stormtools_revisions.py

# Copyright 2013-2015 The Salish Sea MEOPAR contributors
# and The University of British Columbia

# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at

#    http://www.apache.org/licenses/LICENSE-2.0

# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

"""A collection of tools for storm surge results from the Salish Sea Model.
"""
from __future__ import division

import datetime
import netCDF4 as NC
import numpy as np
import arrow
import cStringIO
import requests
from dateutil import tz
from xml.etree import cElementTree as ElementTree
import pandas as pd
from salishsea_tools import tidetools
import csv

#msl over 1983-2001 in metres
MSLS = {'PointAtkinson': 3.095, 'Victoria': 1.902, 'PatriciaBay': 2.298, 'CampbellRiver': 2.896}

def convert_date_seconds(times, start):
    """
    This function converts model output time in seconds to datetime objects.
    Note: Doug has a better version of this in nc_tools.timestamp

    :arg times: array of seconds since the start date of a simulation. From time_counter in model output.
    :type times: int

    :arg start: string containing the start date of the simulation in format '01-Nov-2006'
    :type start: str

    :arg diff: string indicating the time step in the times data E.g. months, seconds, days
    :type diff: str

    :returns: array of datetime objects representing the time of model outputs.
    """
    arr_times=[]
    for ii in range(0,len(times)):
        arr_start =arrow.Arrow.strptime(start,'%d-%b-%Y')
        arr_new=arr_start.replace(seconds=times[ii])
        arr_times.append(arr_new.datetime)

    return arr_times

def convert_date_hours(times, start):
    """
    This function converts model output time in hours to datetime objects.

    :arg times: array of hours since the start date of a simulation. From time_counter in model output.
    :type times: int

    :arg start: string containing the start date of the simulation in format '01-Nov-2006'
    :type start: str

    :returns: array of datetime objects representing the time of model outputs.
    """
    arr_times=[]
    for ii in range(0,len(times)):
        arr_start =arrow.Arrow.strptime(start,'%d-%b-%Y')
        arr_new=arr_start.replace(hours=times[ii])
        arr_times.append(arr_new.datetime)

    return arr_times

def get_CGRF_weather(start,end,grid):
    """
    Returns the CGRF weather between the dates start and end at the grid point defined in grid.

    :arg start: string containing the start date of the CGRF collection in format '01-Nov-2006'
    :type start: str

    :arg start: string containing the end date of the CGRF collection in format '01-Nov-2006'
    :type start: str

    :arg grid: array of the CGRF grid coordinates for the point of interest eg. [244,245]
    :arg type: arr of ints

    :returns: windspeed, winddir pressure and time array from CGRF data for the times indicated
    """
    u10=[]; v10=[]; pres=[]; time=[];
    st_ar=arrow.Arrow.strptime(start, '%d-%b-%Y')
    end_ar=arrow.Arrow.strptime(end, '%d-%b-%Y')

   # CGRF_path = '/ocean/dlatorne/MEOPAR/CGRF/NEMO-atmos/'
    CGRF_path = '/ocean/nsoontie/MEOPAR/CGRF/NEMO-atmos/'
    for r in arrow.Arrow.range('day', st_ar, end_ar):
        #u
	m=r.month;
	if m<10:
	  mstr='0' + str(m)
	else:
	  mstr=str(m)
	d=r.day;
	if d<10:
	  dstr='0' + str(d)
	else:
	  dstr=str(d)
        strU='u10_y' + str(r.year) +'m' +mstr + 'd'+ dstr +'.nc'
        fU=NC.Dataset(CGRF_path+strU)
        var=fU.variables['u_wind'][:,grid[0],grid[1]]; u10.extend(var[:])

        #time
        tim=fU.variables['time_counter']; time.extend(tim[:] + (r.day-st_ar.day)*24)
        times =convert_date_hours(time,start)

        #v
        strV='v10_y' + str(r.year) +'m' +mstr + 'd'+ dstr +'.nc'
        fV=NC.Dataset(CGRF_path+strV)
        var=fV.variables['v_wind'][:,grid[0],grid[1]]; v10.extend(var[:])

