient samples (1280,)

In [37]:

fig, ax = plt.subplots(figsize = (10,10))
viz_tools.set_aspect(ax, coords = 'map')
ax.plot(list_of_lons, list_of_lats, 'ro')
ax.plot(list_of_lons[(list_of_lons < -123.5) & (list_of_lats < 48.6)], 
        list_of_lats[(list_of_lons < -123.5) & (list_of_lats < 48.6)], 
        'bo', label = 'Juan de Fuca')
viz_tools.plot_coastline(ax, grid, coords = 'map')
ax.set_ylim(48, 50.5)
ax.legend()
ax.set_xlim(-125.7, -122.5);

In [55]:

fig, ax = plt.subplots(figsize = (10,10))
viz_tools.set_aspect(ax, coords = 'map')
ax.plot(list_of_lons[(list_of_cs_ni >= 25) & (list_of_cs_ni <= 32) 
                     & (list_of_cs_si >= 58) & (list_of_cs_si <= 93)], 
        list_of_lats[(list_of_cs_ni >= 25) & (list_of_cs_ni <= 32) 
                     & (list_of_cs_si >= 58) & (list_of_cs_si <= 93)], 
        'go')
viz_tools.plot_coastline(ax, grid, coords = 'map')
ax.set_title('Stations with  25 <= Ni <= 32 & 58 <= Si <= 93')
ax.set_ylim(48, 50.5)
ax.set_xlim(-125.7, -122.5);

In [58]:

list_of_cs_ni[(list_of_cs_ni >= 25) & (list_of_cs_ni <= 32) 
                     & (list_of_cs_si >= 58) & (list_of_cs_si <= 93)].shape

Out[58]:

(211,)

All boxed samples have depths > 30 m and are not in Juan de Fuca (211 samples in total)¶

In [57]:

fig, ax = plt.subplots(figsize = (10,10))
ax.plot(list_of_cs_ni[list_of_depths < 15], list_of_model_ni[list_of_depths < 15], 
        '.', color = 'Crimson', alpha = 0.5, label = 'surface')
ax.plot(list_of_cs_ni[(22 > list_of_depths) & (list_of_depths >= 15)], 
        list_of_model_ni[(22 > list_of_depths) & (list_of_depths >= 15)], 
        '.', color = 'DarkOrange', alpha = 0.5, label = 'intermediate')
ax.plot(list_of_cs_ni[list_of_depths >= 22], list_of_model_ni[list_of_depths >= 22], 
        '.', color = 'DeepPink', alpha = 0.5, label = 'deep')
ax.plot(list_of_cs_ni[(list_of_cs_ni >= 25) & (list_of_cs_ni <= 32) 
                      & (list_of_cs_si >= 58) & (list_of_cs_si <= 93)], 
        list_of_model_ni[(list_of_cs_ni >= 25) & (list_of_cs_ni <= 32) 
                         & (list_of_cs_si >= 58) & (list_of_cs_si <= 93)], 
        '.', color = 'Black', alpha = 0.5, label = 'box')

ax.plot(list_of_cs_ni[(list_of_depths < 15) & (list_of_lons < -123.5) & (list_of_lats < 48.6)], 
        list_of_model_ni[(list_of_depths < 15) & (list_of_lons < -123.5) 
                         & (list_of_lats < 48.6)], 
        '.', color = 'ForestGreen', alpha = 0.5, label = ' Juan de Fuca surface')
ax.plot(list_of_cs_ni[(22 > list_of_depths) & (list_of_depths >= 15) 
                      & (list_of_lons < -123.5) & (list_of_lats < 48.6)], 
        list_of_model_ni[(22 > list_of_depths) & (list_of_depths >= 15) 
                         & (list_of_lons < -123.5) & (list_of_lats < 48.6)], 
        '.', color = 'CornflowerBlue', alpha = 0.5, label = ' Juan de Fuca intermediate')
ax.plot(list_of_cs_ni[(list_of_depths >= 22) & (list_of_lons < -123.5) & (list_of_lats < 48.6)], 
        list_of_model_ni[(list_of_depths >= 22) & (list_of_lons < -123.5) 
                         & (list_of_lats < 48.6)], 
        '.', color = 'DarkOrchid', alpha = 0.5, label = ' Juan de Fuca deep')

