def zerodown_data(datpath, sr):
# lists of relevant hourly data
hmean = []
hmax = []
hhts = []
hht10 = []
# loop through all our data files and compute relevant hourly data
"""
The weakness to this approach is that individual waves spanning files are not identified (they could but it would be a hassle). It also means we need to handle border cases (zero-dimension matrices) in our zerodown function.
"""
for file in sorted(os.listdir(datpath)):
# read raw data from file
with open(os.path.join(datpath, file), 'r') as f:
raw = f.read().split('\n')
# remove first line (header) and any empty lines
raw = [r for r in raw[1:] if r]
# retrieve elevation data and convert into meters
data = [r.split(',')[1].replace('-\.','-0\.') for r in raw]
data = np.asarray(data, float)*0.01
# zerodowncross
mean, max, hts, ht10 = wave_util.zerodown(data, sr)
# add to our hourly list
print file, hts
hmean.append(mean)
hmax.append(max)
hhts.append(hts)
hht10.append(ht10)
return (hmean, hmax, hhts, hht10)
def zerodown_data(datpath, sr):
# lists of relevant hourly data
hmean = []
hmax = []
hhts = []
hht10 = []
# loop through all our data files and compute relevant hourly data
"""
The weakness to this approach is that individual waves spanning files are not identified (they could but it would be a hassle). It also means we need to handle border cases (zero-dimension matrices) in our zerodown function.
"""
for file in sorted(os.listdir(datpath)):
# read raw data from file
with open(os.path.join(datpath, file), 'r') as f:
raw = f.read().split('\n')
# remove first line (header) and any empty lines
raw = [r for r in raw[1:] if r]
# retrieve elevation data and convert into meters
data = [r.split(',')[1].replace('-\.', '-0\.') for r in raw]
data = np.asarray(data, float) * 0.01
# zerodowncross
mean, max, hts, ht10 = wave_util.zerodown(data, sr)
# add to our hourly list
print file, hts
hmean.append(mean)
hmax.append(max)
hhts.append(hts)
hht10.append(ht10)
return (hmean, hmax, hhts, hht10)
def synthetic(hsfile, spectrum, sr):
curpath = os.path.dirname(os.path.abspath(__file__))
with open(os.path.join(curpath,hsfile),'r') as f:
raw = f.read().split('\n')
data = [[float(r.split(',')[0]),float(r.split(',')[1])] for r in raw if r]
"""
constants for spectrum and sea water level generation. we define them outside the loop to save some time
"""
max_t = 3600 # maximum duration of time series
nyq_f = sr/2 # nyquist frequency of spectra: highest frequency bin
del_f = 1.0/max_t # interval between frequency bins
f = np.linspace(del_f,nyq_f,max_t) # array containing all frequency bins
t = np.linspace(0,max_t,max_t*sr)
#t = np.reshape(t, (t.size,1))
# lists of relevant hourly data
hmean = []
hmax = []
hhts = []
hht10 = []
index = 0
# jonswap
if spectrum == 0:
for dat in data:
index += 1
# generate spectrum
S = wave_util.jonswap(dat[0], dat[1], 3600, f)
# generate elevation data
E = wave_util.ema(S, 0.5, del_f)
# zerodowncross
mean, max, hts, ht10 = wave_util.zerodown(E, sr)
print index, [dat[0] - hts[0], dat[1] - hts[1]]
# add to our hourly list
#print file, mean, max, hts, ht10
hmean.append(mean)
hmax.append(max)
hhts.append(hts)
hht10.append(ht10)
# wallops
elif spectrum == 1:
for dat in data:
index += 1
# generate spectrum
S = wave_util.wallops(dat[0], dat[1], 10.0, f)
# generate elevation data
E = wave_util.ema(S, 0.5, del_f)
# zerodowncross
mean, max, hts, ht10 = wave_util.zerodown(E, sr)
print index, [dat[0] - hts[0], dat[1] - hts[1]]
# add to our hourly list
#print file, mean, max, hts, ht10
hmean.append(mean)
hmax.append(max)
hhts.append(hts)
hht10.append(ht10)