        #pressure
        strP='slp_corr_y' + str(r.year) +'m' +mstr + 'd'+ dstr +'.nc'
        fP=NC.Dataset(CGRF_path+strP)
        var=fP.variables['atmpres'][:,grid[0],grid[1]]; pres.extend(var[:])

        u10s=np.array(u10); v10s=np.array(v10); press=np.array(pres)
        windspeed=np.sqrt(u10s**2+v10s**2)
	winddir=np.arctan2(v10,u10) * 180 / np.pi
	winddir=winddir + 360 * (winddir<0)
    return windspeed, winddir, press, times

def combine_data(data_list):
    """
    This function combines output from a list of netcdf files into a dict objects of model fields.
    It is used for easy handling of output from thalweg and surge stations.

    :arg data_list: dict object that contains the netcdf handles for the files to be combined. e.g {'Thelweg1': f1, 'Thalweg2': f2,...} where f1 = NC.Dataset('1h_Thalweg1.nc','r')
    :type data_list: dict object

    :returns: dict objects us, vs, lats, lons, sals, tmps, sshs with the zonal velocity, meridional velocity, latitude, longitude, salinity, temperature, and sea surface height for each station. The keys are the same as those in data_list. For example, us['Thalweg1'] contains the zonal velocity from the Thalweg 1 station.

    """

    us={}; vs={}; lats={}; lons={}; sals={}; tmps={}; sshs={};
    for k in data_list:
        net=data_list.get(k)
        us[k]=net.variables['vozocrtx']
        vs[k]=net.variables['vomecrty']
        lats[k]=net.variables['nav_lat']
        lons[k]=net.variables['nav_lon']
        tmps[k]=net.variables['votemper']
        sals[k]=net.variables['vosaline']
        sshs[k]=net.variables['sossheig']
    return us, vs, lats, lons, tmps, sals, sshs

def get_variables(fU,fV,fT,timestamp,depth):
    """
    Generates masked u,v,SSH,S,T from NETCDF handles fU,fV,fT at timestamp and depth.

    :arg fU: netcdf handle for Ugrid model output
    :type fU: netcdf handle

    :arg fV: netcdf handle for Vgrid model output
    :type fV: netcdf handle

    :arg fT: netcdf handle for Tgrid model output
    :type fT: netcdf handle

    :arg timestamp: the timestamp for desired model output
    :type timestamp: int

    :arg depth: the model z-level for desired output
    :type depth: int

    :returns: masked arrays U,V,E,S,T with of zonal velocity,merdional velocity, sea surface height, salinity, and temperature at specified time and z-level.

    """
    # get u and ugrid
    u_vel = fU.variables['vozocrtx']  #u currents and grid
    U = u_vel[timestamp,depth,:,:] #grab data at specified level and time.
    #mask u so that white is plotted on land points
    mu =U== 0
    U= np.ma.array(U,mask=mu)

    #get v and v grid
    v_vel = fV.variables['vomecrty']  #v currents and grid
    V = v_vel[timestamp,depth,:,:] #grab data at specified level and time.

    #mask v so that white is plotted on land points
    mu = V == 0
    V= np.ma.array(V,mask=mu)

    #grid for T points
    eta = fT.variables['sossheig']
    E = eta[timestamp,:,:]
    mu=E==0; E = np.ma.array(E,mask=mu)

    sal = fT.variables['vosaline']
    S= sal[timestamp,depth,:,:]
    mu=S==0; S= np.ma.array(S,mask=mu)

    temp=fT.variables['votemper']
    T = temp[timestamp,depth,:,:]
    mu=T==0; T = np.ma.array(T,mask=mu)

    return U, V, E, S, T

def get_EC_observations(station, start_day, end_day):
    """
    Gather Environment Canada weather observations for the station and dates indicated. The dates should span one month because of how EC data is collected.

    :arg station: string with station name (no spaces). e.g. 'PointAtkinson'
    :type station: str

    :arg start_day: string contating the start date in the format '01-Dec-2006'.
    :type start_day: str

    :arg end_day: string contating the end date in the format '01-Dec-2006'.
    :type end_day: str

    :returns: wind_speed, wind_dir, temperature, times, lat and lon: wind speed and direction, and time (UTC) of data from observations. Also latitude and longitude of the station.

