def doAREMCAnalyssis(stnr, startDate, endDate):
"""
Preforms a Ragnar Ekker variant of Monte Carlo (or whatever). This metod runs an analysis of several conbinations
of gammadistributions for precipitation and temperature affect on cloudcover. Results are written to a sqllite
database for further study.
:param stnr: eKlima station where there is cloudcover for testing method.
:param startDate: simulations and referance data from this data
:param endDate: simulations and referance data to this data
:return:
"""
wsTemp = getMetData(stnr, 'TAM', startDate, endDate, 0, 'list')
wsPrec = getMetData(stnr, 'RR', startDate, endDate, 0, 'list')
wsCC = getMetData(stnr, 'NNM', startDate, endDate, 0, 'list')
temp, date = strip_metadata(wsTemp, get_dates=True)
prec = strip_metadata(wsPrec, False)
clouds = strip_metadata(wsCC, False)
loggdate = datetime.datetime.now()
period = '{0} to {1}'.format(startDate,endDate)
cs = [1] # cloudshift the observed data one day forward
psh = [2.8, 3.0, 3.3] # prec gamma shapefactor
psc = [2.0] # prec gamma scalefactor
pdb = [0.2, 0.3, 0.5, 0.7] # prec days back
pap = [0.3, 0.4, 0.5] # prec amplitude factor
tsh = [4.3, 4.5, 4.8, 5.0] # temp gamma shapefactor
tsc = [1.0] # temp gamma scalefactor
tdb = [4.8, 5.0, 5.5] # temp days back
tap = [0.02, 0.03, 0.05] # temp amplitude factor
estimateSimulations = len(cs)*len(psh)*len(psc)*len(pdb)*len(pap)*len(tsh)*len(tsc)*len(tdb)*len(tap)
doneSimulations = 0
for a in cs:
cloudsShifted = __shiftClouds(clouds, a)
for b in psh:
for c in psc:
for d in pdb:
for e in pap:
for f in tsh:
for g in tsc:
for h in tdb:
for i in tap:
estClouds = dgc.cc_gamma_prec_and_temp_change(prec, temp, [b, c, d, e], [f, g, h, i])
# What is the root mean square of estimatet vs observed clouds?
rms = np.sqrt(((np.array(estClouds) - np.array(cloudsShifted)) ** 2).mean())
__writeRMC2database(database_location, a, b, c, d, e, f, g, h, i, rms, loggdate, stnr, period)
doneSimulations = doneSimulations + 1
print('beregning {0} av {1}'.format(doneSimulations, estimateSimulations))
return
def doAREMCAnalyssis(stnr, startDate, endDate):
"""
Preforms a Ragnar Ekker variant of Monte Carlo (or whatever). This metod runs an analysis of several conbinations
of gammadistributions for precipitation and temperature affect on cloudcover. Results are written to a sqllite
database for further study.
:param stnr: eKlima station where there is cloudcover for testing method.
:param startDate: simulations and referance data from this data
:param endDate: simulations and referance data to this data
:return:
"""
wsTemp = getMetData(stnr, 'TAM', startDate, endDate, 0, 'list')
wsPrec = getMetData(stnr, 'RR', startDate, endDate, 0, 'list')
wsCC = getMetData(stnr, 'NNM', startDate, endDate, 0, 'list')
temp, date = strip_metadata(wsTemp, get_dates=True)
prec = strip_metadata(wsPrec, False)
clouds = strip_metadata(wsCC, False)
loggdate = datetime.datetime.now()
period = '{0} to {1}'.format(startDate,endDate)
cs = [1] # cloudshift the observed data one day forward
psh = [2.8, 3.0, 3.3] # prec gamma shapefactor
psc = [2.0] # prec gamma scalefactor
pdb = [0.2, 0.3, 0.5, 0.7] # prec days back
pap = [0.3, 0.4, 0.5] # prec amplitude factor
tsh = [4.3, 4.5, 4.8, 5.0] # temp gamma shapefactor
tsc = [1.0] # temp gamma scalefactor
tdb = [4.8, 5.0, 5.5] # temp days back
tap = [0.02, 0.03, 0.05] # temp amplitude factor
estimateSimulations = len(cs)*len(psh)*len(psc)*len(pdb)*len(pap)*len(tsh)*len(tsc)*len(tdb)*len(tap)
doneSimulations = 0
for a in cs:
cloudsShifted = __shiftClouds(clouds, a)
for b in psh:
for c in psc:
for d in pdb:
for e in pap:
for f in tsh:
for g in tsc:
for h in tdb:
for i in tap:
estClouds = dgc.cc_gamma_prec_and_temp_change(prec, temp, [b, c, d, e], [f, g, h, i])
# What is the root mean square of estimatet vs observed clouds?
rms = np.sqrt(((np.array(estClouds) - np.array(cloudsShifted)) ** 2).mean())
__writeRMC2database(database_location, a, b, c, d, e, f, g, h, i, rms, loggdate, stnr, period)
doneSimulations = doneSimulations + 1
print('beregning {0} av {1}'.format(doneSimulations, estimateSimulations))
return
def testCloudMaker(stnr, startDate, endDate, method):
"""Gets data from a eKlima station and smooths the data depending on which method we wish to test.