# custom spectrum -- based on bert-schneider
elif spectrum == 2:
for dat in data:
index += 1
#generate spectrum
S = wave_util.genspec(hs=dat[0], ts=dat[1]-1., f=f, foffset=0, A=0.072, C=0)
E = wave_util.ema(S, 0.5, del_f)
mean, max, hts, ht10 = wave_util.zerodown(E, sr)
print index, [dat[0] - hts[0], dat[1] - hts[1]]
# add to our hourly list
#print file, mean, max, hts, ht10
hmean.append(mean)
hmax.append(max)
hhts.append(hts)
hht10.append(ht10)
data[:] = np.asarray(data)
hhts[:] = np.asarray(hhts)
print np.mean(data), np.mean(hhts)
return (hmean, hmax, hhts, hht10, data)
def maximizem(sr):
curpath = os.path.dirname(os.path.abspath(__file__))
with open(os.path.join(curpath,'hts_data'),'r') as f:
hts_raw = f.read().split('\n')
dhts = [[float(r.split(',')[0]),float(r.split(',')[1])] for r in hts_raw if r]
with open(os.path.join(curpath,'mean_data'),'r') as f:
mean_raw = f.read().split('\n')
dmean = [[float(r.split(',')[0]),float(r.split(',')[1])] for r in mean_raw if r]
with open(os.path.join(curpath,'max_data'),'r') as f:
max_raw = f.read().split('\n')
dmax = [[float(r.split(',')[0]),float(r.split(',')[1])] for r in max_raw if r]
with open(os.path.join(curpath,'ht10_data'),'r') as f:
ht10_raw = f.read().split('\n')
dht10 = [[float(r.split(',')[0]),float(r.split(',')[1])] for r in ht10_raw if r]
"""
constants for spectrum and sea water level generation. we define them outside the loop to save some time
"""
max_t = 3600 # maximum duration of time series
nyq_f = sr/2 # nyquist frequency of spectra: highest frequency bin
del_f = 1.0/max_t # interval between frequency bins
f = np.linspace(del_f,nyq_f,max_t) # array containing all frequency bins
t = np.linspace(0,max_t,max_t*sr)
#t = np.reshape(t, (t.size,1))
# list of hourly error rates
emean = []
emax = []
ehts = []
eht10 = []
for m in xrange(2,10):
# lists of relevant hourly data
hmean = []
hmax = []
hhts = []
hht10 = []
index = 0
for dat in dhts:
index += 1
# generate spectrum
S = wave_util.wallops(dat[0], dat[1], m, f)
# generate elevation data
E = wave_util.ema(S, 0.5, del_f)
# zerodowncross
mean, max, hts, ht10 = wave_util.zerodown(E, sr)
print index, [dat[0] - hts[0], dat[1] - hts[1]]
# add to our hourly list
#print file, mean, max, hts, ht10
hmean.append(mean)
hmax.append(max)
hhts.append(hts)
hht10.append(ht10)
hhts[:] = np.asarray(hhts)
hmean[:] = np.asarray(hmean)
hmax[:] = np.asarray(hmax)
hht10[:] = np.asarray(hht10)
meandmean = np.mean(np.asarray(dmean))
meandmax = np.mean(np.asarray(dmax))
meandhts = np.mean(np.asarray(dhts))
meandht10 = np.mean(np.asarray(dht10))
emean.append((np.mean(hmean) - meandmean) / meandmean)
emax.append((np.mean(hmax) - meandmax) / meandmax)
ehts.append((np.mean(hhts) - meandhts) / meandhts)
eht10.append((np.mean(hht10) - meandht10) / meandht10)
return emean, emax, ehts, eht10
def synthetic(hsfile, spectrum, sr):
curpath = os.path.dirname(os.path.abspath(__file__))
with open(os.path.join(curpath, hsfile), 'r') as f:
raw = f.read().split('\n')
data = [[float(r.split(',')[0]), float(r.split(',')[1])] for r in raw if r]
"""
constants for spectrum and sea water level generation. we define them outside the loop to save some time
"""
max_t = 3600 # maximum duration of time series
nyq_f = sr / 2 # nyquist frequency of spectra: highest frequency bin
del_f = 1.0 / max_t # interval between frequency bins
f = np.linspace(del_f, nyq_f, max_t) # array containing all frequency bins
t = np.linspace(0, max_t, max_t * sr)
#t = np.reshape(t, (t.size,1))
# lists of relevant hourly data
hmean = []
hmax = []
hhts = []
hht10 = []
index = 0
# jonswap
if spectrum == 0:
for dat in data:
index += 1