    """
    station_ids = {
	'PamRocks': 6817,
	'SistersIsland': 6813,
	'EntranceIsland': 29411,
        'Sandheads': 6831,
        'YVR':51442, #note, I think YVR station name changed in 2013. Older data use 889
        'YVR_old': 889,
        'PointAtkinson': 844,
        'Victoria': 10944,
        'CampbellRiver': 145,
        'PatriciaBay': 11007, # not exactly at Patricia Bay
	'Esquimalt': 52
    }

    st_ar=arrow.Arrow.strptime(start_day, '%d-%b-%Y')
    end_ar=arrow.Arrow.strptime(end_day, '%d-%b-%Y')
    PST=tz.tzoffset("PST",-28800)

    wind_spd= []; wind_dir=[]; temp=[];
    url = 'http://climate.weather.gc.ca/climateData/bulkdata_e.html'
    query = {
        'timeframe': 1,
        'stationID': station_ids[station],
        'format': 'xml',
        'Year': st_ar.year,
        'Month': st_ar.month,
        'Day': 1,
    }
    response = requests.get(url, params=query)
    tree = ElementTree.parse(cStringIO.StringIO(response.content))
    root = tree.getroot()
    #read lat and lon
    for raw_info in root.findall('stationinformation'):
	lat =float(raw_info.find('latitude').text)
	lon =float(raw_info.find('longitude').text)
    #read data
    raw_data = root.findall('stationdata')
    times = []
    for record in raw_data:
        day = int(record.get('day'))
        hour = int(record.get('hour'))
        year = int(record.get('year'))
        month = int(record.get('month'))
        t = arrow.Arrow(year, month,day,hour,tzinfo=PST)
        selectors = (
            (day == st_ar.day - 1 and hour >= 16)
            or
            (day >= st_ar.day and day < end_ar.day)
            or
            (day == end_ar.day and hour < 16)
        )
        if selectors:
            try:
                wind_spd.append(float(record.find('windspd').text))
                t.to('utc')
                times.append(t.datetime)
            except TypeError:
                wind_spd.append(float('NaN'))
                t.to('utc')
                times.append(t.datetime)
	    try:
                wind_dir.append(float(record.find('winddir').text) * 10)
            except:
                wind_dir.append(float('NaN'))
	    try:
                temp.append(float(record.find('temp').text) +273)
            except:
                temp.append(float('NaN'))
    wind_spd= np.array(wind_spd) * 1000 / 3600
    wind_dir=-np.array(wind_dir)+270
    wind_dir=wind_dir + 360 * (wind_dir<0)
    temp=np.array(temp)
    for i in np.arange(len(times)):
    	times[i] = times[i].astimezone(tz.tzutc())

    return wind_spd, wind_dir, temp, times, lat, lon

def get_SSH_forcing(boundary, date):
    """
    A function that returns the ssh forcing for the month of the date and boundary indicated.

    :arg boundary: A string naming the boundary. e.g 'north' or 'west'
    :type boundary: str

    :arg date: A string indicating the date of interest. e.g. '01-Dec-2006'. The day needs to be the first day of the month.
    :type date: str

    :returns: ssh_forc, time_ssh: arrays of the ssh forcing values and corresponding times

    """
    date_arr = arrow.Arrow.strptime(date, '%d-%b-%Y')
    year = date_arr.year
    month = date_arr.month;    month= "%02d" % (month,)
    if boundary == 'north':
	filen='sshNorth'
    else:
	filen ='ssh'
    ssh_path = '/data/nsoontie/MEOPAR/NEMO-forcing/open_boundaries/' +boundary +'/ssh/' + filen +'_y' + str(year) +'m' +str(month)+ '.nc'
    fS = NC.Dataset(ssh_path);
    ssh_forc=fS.variables['sossheig'];
    tss= fS.variables['time_counter'][:];
    l = tss.shape[0]; t=np.linspace(0,l-1,l) #time array
    time_ssh=convert_date_hours(t,date);

