ccnact 0.0.9

KISS SciPy-based CCN activation model

☁️ KISS SciPy-based CCN activation model

📌 overview

ccnact is a simple, yet complete, adiabatic/hydrostatic air-parcel framework employing moving-sectional/particle-resolved aerosol-cloud microphysics, featuring:

• 🧮 integration using SciPy's interface to LSODA stiff ODE solver
• 📝 ODE system based on Arabas & Shima 2017 (extended to polydisperse aerosol size spectrum)
• 🏁 κ-Köhler wet radii equilibration of input dry-size spectrum with SciPy's elementwise root finder
• 🌪️ capability of resolving aerosol activation, deactivation, drop growth, evaporation and ripening
• 📈 mulit-modal lognormal (using SciPy's stats routines) spectrum specification (concentration at STP)
• 🔌 portable across platforms and architectures (CI on Linux, macOS & Windows, on Intel & ARM CPUs)
• ⚙️ single-function interface allowing to modify every single constant, and returning a tuple of:
  • concentration of activated droplets (at STP)
  • relative humidity (saturation) along the ascent (1D array)
  • wet radii for all sections/particles (2D array)
  • time values (1D array)
• 🧩 effective interfacing options for Matlab, IDL, Julia, etc
• ⚖️ unit-aware implemetation using Pint (dimensional analysis enabled for tests only)
• 🚀 subsecond execution times for common parameter settings
• 🔗 KISS design: depends on SciPy, NumPy & Pint only; single ~500 LOC file (physics + setup + tests)

The last five points were the key motivating factors for the development - the project originated from a search for a simple, lightweight (in dependencies) and fast CCN activation air-parcel model with concise code, automated testing and no hardcoded constants.

💡 example notebooks

• exploring dependence of activatied fraction and maximal supersaturation on updraft velocity:
  
• plotting the raw state of the model (saturation and wet radii profiles):
  

💻 notes for users

To install the package, try: pip install ccnact

Using from Python:

from ccnact import parcel
help(parcel)
n_act, rh, r_w, time = parcel(...)

Interfacing from Matlab (using the built-in Python bridge):

ccnact = py.importlib.import_module('ccnact');
ccnact.parcel(pyargs(...
   'MAC', 1,...
   'n_bins', int32(100),...
   'p', 101300,...
   'T', 300,...
   'RH', .99,...
   'dt', 1,...
   'nt', int32(100),...
   'w', 2,...
   'sigma', 0.072,...
   'kappa', py.tuple({1}),...
   'meanr', py.tuple({3e-8}),...
   'gstdv', py.tuple({1.5}),...
   'n_tot', py.tuple({1e9})...
))

⚙ notes for developers

To execute the tests after checking out from git: pip install -e .[dev]; pytest ccnact.py

To set-up pre-commit: pip install pre-commit; pre-commit install