Python 3 6 support (josephhardinee#46)

This adds py3.6 support to the travis CI tests, pins hdf for testing, and had a few of my ports from pyparticle probe leak in. 

* Moved dsd fitting from pyparticle probe ua98 module into a named ua98 module instead of the previous dsd_fits module.

* Added py3.6 to travis.yml file

* Adding changes to fix travis CI

.travis.yml

@@ -25,6 +25,12 @@ matrix:
         - PYTHON_VERSION="3.5"
         - NOSE_ARGS="-v --no-skip --with-cov --cov pydisdrometer pydisdrometer"
         - COVERALLS="true"
+      - python: "3.6"
+        env: 
+        - PYTHON_VERSION="3.6"
+        - NOSE_ARGS="-v --no-skip --with-cov --cov pydisdrometer pydisdrometer"
+        - COVERALLS="true"
+
 
 
 before_install:

continuous_integration/install.sh

@@ -29,23 +29,30 @@ conda create -n testenv --yes pip python=$PYTHON_VERSION
 source activate testenv
 
 # Install dependencies
-conda install --yes numpy scipy matplotlib netcdf4 nose sphinx numpydoc h5py
+conda install --yes numpy scipy matplotlib netcdf4 nose sphinx numpydoc hdf4=4.2.12
 conda install --yes sphinx_rtd_theme
 
 
 if [[ $PYTHON_VERSION == '2.7' ]]; then
-    conda install --yes sphinx numpydoc h5py
+    conda install --yes sphinx numpydoc hdf4=4.2.12
     conda install --yes sphinx_rtd_theme
     pip install sphinxcontrib-bibtex
     pip install xmltodict
     pip install pytmatrix
 fi
 if [[ $PYTHON_VERSION == '3.4' ]]; then
     pip install pytmatrix
+    conda install --yes hdf4=4.2.12
 fi
 if [[ $PYTHON_VERSION == '3.5' ]]; then
     pip install pytmatrix
+    conda install --yes hdf4=4.2.12
 fi
+if [[ $PYTHON_VERSION == '3.6' ]]; then
+    pip install pytmatrix
+    conda install --yes hdf4=4.2.12
+fi
+
 
 # install coverage modules
 pip install nose-cov

pydisdrometer/__init__.py

@@ -14,3 +14,4 @@
 
 from . import partition
 from . import utility
+from . import fit

pydisdrometer/fit/ua98.py

@@ -0,0 +1,292 @@
+# -*- coding: utf-8 -*-
+'''
+The calc_dsd module contains various fitting functions to calculate dsd parameters
+based on measured drop size distributions. Many of these (all currently) were ported from the 
+pyparticleprobe package. 
+'''
+
+from __future__ import print_function, division
+
+import numpy as np
+import scipy.special as scifunct
+
+from ..utility import configuration
+
+"""
+
+A grouping of functions to calculate parameters of a drop size distribution
+ using the methodology of  Ulbrich and Atlas (JAM 1998, JAMC 2007).
+That study assumes a gamma distribution
+ (using the complete gamma function) of the DSD
+      N(D) = N0 * D^mu * exp(-Lambda*D)
+
+ and uses the method of moments, with the nth moment expressed:
+                               Gamma(n + mu + 1)
+      Mn = D^n*N(D)*dD = N0 * -------------------
+                              Lambda^(n + mu + 1)
+
+ where N0 is the intercept parameter (mm^(-1-m) m^-3), mu is the shape
+  parameter (unitless) and Lambda is the slope paramter (m).
+
+            (7-11eta) - [(7-11eta)^2 -4(eta-1)(30eta-12)]^(1/2)
+      mu =  -----------------------------------------------------
+                                 2(eta-1)
+ where
+               M4^2
+      eta = -----------
+             M2 * M6
+
+ and
+               ( (4+mu)(3+mu)M2 )^(1/2)
+      Lambda =  (----------------)
+               (       M4       )
+
+
+ and N0 can be found using the 6th moment (M6 or Ze) and solving for N0
+           M6 * Lambda^(7+mu)
+      N0 = -----------------
+              Gamma(7+mu)
+
+Adapted by Nick Guy.
+
+"""
+
+# Define various constants that may be used for calculations
+rho0=1.2 # reference density at the surface [kg m^-3]
+rhoL=1000. # density of water [kg m^-3]
+
+def eta_ratio(M2,M4,M6):
+    """Compute the ratio eta using the method of moments
+
+    Ulbrich and Atlas (JAM 1998), Eqn 8
+
+    Parameters
+    ----------
+    M2: float
+        Second moment of the DSD [m^-1]
+    M4: float
+        Fourth moment of the DSD [m]
+    M6: float
+        Sixth moment of the DSD [m^3]
+
+    Returns
+    -------
+    Eta: float
+        Ratio of M4^2/(M2*M6) [unitless]
+
+    Notes
+    -----
+    This particular methodology uses the 2nd, 4th, and 6th moments.
+    """
+
+    Eta = M4**2 / (M2 * M6)
+    return Eta
+
+def shape(M2,M4,M6):
+    """Compute the shape parameter mu using the method of moments
+
+    Ulbrich and Atlas (JAM 1998), Eqns 8 and 9
+
+    Parameters
+    ----------
+    M2: float
+        Second moment of the DSD [m^-1]
+    M4: float
+        Fourth moment of the DSD [m]
+    M6: float
+        Sixth moment of the DSD [m^3]
+
+    Returns
+    -------
+    mu: float
+        Shape parameter of gamma DSD model [unitless]
+    Notes
+    -----
+
+  This particular methodology uses the 2nd, 4th, and 6th moments.
+  First the 3rd moment normalized by Dm (M4/M3) is calculated.
+  Next the shape paramter is computed.
+  The denominator [2*(Eta-1)] values are filtered between -1E-1 and +1E-1
+  to ensure that unrealistically large shape parameter values 
+  are not calculated.
+    """
+    # Calculate the eta ratio [eq. 8]
+    eta = eta_ratio(M2,M4,M6)
+
+    # Now calculate the shape parameter [eq. 3]
+    muNumer = (7.0 - 11.0 * eta) - np.sqrt((7.0 - 11.0 * eta)**2.0 - 4.0 * (eta - 1.0)
+                                    * (30. * eta - 12.0))
+    muDenom = 2.0 * (eta - 1.0)
+
+    # Mask any zero or unrealistically low values in denominator
+    muDenom = np.ma.masked_inside(muDenom,-1E-1,1E-1)
+
+    mu = muNumer / muDenom
+
+    return mu
+
+def slope(M2,M4,mu):
+    """Compute the slope (Lambda) using the method of moments
+
+    Ulbrich and Atlas (JAM 1998), Eqn 10
+    Parameters
+    ----------
+    M2: float
+        Second moment of the DSD [m^-1]
+    M4: float
+        Fourth moment of the DSD [m]
+    mu: float
+        Shape parameter of gamma DSD model [unitless]
+
+    Returns
+    -------
+    Lambda: float
+        Slope [m^-1]
+    Notes:
+    ------
+    This particular methodology uses the 3rd and 4th moments and shape
+    paramter to compute the slope is computed.
+    """
+    Lambda = np.ma.sqrt((4 + mu) * (3 + mu) * M2 / M4)
+
+    return Lambda
+
+def intercept(M6,mu,Lambda):
+    """Compute the intercept parameter (N0) using the method of moments 
+
+    Ulbrich and Atlas (JAM 1998), Eqn 6 solved for N0
+
+    Parameters
+    ----------
+    M6: float
+        Sixth moment of the DSD [m^3]
+    mu: float
+        Shape parameter of gamma DSD model [unitless]
+    Lambda: float
+        Slope [m^-1]
+    Returns
+    -------
+    N0: float
+        Intercept parameter [m^(-1-shape) m^-3]
+
+    Notes
+    -----
+    The exponents work out in the following way:
+    M6*Lambda^(7+mu) = m^-3*m^-1(7+mu) = m^-3*m^-7*m^-mu = m^(-1-mu) * m^-3
+    """
+
+    # Calculate numerator
+    IntNumer = M6 * Lambda**(7. + mu)
+
+    # Calculate denominator
+    # Mask values near 0., otherwise gamma function returns "inf"
+    mucopy = np.ma.masked_inside(mu,-1E-1,1E-1)
+    IntDenom = scifunct.gamma(7 + mucopy)
+    # Mask any invalid values
+    np.ma.masked_invalid(IntDenom)
+
+    # Mask any zero values from Denom (-999. a problem)
+    IntDenom = np.ma.masked_equal(IntDenom,0.)
+    #print(mu(:,0)+"  "+Lambda(:,0)+"  "+IntNumer(:,0)+"  "+IntDenom(:,0))
+
+    # Calculate N0
+    N0 = IntNumer / IntDenom
+
+    return N0
+
+def mom_d0(mu,Lambda):
+    """Compute the median volume diameter (D0) using the method of moments.
+
+    Ulbrich and Atlas (JAM 1998), Eqn 11
+
+    Parameters
+    ----------
+    mu: float
+        Shape parameter of gamma DSD model [unitless]
+    Lambda: float
+        Slope paramter of gamma DSD model [m^-1]
+
+    Returns
+    -------
+    D0: float
+        Median volume diameter [m]
+    """
+
+    D0 = (3.67 + mu) / Lambda
+    return D0
+
+def zr_a(mu,N0):
+    """Assuming a rainfall parameter relationship of Z=AR^b,
+    Compute the A prefactor using gamma distribution.
+
+    Ulbrich and Atlas (JAMC 2007), Eqn T5
+
+    Parameters
+    ----------
+    mu: float
+        Shape parameter of gamma DSD model [unitless]
+    N0: float
+        Intercept parameter [m^(-1-shape) m^-3]
+
+    Returns
+    -------
+    A: float
+        Z-R prefactor (see description)
+    """
+
+    # Mask 0. values, otherwise gamma function returns "inf"
+    mucopy = np.ma.masked_equal(mu,0.)
+
+    # gamma_fix is a patch for versions earlier than NCL v6.2,
+    # which cannot handle missing data in the gamma function
+    ANumer = 10E6 * scifunct.gamma(7 + mucopy) * N0**(-2.33/(4.67 + mu))
+    ADenom = (33.31 * scifunct.gamma(4.67 + mucopy))**((7 + mu)/(4.67 + mu))
+
+    # Mask any zero values from Denom
+    ADenom = np.ma.masked_equal(ADenom,0.)
+
+    A = ANumer / ADenom
+
+    return A
+
+def zr_b(mu):
+    """Assuming a rainfall parameter relationship of Z=AR^b,
+    Compute the b exponent using gamma distribution.  
+
+    Ulbrich and Atlas (JAMC 2007), Eqn T5
+    Parameters
+    ----------
+    mu: float
+        Shape parameter of gamma DSD model [unitless]
+
+    Returns
+    -------
+    b: float
+        Z-R prefactor (see description)
+    """
+
+    b = (7 + mu) / (4.67 + mu)
+    return b
+
+def norm_intercept(LWC,Dm):
+    """Calculates the normalized intercept parameter, which
+    is more physically meaningful than N0.
+
+    Ulbrich and Atlas (JAMC 2007), Eqn T9
+
+    Parameters
+    ----------
+    LWC: float
+        Liquid water content [g m^-3]
+    Dm: float
+        Volume weigthed mean diameter [m]
+
+    Returns
+    -------
+    Nw: float
+        Normalized intercept parameter
+    """
+    # The factor of 1000 converts the water density from kg/m^3 to g/m^3
+    Nw = (256 / (np.pi * rhoL * 1000.)) * (LWC / Dm**4)
+
+    return Nw