    return ssh_forc, time_ssh

def dateParserMeasured2(s):
    """
    converts string in %d-%b-%Y %H:%M:%S format Pacific time to a datetime object UTC time.
    """
    PST=tz.tzoffset("PST",-28800)
    #convert the string to a datetime object
    unaware = datetime.datetime.strptime(s, "%d-%b-%Y %H:%M:%S ")
    #add in the local time zone (Canada/Pacific)
    aware = unaware.replace(tzinfo=PST)
    #convert to UTC
    return aware.astimezone(tz.tzutc())

def dateParserMeasured(s):
    """
    converts string in %Y/%m/%d %H:%M format Pacific time to a datetime object UTC time.
    """
    PST=tz.tzoffset("PST",-28800)
    #convert the string to a datetime object
    unaware = datetime.datetime.strptime(s, "%Y/%m/%d %H:%M")
    #add in the local time zone (Canada/Pacific)
    aware = unaware.replace(tzinfo=PST)
    #convert to UTC
    return aware.astimezone(tz.tzutc())

def load_tidal_predictions(filename):
    """
    Load tidal prediction from a file.

    :arg filename: A string representing the path of the csv file that contains the tidal predictions. This file should be generated with get_ttide_8.m
    :type filename: string

    :returns: ttide: a dict object that contains tidal predictions and msl the mean component from the harmonic analysis
    """

    #read msl
    line_number = 1
    with open(filename, 'rb') as f:
        mycsv = csv.reader(f); mycsv = list(mycsv)
        msl = mycsv[line_number][1]
        msl=float(msl)

    ttide = pd.read_csv(filename,skiprows=3,parse_dates=[0],date_parser=dateParserMeasured2)
    ttide = ttide.rename(columns={'Time_Local ': 'time', ' pred_8 ': 'pred_8', ' pred_all ': 'pred_all', ' pred_noshallow ': 'pred_noshallow'})

    return ttide, msl

def load_observations(filename):
    """
    Loads tidal observations from the DFO website using tidetools function

    :arg start: a string representing the starting date of the observations.
    :type start: string in format %d-%b-%Y

    :arg end: a string representing the end date of the observations.
    :type end: string in format %d-%b-%Y

    :arg location: a string representing the location for observations
    :type location: a string from the following - PointAtkinson, Victoria, PatriciaBay, CampbellRiver

    :returns: wlev_meas: a dict object with the water level measurements reference to Chart Datum
    """

    #stations = {'PointAtkinson': 7795, 'Victoria': 7120, 'PatriciaBay': 7277, 'CampbellRiver': 8074}
    #statID_PA = stations[location]
    #filename = 'wlev_' +str(statID_PA) + '_' + start +'_' +end +'.csv'
    #tidetools.get_dfo_wlev(statID_PA,start,end)
    wlev_meas = pd.read_csv(filename,skiprows=7,parse_dates=[0],date_parser=dateParserMeasured)
    wlev_meas = wlev_meas.rename(columns={'Obs_date': 'time', 'SLEV(metres)': 'slev'})

    return wlev_meas

def observed_anomaly(ttide,wlev_meas,msl):
    """
    Calculates the observed anomaly (water level obs - tidal predictions).

    :arg ttide: A struc object that contains tidal precitions from get_ttide_8.m
    :type ttide: struc with dimensions time, pred_all, pred_8

    :arg wlev_meas: A struc object with observations from DFO
    :type wlev_meas: sruc with dimensions time, slev

    :arg msl: The mean sea level from tidal predictions
    :type msl: float

    :returns: ssanomaly: the ssh anomaly (wlev_meas.slev-(ttide.pred_all+msl))
    """
    ssanomaly = np.zeros(len(wlev_meas.time))
    for i in np.arange(0,len(wlev_meas.time)):
    #check that there is a corresponding time
    #if any(wlev_pred.time == wlev_meas.time[i]):
        ssanomaly[i] =(wlev_meas.slev[i] - (ttide.pred_all[ttide.time==wlev_meas.time[i]]+msl))
        if not(ssanomaly[i]):
            ssanomaly[i]=float('Nan')

    return ssanomaly

def modelled_anomaly(ssh_m, ssh_tides):
    """
    Calculates the modelled ssh anomaly by finding the difference between a simulation with all forcing and a simulation with tides only.