# generate spectrum
S = wave_util.jonswap(dat[0], dat[1], 3600, f)
# generate elevation data
E = wave_util.ema(S, 0.5, del_f)
# zerodowncross
mean, max, hts, ht10 = wave_util.zerodown(E, sr)
print index, [dat[0] - hts[0], dat[1] - hts[1]]
# add to our hourly list
#print file, mean, max, hts, ht10
hmean.append(mean)
hmax.append(max)
hhts.append(hts)
hht10.append(ht10)
# wallops
elif spectrum == 1:
for dat in data:
index += 1
# generate spectrum
S = wave_util.wallops(dat[0], dat[1], 10.0, f)
# generate elevation data
E = wave_util.ema(S, 0.5, del_f)
# zerodowncross
mean, max, hts, ht10 = wave_util.zerodown(E, sr)
print index, [dat[0] - hts[0], dat[1] - hts[1]]
# add to our hourly list
#print file, mean, max, hts, ht10
hmean.append(mean)
hmax.append(max)
hhts.append(hts)
hht10.append(ht10)
# custom spectrum -- based on bert-schneider
elif spectrum == 2:
for dat in data:
index += 1
#generate spectrum
S = wave_util.genspec(hs=dat[0],
ts=dat[1] - 1.,
f=f,
foffset=0,
A=0.072,
C=0)
E = wave_util.ema(S, 0.5, del_f)
mean, max, hts, ht10 = wave_util.zerodown(E, sr)
print index, [dat[0] - hts[0], dat[1] - hts[1]]
# add to our hourly list
#print file, mean, max, hts, ht10
hmean.append(mean)
hmax.append(max)
hhts.append(hts)
hht10.append(ht10)
data[:] = np.asarray(data)
hhts[:] = np.asarray(hhts)
print np.mean(data), np.mean(hhts)
return (hmean, hmax, hhts, hht10, data)
def maximizem(sr):
curpath = os.path.dirname(os.path.abspath(__file__))
with open(os.path.join(curpath, 'hts_data'), 'r') as f:
hts_raw = f.read().split('\n')
dhts = [[float(r.split(',')[0]),
float(r.split(',')[1])] for r in hts_raw if r]
with open(os.path.join(curpath, 'mean_data'), 'r') as f:
mean_raw = f.read().split('\n')
dmean = [[float(r.split(',')[0]),
float(r.split(',')[1])] for r in mean_raw if r]
with open(os.path.join(curpath, 'max_data'), 'r') as f:
max_raw = f.read().split('\n')
dmax = [[float(r.split(',')[0]),
float(r.split(',')[1])] for r in max_raw if r]
with open(os.path.join(curpath, 'ht10_data'), 'r') as f:
ht10_raw = f.read().split('\n')
dht10 = [[float(r.split(',')[0]),
float(r.split(',')[1])] for r in ht10_raw if r]
"""
constants for spectrum and sea water level generation. we define them outside the loop to save some time
"""
max_t = 3600 # maximum duration of time series
nyq_f = sr / 2 # nyquist frequency of spectra: highest frequency bin
del_f = 1.0 / max_t # interval between frequency bins
f = np.linspace(del_f, nyq_f, max_t) # array containing all frequency bins
t = np.linspace(0, max_t, max_t * sr)
#t = np.reshape(t, (t.size,1))
# list of hourly error rates
emean = []
emax = []
ehts = []
eht10 = []
for m in xrange(2, 10):
# lists of relevant hourly data
hmean = []
hmax = []
hhts = []
hht10 = []
index = 0
for dat in dhts:
index += 1
# generate spectrum
S = wave_util.wallops(dat[0], dat[1], m, f)
# generate elevation data
E = wave_util.ema(S, 0.5, del_f)
# zerodowncross
mean, max, hts, ht10 = wave_util.zerodown(E, sr)
print index, [dat[0] - hts[0], dat[1] - hts[1]]
# add to our hourly list
#print file, mean, max, hts, ht10
hmean.append(mean)
hmax.append(max)
hhts.append(hts)
hht10.append(ht10)
hhts[:] = np.asarray(hhts)
hmean[:] = np.asarray(hmean)
hmax[:] = np.asarray(hmax)
hht10[:] = np.asarray(hht10)
meandmean = np.mean(np.asarray(dmean))
meandmax = np.mean(np.asarray(dmax))
meandhts = np.mean(np.asarray(dhts))
meandht10 = np.mean(np.asarray(dht10))
emean.append((np.mean(hmean) - meandmean) / meandmean)
emax.append((np.mean(hmax) - meandmax) / meandmax)
ehts.append((np.mean(hhts) - meandhts) / meandhts)
eht10.append((np.mean(hht10) - meandht10) / meandht10)
return emean, emax, ehts, eht10