We find the nash-sutcliffe to observations of clouds. And we plot and save the result to Plots folder.
:param stnr: eKlima station where there is cloud cover for testing method.
:param startDate: simulations and reference data from this data
:param endDate: simulations and reference data to this data
:param method: specify which method to be tested
:return:
Available methods to test:
cc_from_prec
ccFromAveragePrec
ccFromAverageObsCc
ccGammaPrec
ccGammaTemp
ccGammaPrecAndTemp"""
wsTemp = getMetData(stnr, 'TAM', startDate, endDate, 0, 'list')
wsPrec = getMetData(stnr, 'RR', startDate, endDate, 0, 'list')
wsCC = getMetData(stnr, 'NNM', startDate, endDate, 0, 'list')
temp, date = strip_metadata(wsTemp, get_dates=True)
prec = strip_metadata(wsPrec)
clouds = strip_metadata(wsCC)
dayShift = 1
clouds = __shiftClouds(clouds, dayShift)
if method == 'ccFromRandomThomas':
estClouds = dpz.clouds_from_precipitation(prec, method='Random Thomas')
gammaFigtext = 'No gamma smoothing and dayshift = {0}'.format(dayShift)
elif method == 'ccFromPrec':
estClouds = dpz.clouds_from_precipitation(prec, method='Binary')
gammaFigtext = 'No gamma smoothing and dayshift = {0}'.format(dayShift)
elif method == 'ccFromAveragePrec':
estClouds = dpz.clouds_from_precipitation(prec, method='Average')
gammaFigtext = 'No gamma smoothing and dayshift = {0}'.format(dayShift)
elif method == 'ccFromPrecAndAveragePrec':
estClouds = dpz.clouds_from_precipitation(prec, method='Binary and average')
gammaFigtext = ''.format()
elif method == 'ccFromAverageObsCc':
estClouds = [sum(clouds)/float(len(clouds))] * len(clouds)
gammaFigtext = ''.format()
elif method == 'ccGammaPrec':
gammaPrec = [2.8, 2., 0.3, 0.4]
estClouds = dpz.clouds_from_precipitation(prec, method='Binary')
estClouds = dgc.cc_gamma_smoothing(estClouds, gammaPrec)
gammaFigtext = "gamma smoothing prec = {0} and dayshift = {1}".format(gammaPrec, dayShift)
elif method == 'ccGammaTempChange':
gammaTemp = [4.8, 1., 5., 0.03]
estClouds = dgc.cc_gamma_temp(temp, gammaTemp)
gammaFigtext = "gamma smoothing temp = {0} and dayshift = {1}".format(gammaTemp, dayShift)
elif method == 'ccGammaPrecAndTempChange':
gammaPrec = [2.8, 2., 0.3, 0.4]
gammaTemp = [4.8, 1., 5., 0.03]
estClouds = dgc.cc_gamma_prec_and_temp_change(prec, temp, gammaPrec, gammaTemp)
gammaFigtext = "prec = {0} and temp = {1} and dayshift = {2}".format(gammaPrec, gammaTemp, dayShift)
fileName = "{3} {0} {1} {2}.png".format(stnr, startDate[0:7], endDate[0:7], method)
# What is the root mean square of estimatet vs observed clouds?
# rms = np.sqrt(((np.array(estClouds) - np.array(clouds)) ** 2).mean())
# Nash–Sutcliffe model efficiency coefficient from
# https://en.wikipedia.org/wiki/Nash%E2%80%93Sutcliffe_model_efficiency_coefficient
numerator = 0
denominator = 0
mean_clouds = sum(clouds)/float(len(clouds))
for i in range(0, len(clouds), 1):
numerator += (clouds[i] - estClouds[i])**2
denominator += (clouds[i] - mean_clouds)**2
nash_sutcliffe = 1 - numerator/denominator
# Figure dimensions
fsize = (16, 10)
plt.figure(figsize=fsize)
plt.clf()
# plot total snowdepth on land
plt.bar(date, prec, width=1, color="0.4")
# plot the estimated cloud cover
for i in range(0, len(estClouds) - 1, 1):
if estClouds[i] > 0:
plt.hlines(max(prec) * 1.2, date[i], date[i + 1], lw=45, color=str(-(estClouds[i] - 1.)))
elif estClouds[i] == None:
plt.hlines(max(prec) * 1.2, date[i], date[i + 1], lw=45, color="pink")
else:
plt.hlines(max(prec) * 1.2, date[i], date[i + 1], lw=45, color=str(-(estClouds[i] - 1.)))