    :arg ssh_m: An array of modelled ssh
    :type ssh_m: numpy array

    :arg ssh_tides: Array tides only simulation
    :type ssh_tides: numpy array

    :returns: anom: the difference between all_forcing and tidesonly
    """
    anom=ssh_m-ssh_tides
    return anom

def correct_model(ssh,ttide,sdt,edt):
    """
    Adjusts model output by correcting for error in using only 8 constituents

    :arg ssh: an array with model ssh data
    :type ssh: array of numbers

    :arg ttide: struc with tidal predictions
    :type ttide: struc with dimension time, pred_all, pred_8

    :arg sdt: datetime object representing start date of simulation
    :type sdt: datetime object

    :arg edt: datetime object representing end date of simulation
    :type edt: datetime object

    :returns: corr_model: the corrected model output
    """
    #find index of ttide.time at start and end
    inds = ttide.time[ttide.time==sdt].index[0]
    inde = ttide.time[ttide.time==edt].index[0]

    difference = ttide.pred_noshallow-ttide.pred_8
    difference = np.array(difference)
    #average correction over two times to shift to the model 1/2 outputs
    corr = 0.5*(difference[inds:inde] + difference[inds+1:inde+1])

    corr_model = ssh+corr
    return corr_model

def surge_tide(ssh,ttide,sdt,edt):
    """
    Calculates the sea surface height from the model run with surge only. That is, addes tidal prediction to modelled surge.
    :arg ssh: shh from surge only model run
    :type ssh: array of numbers

    :arg ttide: struc with tidal predictions
    :type ttide: struc with dimension time, pred_all, pred_8

    :arg sdt: datetime object representing start date of simulation
    :type sdt: datetime object

    :arg edt: datetime object representing end date of simulation
    :type edt: datetime object

    :returns: surgetide: the surge only run with tides added (mean not inculded)
    """
    #find index of ttide.time at start and end
    inds = ttide.time[ttide.time==sdt].index[0]
    inde = ttide.time[ttide.time==edt].index[0]

    #average correction over two times to shift to the model 1/2 outputs
    tide = np.array(ttide.pred_all)
    tide_corr = 0.5*(tide[inds:inde] + tide[inds+1:inde+1])

    surgetide = ssh+tide_corr
    return surgetide


def get_statistics(obs, model, t_obs, t_model, sdt, edt):
    """
    Calculates several statisitcs, such as mean error, maximum value, etc.
    for model and observations in a given time period.

    :arg obs: observation data
    :type obs: array

    :arg model: model data
    :type model: array

    :arg t_obs: observations time
    :type t_obs: array

    :arg t_model: model time
    :type t_model: array

    :arg sdt: datetime object representing start date of analysis period
    :type sdt: datetime object

    :arg edt: datetime object representing end date of analysis period
    :type edt: datetime object

    :returns: max_obs, max_model, tmax_obs, tmax_model, mean_error,
              mean_abs_error, rms_error, gamma2 (see Bernier Thompson 2006),
              correlation matrix, willmott score, mean_obs, mean_model,
              std_obs, std_model
    """
    # truncate model
    trun_model, trun_tm = truncate(
        model, t_model, sdt.replace(minute=30), edt.replace(minute=30))
    trun_model = trun_model[:-1]
    trun_tm = trun_tm[:-1]
    # truncate observations
    trun_obs, trun_to = truncate(obs, t_obs, sdt, edt)
    # rebase observations
    rbase_obs, rbase_to = rebase_obs(trun_obs, trun_to)
    error = trun_model-rbase_obs
    # calculate statisitcs
    gamma2 = np.var(error)/np.var(rbase_obs)
    mean_error = np.mean(error)
    mean_abs_error = np.mean(np.abs(error))
    rms_error = _rmse(error)
    corr = np.corrcoef(rbase_obs, trun_model)
    max_obs, tmax_obs = _find_max(rbase_obs, rbase_to)
    max_model, tmax_model = _find_max(trun_model, trun_tm)
    mean_obs = np.mean(rbase_obs)
    mean_model = np.mean(trun_model)
    std_obs = np.std(rbase_obs)
    std_model = np.std(trun_model)

