condition, depending on the ``lag`` optional parameter.
By default, the ``lag`` parameter is calculated as the time difference between
the first month of the ENSO indicator reference (or October, if the ENSO indicator
does not have any reference season) and the ``starting_month`` parameter.
For example, if ``starting_month`` is 4 (April) and the reference season runs from
October to December included, the lag will be 6 months, and each water year will be
associated with the ENSO conditions prevailing the previous October.
In both cases, the colors corresponding to each ENSO conditions are available
through the :data:`ENSOcolors` dictionary.
Let us replot the previous FDCs with ENSO information.
First, we need to associate an ENSO indicator to the series.
"""
ONI = enso.load_oni()
series = enso.set_ensoindicator(series, ONI)
"""
We can now create a new figure and a new plot.
"""
fig = cpl.figure()
cpl.plot_fdc(series, enso=True, marker='o')
cpl.plot_fdc(series, enso=True, starting_month=4)
cpl.gca().legend()
fig.savefig('athens_fdc_enso.png')
"""
.. image:: athens_fdc_enso.*
"""
import numpy as np
import scikits.hydroclimpy as hydro
import scikits.hydroclimpy.enso as enso
import scikits.hydroclimpy.io.usgs as usgs
import scikits.hydroclimpy.plotlib as cpl
ONI = enso.load_oni()
series = usgs.load_usgs_flows('02217770').fill_missing_dates()
series = enso.set_ensoindicator(series, ONI)
fig = cpl.figure()
fsp = fig.add_subplot(111)
cpl.plot_fdc(series, plot_enso=True, marker='o', markersize=6, ax=fsp)
cpl.plot_fdc(series, plot_enso=True, starting_month=4, ax=fsp)
fsp.legend()
fsp.set_ylabel("Flows (cfs)", fontweight='bold')
fig.suptitle("Flow duration curve for the North Oconee River at Athens, GA",
fontweight="bold", fontsize=12)
def dump_seasonal_averages(observations, dbname="coaps.sqlite",
period="monthly", ensostr="ONI"):
"""
Dump seasonal averages of observations in a SQLite database.
The database is named following the template ``avg%(observations)s.sdb``.
For each season, there are four tables named following the template
``%(period)s_%(phase)s_%(ENSOindicator)``.
Each table has N+1 columns:
* Station COOP ID (format ``COOP%06i``)
* ENSO indicator (used to define the phase): in ('ONI', 'JMAI', 'MEI')
* ENSO phase : in ('A', 'C', 'N', 'W')
* N values for each season
Parameters
----------
observations: string
String representing the observations to store.
period : ('M', 'NDJ', 'DJF', 'JFM'), optional
String representing the seasons.
Use 'M' for monthly, 'DJF' for 'DJF,MAM,JJA,SON'...
ensostr : ('ONI', 'JMAI', 'MEI'), optional
String representing the ENSO indicator to define ENSO phases
"""
seasons = dict(monthly=['JAN', 'FEB', 'MAR', 'APR', 'MAY', 'JUN',
'JUL', 'AUG', 'SEP', 'OCT', 'NOV', 'DEC', ],
JFM=['JFM', 'AMJ', 'JAS', 'OND'],
DJF=['DJF', 'MAM', 'JJA', 'SON'],
NDJ=['NDJ', 'FMA', 'MJJ', 'ASO'])
# Get the seasonal frequency
period = period.upper()
period_to_freq = dict(M=hydro._c.FR_MTH,
MONTHLY=hydro._c.FR_MTH,
NDJ=hydro._c.FR_QTREOCT,
DJF=hydro._c.FR_QTRENOV,
JFM=hydro._c.FR_QTREDEC,)
try:
freq = period_to_freq[period]
except KeyError:
errmsg = "Invalid period '%s': should be in %s."