# plot cloud cover from met
for i in range(0, len(clouds) - 1, 1):
if clouds[i] > 0:
plt.hlines(max(prec) * 1.1, date[i], date[i + 1], lw=45, color=str(-(clouds[i] - 1.)))
elif clouds[i] == None:
plt.hlines(max(prec) * 1.1, date[i], date[i + 1], lw=45, color="pink")
else:
plt.hlines(max(prec) * 1.1, date[i], date[i + 1], lw=45, color=str(-(clouds[i] - 1.)))
# this plots temperature on separate right side axis
plt.twinx()
temp_pluss = []
temp_minus = []
for i in range(0, len(temp), 1):
if temp[i] >= 0:
temp_pluss.append(temp[i])
temp_minus.append(np.nan)
else:
temp_minus.append(temp[i])
temp_pluss.append(np.nan)
plt.plot(date, temp, "black")
plt.plot(date, temp_pluss, "red")
plt.plot(date, temp_minus, "blue")
# title and text fields
plt.title("{3} {0} {1} {2}".format(stnr, startDate[0:7], endDate[0:7], method))
plt.text(date[len(date)/2], min(temp)*1.2, 'gamma smoothing [shape, scale, days back, amplification]')
plt.text(date[len(date)/2], min(temp)*1.3, gammaFigtext)
# this is a scatter plot of modelled and estimated cloud cover
xfrac = 0.15
yfrac = (float(fsize[0])/float(fsize[1])) * xfrac
xpos = 0.95-xfrac
ypos = 0.42-yfrac
a = plt.axes([xpos, ypos, xfrac, yfrac])
a.scatter(clouds, estClouds)
plt.setp(a, xticks=[0, 0.5, 1], yticks=[0, 0.5, 1])
plt.text(0.0, 0.1, 'na_su = {0}'.format(round(nash_sutcliffe, 2)), color='yellow', bbox={'facecolor':'black'})
plt.savefig("{0}{1}".format(plot_folder, fileName))
return nash_sutcliffe
def testCloudMaker(stnr, startDate, endDate, method):
"""Gets data from a eKlima station and smooths the data depending on which method we wish to test.
We find the nash-sutcliffe to observations of clouds. And we plot and save the result to Plots folder.
:param stnr: eKlima station where there is cloud cover for testing method.
:param startDate: simulations and reference data from this data
:param endDate: simulations and reference data to this data
:param method: specify which method to be tested
:return:
Available methods to test:
cc_from_prec
ccFromAveragePrec
ccFromAverageObsCc
ccGammaPrec
ccGammaTemp
ccGammaPrecAndTemp"""
wsTemp = getMetData(stnr, 'TAM', startDate, endDate, 0, 'list')
wsPrec = getMetData(stnr, 'RR', startDate, endDate, 0, 'list')
wsCC = getMetData(stnr, 'NNM', startDate, endDate, 0, 'list')
temp, date = strip_metadata(wsTemp, get_dates=True)
prec = strip_metadata(wsPrec)
clouds = strip_metadata(wsCC)
dayShift = 1
clouds = __shiftClouds(clouds, dayShift)
if method == 'ccFromRandomThomas':
estClouds = dpz.clouds_from_precipitation(prec, method='Random Thomas')
gammaFigtext = 'No gamma smoothing and dayshift = {0}'.format(dayShift)
elif method == 'ccFromPrec':
estClouds = dpz.clouds_from_precipitation(prec, method='Binary')
gammaFigtext = 'No gamma smoothing and dayshift = {0}'.format(dayShift)
elif method == 'ccFromAveragePrec':
estClouds = dpz.clouds_from_precipitation(prec, method='Average')
gammaFigtext = 'No gamma smoothing and dayshift = {0}'.format(dayShift)
elif method == 'ccFromPrecAndAveragePrec':
estClouds = dpz.clouds_from_precipitation(prec, method='Binary and average')
gammaFigtext = ''.format()
elif method == 'ccFromAverageObsCc':
estClouds = [sum(clouds)/float(len(clouds))] * len(clouds)
gammaFigtext = ''.format()
elif method == 'ccGammaPrec':
gammaPrec = [2.8, 2., 0.3, 0.4]
estClouds = dpz.clouds_from_precipitation(prec, method='Binary')
estClouds = dgc.cc_gamma_smoothing(estClouds, gammaPrec)
gammaFigtext = "gamma smoothing prec = {0} and dayshift = {1}".format(gammaPrec, dayShift)
elif method == 'ccGammaTempChange':
gammaTemp = [4.8, 1., 5., 0.03]
estClouds = dgc.cc_gamma_temp(temp, gammaTemp)
gammaFigtext = "gamma smoothing temp = {0} and dayshift = {1}".format(gammaTemp, dayShift)
elif method == 'ccGammaPrecAndTempChange':
gammaPrec = [2.8, 2., 0.3, 0.4]
gammaTemp = [4.8, 1., 5., 0.03]
estClouds = dgc.cc_gamma_prec_and_temp_change(prec, temp, gammaPrec, gammaTemp)
gammaFigtext = "prec = {0} and temp = {1} and dayshift = {2}".format(gammaPrec, gammaTemp, dayShift)
fileName = "{3} {0} {1} {2}.png".format(stnr, startDate[0:7], endDate[0:7], method)