    ws = willmott_skill(rbase_obs, trun_model)

    return (
        max_obs, max_model, tmax_obs, tmax_model, mean_error, mean_abs_error,
        rms_error, gamma2, corr, ws, mean_obs, mean_model, std_obs, std_model,
    )


def truncate(data,time,sdt,edt):
    """
    Returns truncated array for the time period of interest
    :arg data: data to be truncated
    :type data: array

    :arg time: time output associated with data
    :type time: array

    :arg sdt: datetime object representing start date of analysis period
    :type sdt: datetime object

    :arg edt: datetime object representing end date of analysis period
    :type edt: datetime object

    :returns: data_t, time_t, truncated data and time arrays
    """
    inds = np.where(time==sdt)[0]
    inde = np.where(time==edt)[0]

    data_t=np.array(data[inds:inde+1])
    time_t = np.array(time[inds:inde+1])

    return data_t, time_t

def rebase_obs(data,time):
    """
    Rebases the observations so that they are given on the half hour instead of hour.
    Half hour outputs caclulated by averaging between two hourly outputs.
    :arg data: data to be rebased
    :type data: array

    :arg time: time outputs associated with data
    :type time: array

    :returns: rebase_data, rebase_time, the data and times shifted by half an hour
    """
    rebase_data = 0.5*(data[1:]+data[:-1])
    rebase_time=[]
    for k in range(time.shape[0]):
        rebase_time.append(time[k].replace(minute=30))
    rebase_time=np.array(rebase_time)
    rebase_time=rebase_time[0:-1]
    return rebase_data, rebase_time

def _rmse(diff):
    return np.sqrt(np.mean(diff**2))

def _find_max(data,time):
    max_data = np.nanmax(data)
    time_max =time[np.nanargmax(data)]

    return max_data, time_max

def willmott_skill(obs,model):
    """Caclulates the Willmott skill score of the model. See Willmott 1982.
    :arg obs: observations data
    :type obs: array

    :arg model: model data
    :type model: array

    :returns: ws, the Willmott skill score
    """
    obar = np.nanmean(obs)
    mprime = model -obar
    oprime = obs -obar

    diff_sq = np.sum((model-obs)**2)
    add_sq = np.sum((np.abs(mprime) +np.abs(oprime))**2)

    ws = 1-diff_sq/add_sq
    return ws

def get_NOAA_wlev(station_no, start_date, end_date):
    """Download water level data from NOAA site for one NOAA station
    for specified period.

    :arg station_no: Station number e.g. 9443090.
    :type station_no: int

    :arg start_date: Start date; e.g. '01-JAN-2010'.
    :type start_date: str

    :arg end_date: End date; e.g. '31-JAN-2010'
    :type end_date: str

    :returns: Saves text file with water level data in meters at one station. Time zone is UTC
    """
    # Name the output file
    outfile = 'wlev_'+str(station_no)+'_'+str(start_date)+'_'+str(end_date)+'.csv'
    # Form urls and html information

    st_ar=arrow.Arrow.strptime(start_date, '%d-%b-%Y')
    end_ar=arrow.Arrow.strptime(end_date, '%d-%b-%Y')

    base_url = 'http://tidesandcurrents.noaa.gov'
    form_handler = (
        '/stationhome.html?id='
        + str(station_no))
    data_provider = (
        '/api/datagetter?product=hourly_height&application=NOS.COOPS.TAC.WL'
        + '&begin_date=' +st_ar.format('YYYYMMDD') +'&end_date='+end_ar.format('YYYYMMDD')
        + '&datum=MLLW&station='+str(station_no)
        + '&time_zone=GMT&units=metric&interval=h&format=csv')
    # Go get the data from the DFO site
    with requests.Session() as s:
        s.post(base_url)
        r = s.get(base_url + data_provider)
    # Write the data to a text file
    with open(outfile, 'w') as f:
        f.write(r.text)

def get_NOAA_predictions(station_no, start_date, end_date):
    """Download tide predictions from NOAA site for one NOAA station
    for specified period.