ax.plot(np.arange(0,35), color = 'grey')
ax.grid('on')
ax.set_title('DFO Nitrate')
ax.set_xlabel('DFO')
ax.set_ylabel('Nowcast-green');
ax.legend()
print('surface bias =  ' + str(-np.mean(list_of_cs_ni[list_of_depths < 15]) 
                               + np.mean(list_of_model_ni[list_of_depths < 15])))
print('surface RMSE = ' + str(np.sqrt(np.sum((list_of_model_ni[list_of_depths < 15] 
                                              - list_of_cs_ni[list_of_depths < 15])**2) /
                              len(list_of_cs_ni[list_of_depths < 15]))))
xbar = np.mean(list_of_cs_ni[list_of_depths < 15])
print('surface Willmott = ' + str(1-(np.sum((list_of_model_ni[list_of_depths < 15] 
                                             - list_of_cs_ni[list_of_depths < 15])**2)  / 
                             np.sum((np.abs(list_of_model_ni[list_of_depths < 15] - xbar) 
                                     + np.abs(list_of_cs_ni[list_of_depths < 15] - xbar))**2))))

print('intermediate bias =  ' + str(-np.mean(list_of_cs_ni[(22 > list_of_depths) 
                                                           & (list_of_depths >= 15)]) 
                               + np.mean(list_of_model_ni[(22 > list_of_depths) 
                                                          & (list_of_depths >= 15)])))
print('intermediate RMSE = ' + str(np.sqrt(np.sum((list_of_model_ni[(22 > list_of_depths) 
                                                                    & (list_of_depths >= 15)] 
                                              - list_of_cs_ni[(22 > list_of_depths) 
                                                              & (list_of_depths >= 15)])**2) /
                              len(list_of_cs_ni[(22 > list_of_depths) & (list_of_depths >= 15)]))))
xbar = np.mean(list_of_cs_ni[(22 > list_of_depths) & (list_of_depths >= 15)])
print('intermediate Willmott = ' + str(1-(np.sum((list_of_model_ni[(22 > list_of_depths) 
                                                                   & (list_of_depths >= 15)] 
                                             - list_of_cs_ni[(22 > list_of_depths) 
                                                             & (list_of_depths >= 15)])**2)  / 
                             np.sum((np.abs(list_of_model_ni[(22 > list_of_depths) 
                                                             & (list_of_depths >= 15)] - xbar) 
                                     + np.abs(list_of_cs_ni[(22 > list_of_depths) 
                                                            & (list_of_depths >= 15)] - xbar))**2))))

print('deep bias =  ' + str(-np.mean(list_of_cs_ni[list_of_depths >= 22]) 
                               + np.mean(list_of_model_ni[list_of_depths >= 22])))
print('deep RMSE = ' + str(np.sqrt(np.sum((list_of_model_ni[list_of_depths >= 22] 
                                              - list_of_cs_ni[list_of_depths >= 22])**2) /
                              len(list_of_cs_ni[list_of_depths >= 22]))))
xbar = np.mean(list_of_cs_ni[list_of_depths >= 22])
print('deep Willmott = ' + str(1-(np.sum((list_of_model_ni[list_of_depths >= 22] 
                                             - list_of_cs_ni[list_of_depths >= 22])**2)  / 
                             np.sum((np.abs(list_of_model_ni[list_of_depths >= 22] - xbar) 
                                     + np.abs(list_of_cs_ni[list_of_depths >= 22] - xbar))**2))))

print('bias =  ' + str(-np.mean(list_of_cs_ni) + np.mean(list_of_model_ni)))
print('RMSE = ' + str(np.sqrt(np.sum((list_of_model_ni - list_of_cs_ni)**2) /
                              len(list_of_cs_ni))))
xbar = np.mean(list_of_cs_ni)
print('Willmott = ' + str(1-(np.sum((list_of_model_ni - list_of_cs_ni)**2)  / 
                             np.sum((np.abs(list_of_model_ni - xbar) 
                                     + np.abs(list_of_cs_ni - xbar))**2))))

surface bias =  -3.538846003674042
surface RMSE = 6.917660684363097
surface Willmott = 0.8309561362318326
intermediate bias =  -4.384472351795072
intermediate RMSE = 6.776651945388753
intermediate Willmott = 0.5530268758659311
deep bias =  -1.1940150120221134
deep RMSE = 3.8010982374846325
deep Willmott = 0.7678296473713924
bias =  -2.35359531664837
RMSE = 5.36571883921976
Willmott = 0.912549833123462