def activate(self, mode=None):
"""
If there are no analysis parameters given, eg no user input of mean or sigma, we will update those based on the activation function.
"""
if self.activation_function == 'gauss':
if mode == 'mean':
return np.mean(self.universe)
if mode == 'sigma':
return np.sigma(self.universe)
elif self.activation_function == 'composite_gauss':
if mode == 'mean':
return np.mean(self.universe)
if mode == 'sigma':
return np.sigma(self.universe)
else:
raise Exception("Hey man you need a proper activation function here")
def generate_data(nbr_iterations):
dims = (3, 1)
start_positions = np.array(np.sigma(0, 0.5))
# Computing trajectory
data = [start_positions]
for iteration in range(nbr_iterations):
previous_positions = data[-1]
new_positions = np.array(sigma(iteration / (2 * np.pi), 0.5))
data.append(new_positions)
return data
def cloudfilter(allprf, tarr, z, datestring):
r"""
Cloud filter
This function computes the cloud filter equations described in [Teschke (2008),Garcia-Franco (2017), Garcia-Franco et.al. (2018)]
**Parameters**
**allprf**: `np.nadarray`
backscattering matrix
**tarr** : `np.array`
Time-array, usually decimal hours.
**z** : `np.array`
Height-array usually length 500, start=10, end=5000, t_step=10 [m]
**datestring** : `string`
String for date, typically %Y%m%d, e.g., 20160305
:rtype: datetimearray, flag vector: temps returns all datetimes where cloud or precipitation
has been found, flag vector is a numpy array with the dimensions of tarr where clouds are depicted
as 1 and clear-sky conditions as 0.
**Statistical filter**
.. math:: \beta_\sigma (z,t)=B(z,t)\sigma(t)
.. math:: \mu=\frac{1}{N_zN_t}\sum_z\sum_t\beta_\sigma(z,t)
.. math:: \sum=\frac{1}{N_zN_t-1}\sum_z\sum_t[\beta_\sigma(z,t)-\mu]^2.
.. math:: B_N=\mu+3\sqrt{\sum}
* :math:`B(z,t)` Backscattering matrix
* :math:`\sigma(t)` Variance over time $t$.
* :math:`\mu` Global mean of \beta_\sigma(z,t)
* :math:`\sum` Global variance of \beta_\sigma(z,t)
* :math:`z_{max}` Maximum integration level [m]
* :math:`B_N` Threshold for determining cloud or no cloud, function of both global mean and variance.
`B_N` defines the threshold value used for determining whether or not a profile at time `t` presents
cloud and precipiation or not. If B(z,t)>B_N then cloud and precipitation are present. Else, clear-sky conditions
are considered.
"""
thrs = 1750.0
day = datestring
dia = int(day[6:8])
year = int(day[0:4])
mes = int(day[4:6])
zi = np.where(z == 200.)
zi = zi[0]
zmax = np.where(z == 4000.)
zmax = zmax[0]
#print zi,zmax
a = allprf[zi:zmax, :]
nbs = []
tmps = []
sigma_t = np.sigma(tarr)
count = len(tarr) * len(z[zi:zmax])
sumi = 0
for i, t in enumerate(tarr):
for j, z1 in enumerate(z[zi:zmax]):
try:
sumi = sumi + a[j + zi, i]
except IndexError:
# print fl
continue
mu = sumi / count
deviat = 0
for i, t in enumerate(tarr):
for j, z1 in enumerate(z[zi:zmax]):
try:
deviat = deviat + ((a[j + zi, i] - mu)**2)
except IndexError:
continue
sigma = deviat / (count - 1)
ec = mu + (3 * np.sqrt(sigma))
# print 'media','sigma','3','s'
# print mu,sigma,three,two
#if ec > 1400:
# print fl, ec
for i, t in enumerate(tarr):
cloud = False
for j, z1 in enumerate(z[zi:zmax]):
try:
if a[j + zi, i] > ec or a[j + zi, i] > thrs:
horas = int(math.floor(t))
minutos = int(round((t - horas) * 60, -1))
tim = datetime.datetime(year, mes, dia, horas, minutos)
tmps.append(tim)
#nbs.append(1)
cloud = True
break
except IndexError:
continue
if cloud:
nbs.append(1)
else:
nbs.append(0)
# print tmps
# print nbs
return tmps, nbs