    :arg station_no: Station number e.g. 9443090.
    :type station_no: int

    :arg start_date: Start date; e.g. '01-JAN-2010'.
    :type start_date: str

    :arg end_date: End date; e.g. '31-JAN-2010'
    :type end_date: str

    :returns: Saves text file with predictions in meters at one station. Time zone is UTC
    """
    # Name the output file
    outfile = 'predictions_'+str(station_no)+'_'+str(start_date)+'_'+str(end_date)+'.csv'
    # Form urls and html information

    st_ar=arrow.Arrow.strptime(start_date, '%d-%b-%Y')
    end_ar=arrow.Arrow.strptime(end_date, '%d-%b-%Y')

    base_url = 'http://tidesandcurrents.noaa.gov'
    form_handler = (
        '/stationhome.html?id='
        + str(station_no))
    data_provider = (
        '/api/datagetter?product=predictions&application=NOS.COOPS.TAC.WL'
        + '&begin_date=' +st_ar.format('YYYYMMDD') +'&end_date='+end_ar.format('YYYYMMDD')
        + '&datum=MLLW&station='+str(station_no)
        + '&time_zone=GMT&units=metric&interval=h&format=csv')
    # Go get the data from the DFO site
    with requests.Session() as s:
        s.post(base_url)
        r = s.get(base_url + data_provider)
    # Write the data to a text file
    with open(outfile, 'w') as f:
        f.write(r.text)

def get_operational_weather(start,end,grid):
    """
    Returns the CGRF weather between the dates start and end at the grid point defined in grid.

    :arg start: string containing the start date of the CGRF collection in format '01-Nov-2006'
    :type start: str

    :arg start: string containing the end date of the CGRF collection in format '01-Nov-2006'
    :type start: str

    :arg grid: array of the CGRF grid coordinates for the point of interest eg. [244,245]
    :arg type: arr of ints

    :returns: windspeed, winddir pressure and time array from CGRF data for the times indicated
    """
    u10=[]; v10=[]; pres=[]; time=[];
    st_ar=arrow.Arrow.strptime(start, '%d-%b-%Y')
    end_ar=arrow.Arrow.strptime(end, '%d-%b-%Y')

    ops_path = '/ocean/sallen/allen/research/Meopar/Operational/'
    opsp_path = '/ocean/nsoontie/MEOPAR/GEM2.5/ops/'

    for r in arrow.Arrow.range('day', st_ar, end_ar):
        #u
	m=r.month;
	if m<10:
	  mstr='0' + str(m)
	else:
	  mstr=str(m)
	d=r.day;
	if d<10:
	  dstr='0' + str(d)
	else:
	  dstr=str(d)
        fstr='ops_y' + str(r.year) +'m' +mstr + 'd'+ dstr +'.nc'
        f=NC.Dataset(ops_path+fstr)
	#u        
	var=f.variables['u_wind'][:,grid[0],grid[1]]; u10.extend(var[:])
	#v
	var=f.variables['v_wind'][:,grid[0],grid[1]]; v10.extend(var[:])    
	#pressure    
	fpstr = 'slp_corr_ops_y' + str(r.year) +'m' +mstr + 'd'+ dstr +'.nc'
        fP=NC.Dataset(opsp_path+fpstr)
        var=fP.variables['atmpres'][:,grid[0],grid[1]]; pres.extend(var[:])
        #time
        tim=f.variables['time_counter']; time.extend(tim[:])
        times =convert_date_seconds(time,'01-Jan-1970')

        u10s=np.array(u10); v10s=np.array(v10); press=np.array(pres)
        windspeed=np.sqrt(u10s**2+v10s**2)
	winddir=np.arctan2(v10,u10) * 180 / np.pi
	winddir=winddir + 360 * (winddir<0)
    return windspeed, winddir, press, times