# What is the root mean square of estimatet vs observed clouds?
# rms = np.sqrt(((np.array(estClouds) - np.array(clouds)) ** 2).mean())
# Nash–Sutcliffe model efficiency coefficient from
# https://en.wikipedia.org/wiki/Nash%E2%80%93Sutcliffe_model_efficiency_coefficient
numerator = 0
denominator = 0
mean_clouds = sum(clouds)/float(len(clouds))
for i in range(0, len(clouds), 1):
numerator += (clouds[i] - estClouds[i])**2
denominator += (clouds[i] - mean_clouds)**2
nash_sutcliffe = 1 - numerator/denominator
# Figure dimensions
fsize = (16, 10)
plt.figure(figsize=fsize)
plt.clf()
# plot total snowdepth on land
plt.bar(date, prec, width=1, color="0.4")
# plot the estimated cloud cover
for i in range(0, len(estClouds) - 1, 1):
if estClouds[i] > 0:
plt.hlines(max(prec) * 1.2, date[i], date[i + 1], lw=45, color=str(-(estClouds[i] - 1.)))
elif estClouds[i] == None:
plt.hlines(max(prec) * 1.2, date[i], date[i + 1], lw=45, color="pink")
else:
plt.hlines(max(prec) * 1.2, date[i], date[i + 1], lw=45, color=str(-(estClouds[i] - 1.)))
# plot cloud cover from met
for i in range(0, len(clouds) - 1, 1):
if clouds[i] > 0:
plt.hlines(max(prec) * 1.1, date[i], date[i + 1], lw=45, color=str(-(clouds[i] - 1.)))
elif clouds[i] == None:
plt.hlines(max(prec) * 1.1, date[i], date[i + 1], lw=45, color="pink")
else:
plt.hlines(max(prec) * 1.1, date[i], date[i + 1], lw=45, color=str(-(clouds[i] - 1.)))
# this plots temperature on separate right side axis
plt.twinx()
temp_pluss = []
temp_minus = []
for i in range(0, len(temp), 1):
if temp[i] >= 0:
temp_pluss.append(temp[i])
temp_minus.append(np.nan)
else:
temp_minus.append(temp[i])
temp_pluss.append(np.nan)
plt.plot(date, temp, "black")
plt.plot(date, temp_pluss, "red")
plt.plot(date, temp_minus, "blue")
# title and text fields
plt.title("{3} {0} {1} {2}".format(stnr, startDate[0:7], endDate[0:7], method))
plt.text(date[len(date)/2], min(temp)*1.2, 'gamma smoothing [shape, scale, days back, amplification]')
plt.text(date[len(date)/2], min(temp)*1.3, gammaFigtext)
# this is a scatter plot of modelled and estimated cloud cover
xfrac = 0.15
yfrac = (float(fsize[0])/float(fsize[1])) * xfrac
xpos = 0.95-xfrac
ypos = 0.42-yfrac
a = plt.axes([xpos, ypos, xfrac, yfrac])
a.scatter(clouds, estClouds)
plt.setp(a, xticks=[0, 0.5, 1], yticks=[0, 0.5, 1])
plt.text(0.0, 0.1, 'na_su = {0}'.format(round(nash_sutcliffe, 2)), color='yellow', bbox={'facecolor':'black'})
plt.savefig("{0}{1}".format(plot_folder, fileName))
return nash_sutcliffe
