#!/usr/bin/env python
# coding: utf-8

# Looking at high-frequency oscillation in nowcast green ssh at Point Atkinson 15 minute files.

# In[1]:


import netCDF4 as nc
import matplotlib.pyplot as plt
import numpy as np
import os
import datetime

from salishsea_tools import nc_tools

get_ipython().run_line_magic('matplotlib', 'inline')


# ## A sample nowcast-green output

# In[2]:


f = nc.Dataset('/results/SalishSea/nowcast-green/08mar16/PointAtkinson.nc')

ssh = f.variables['sossheig'][:,0,0]
ts =nc_tools.timestamp(f,np.arange(ssh.shape[0]))
dates = [d.datetime for d in ts]


# In[3]:


plt.plot(dates,ssh)
plt.title(dates[0].strftime('%Y-%m-%d'))
plt.ylabel('ssh [m]')


# In[4]:


ssh[-10:]


# * Saw-tooth like behaviour and repeated values.
# * Could this be because 15 minutes is not divisible by out 40 second time step?

# ### Other NEMO 3.6 output (with 40s timestep)
# Tested
# * 15 minute output
# * 10 minute output
# * 20 minute output

# In[5]:


rundir = '/ocean/nsoontie/MEOPAR/SalishSea/results/mixing_paper/holl_jul/'
times = [10,15,20]
fig,ax = plt.subplots(1,1,figsize=(15,5))
for time in times:
    f = nc.Dataset(os.path.join(rundir,'PointAtkinson_{}.nc'.format(time)))
    ssh = f.variables['sossheig'][:,0,0]
    ts =nc_tools.timestamp(f,np.arange(ssh.shape[0]))
    dates = [d.datetime for d in ts]
    ax.plot(dates, ssh,label ='{} minute output'.format(time) )
    ax.legend(loc=0)
    ax.set_ylabel('ssh [m]')
    print('{} minute output'.format(time))
    print('last 8 ssh: ', ssh[-8:])
ax.set_xlim([datetime.datetime(2015,7,11),datetime.datetime(2015,7,12)])


# * This saw-tooth behaviour is not present in the 10 minute and 20 minute outputs. So, I think we should switch to 10 minute output in nowcast-green. The behaviors is an artifact of 
# * Unfortunately, this run went unstable in Boundary Pass at 15 m depth after ~15000 time steps. Using 
#    * 40s timestep
#    * kappa/mu=10
#    * background visc= 1e-5
#    * background diff=1e-6 
#    * hollingsworth correction
#    * rn_bfri2 = 2e-3
#    * ln_loglayer=T
#    * rn_bfri2_max = 1.e-1
#    * rn_bfrz0=0.0015
#    * bathy2
#    * TS4 tides
# 
# It was in the middle of a spring tides. A run without hollingsworth correction was stable.
# 

# In[ ]: