First release packaged for PyPI and Conda-Forge.
- Changed membership beta functions for vertical ice and ice crystals.
- Removed executable portion (when running as main script) that was never used.
- Conform project to modern, best practice standards.
- Add basic Travis CI tests.
- Fix and enhance setup script and prepare for proper packaging.
- Drop support for EoL Python 2.7 and add it for 3.7.
- Add CHANGELOG and RELEASE instructions and update/improve README.
Version bump; no non-trivial changes.
Initial Github release.
- Updates to the example Jupyter notebooks.
- Added the dBZ + Kdp threshold back in to eliminate small reflectivities with non-zero Kdp blowing up the rain rates.
- Changed import location for _check_for_array
- Moved base rainfall functions to common, and also now import _check_for_array from common.
- Moved functions from csu_blended_rain here.
- Added ability for user to provide custom parameters to polarimetric rainfall equations via the blended rainfall routines. Also customized blended routines to handle non-S bands. In this case, only R-Z and R-Kdp are used.
- Python 3 compatible.
- Cython now an option for speeding up KDP routines.
- Updated for Cython. Confirmed ~50 times faster than 1-deg np.polyfit().
- Cython now an option for speeding up the hid_beta routines.
- Made number of gates used for standard deviation calculation adjustable.
- Now using a Fortran shared object (calc_kdp_ray_fir) to do the ray-based KDP calculations. This has vastly sped up the overall KDP processing (> 100x). f2py FTW!
- Updated calc_kdp_bringi to fail softly when (window/gs) is not even.
- Sped up hid_beta by using f2py + working w/ 1-D flattened arrays that are later reshaped to the necessary shape.
- Performance improvements.
- Fixed logical inconsistencies leading to lack of rainfall calculation in HID = rain + low Z + high Kdp/Zdr.
- Added window keyword to enable stretching the FIR window (e.g., use a 21-pt filter over 5 km with 250-m gate spacing).
- Forcing FIR order to be even, _calc_kdp_ray will crash otherwise.
- Vastly sped up despeckle routine using scipy.
- Made Python 3 compatible.
- Fixed issue with non-integer array indices.
- Python 3.
- Python 3.
- Python 3 compliant.
- Made Python 3 compatible.
- Made pep8 compatible.
- Made sub-module pep8 compliant.
- Added despeckle() along with a private helper function.
- Added warnings.warn import.
- Added despeckle() along with a private helper function.
- Added warnings.warn import.
- Made algorithm work with a user-defined gate spacing (via gs keyword). Untested on gate spacings that do not divide evenly into the 3-km window used for filtering the PHIDP data, however. But common gate spacings like 50, 100, 150, 200, 250, and 300 meters should all work fine.
- Made the algorithm capable of receiving 2D array inputs (i.e., azimuth & range) as well as 1D inputs (range only). If 2D, rng needs to be 2D as well. However, thsd should remain a scalar, or 1D and only vary by range.
- Pythonized
- Adjusted MBFs for performance.
- Changed the categories and the MBF values based on scattering simulations. NOTE: LDR values from simulations were added.
- Changed the MBFS to the theory-based S-band values. NOTE: LDR VALUES WERE NOT MODIFIED.
- Changed MBFs for wet/dry graupel and Wet Snow to conform to Liu and Chandrasekar (2000).
- Changed MBF for vertical ice to conform to Carey and Rutledge (1998).
- Changed KDP MBF for vertical ice to be unity from -0.2 to -0.6. Used to be from -0.6 to 0.0. This didn't make much difference.
- Changed the fuzzy sets to include vertical ice and get rid of low/high density snow.