raise ValueError(errmsg % (period, period_to_freq.keys()))
# Get the names of the fields
# fieldslist = ["COOPID text primary key", "ENSOI text", "phase text"]
fieldslist = ["id integer primary key autoincrement",
"COOPID text", "ENSOI text", "phase text"]
if period in ("M", "MONTHLY"):
period = "monthly"
fieldslist += ["%s real" % _ for _ in seasons[period]]
nbfields = len(fieldslist)
# Get the conversion function:
if freq == hydro._c.FR_MTH:
conversion_func = None
else:
if observations == 'rain':
conversion_func = ma.sum
else:
conversion_func = ma.mean
# Load the ENSO information
ensostr = ensostr.upper()
if ensostr == 'ONI':
ENSO = enso.load_oni()
elif ensostr == 'JMAI':
ENSO = enso.load_jma()
elif ensostr == 'MEI':
raise NotImplementedError
else:
errmsg = "Invalid ENSO indicator '%s': should be in ('ONI','JMAI','MEI')"
raise ValueError(errmsg % ensostr)
# Get the monthly series from the datatable and set the ENSO indicator
tbname = "series_monthly_%s" % observations
monthly = sql.tsfromsqlite(dbname, tbname, freq="M")
monthly = enso.set_ensoindicator(monthly, ENSO)
# Define dictionaries storing the averages
avg_r = {}
# Loop on the stations
for station in monthly.dtype.names:
# Get the current station
current = monthly[station]
# Get the season
seasonal = current.convert(freq, func=conversion_func)
if (observations == "rain") and (freq != hydro._c.FR_MTH):
mask = hydro.time_series(current.mask, dates=current.dates)
seasonal.mask = mask.convert(freq, func=ma.sum)
# Get the values per phase
coopid = station[-6:]
avg_r[(coopid, ensostr, 'A')] = seasonal.convert("A").mean(0)
avg_r[(coopid, ensostr, 'C')] = seasonal.cold.convert("A").mean(0)
avg_r[(coopid, ensostr, 'N')] = seasonal.neutral.convert("A").mean(0)
avg_r[(coopid, ensostr, 'W')] = seasonal.warm.convert("A").mean(0)
# Get the database name
detect_types = sqlite3.PARSE_DECLTYPES | sqlite3.PARSE_COLNAMES
connection = sqlite3.connect(dbname, detect_types=detect_types)
# Define the creation/insertion lines
tbname = "averages_%(period)s_%(observations)s" % locals()
createstr = "create table %s (%s)" % (tbname, ", ".join(fieldslist))
insertstr_template = "insert into %s values (%s)"
insertstr = insertstr_template % (tbname, ", ".join(['?']*nbfields))
# Create the table
try:
connection.execute(createstr)
except sqlite3.OperationalError:
pass
# Define a generator for the values
generator = ([None, ] + list(k) + list(v.astype(np.object).filled((None,)))
for (k, v) in avg_r.items())
connection.executemany(insertstr, generator)
connection.commit()
connection.close()
mrainfall = rainfall.convert('M',func=ma.sum)
arainfall = mrainfall.convert('A')
assert(arainfall.shape == (64, 12))
monthly_means = arainfall.mean(axis=0).round(1)
print monthly_means
""" [ 115.9 112.4 130.8 94.1 99.4 103.5 122.3 90.2 94.5 77.8
92.7 98.6]"""
ONI = enso.load_oni()
print ONI.dates[[0,-1]]
""" [Jan-1950 Dec-2008]"""
mrainfall = enso.set_ensoindicator(mrainfall, ONI)
print type(mrainfall)
assert(isinstance(mrainfall, enso.ClimateSeries))
print mrainfall.ensoindicator.dates[[0,-1]]
""" [Jan-1944 Dec-2007]"""
mrainfall_2K = mrainfall[(mrainfall.year == 2000)]
print mrainfall_2K.cold
"""[113.792 50.546 86.614 43.18 54.864 50.292 -- -- -- 5.842 106.68 87.884]"""
print mrainfall_2K.neutral
"""[-- -- -- -- -- -- 85.344 93.218 122.174 -- -- --]"""
print mrainfall_2K.warm
"""[-- -- -- -- -- -- -- -- -- -- -- --]"""
mdtype = [('cold', float), ('neutral', float), ('warm', float),
