Python scripts for tropical diagnostics of NWP forecasts.
The diagnostics are meant to be applied to gridded forecast data and example scripts are provided to show how to apply the diagnostics at different lead times.
Required model output is primarily precipitation. This is enough to compute Hovmoeller diagrams and compare to observations and to project onto the convectively coupled equatorial wave (CCEW) EOFs to analyze CCEW activity and skill in model forecasts.
For the moisture convection coupling analysis vertical level data of specific humidity and temperature are needed, in addition to surface pressure and a land-sea mask. The vertical coherence diagnostic also uses data at vertical levels: temperature, specific humidity, divergence, zonal and meridional winds. However, it is up to the user to decide which variables are most useful to the user's specific application.
Contains the functions and modules necessary to compute the various diagnostics. The main diagnostics included are:
Functions to compute hovmoeller latitudinal averages and pattern correlation are included in hovmoeller_calc.py. Plotting routines are included in hovmoeller_plotly.py.
Functions for computing 2D Fourier transforms and 2D power and cross-spectra are included in spacetime.py. To plot the spectra spacetime_plot.py uses pyngl, which is based on NCL and provides similar control over plotting resources.
Functions to project precipitation (either from model output or observations) onto CCEW EOF patterns and compute wave activity and a CCEW skill score are included in CCEWactivity.py. Also included are routines to plot the activity and the skill compared to observations.
These functions compute the coherence between two data sets at multiple vertical levels. The assumption is that the first data set is filtered precipitation (filtered for a CCEW or the MJO) and the second data set is a multi- level dynamical field, either from reanalysis or model output. Once coherence between all variables and filtered precipitation has been computed and saved the plotting routines to plot vertical profiles of coherence and phase angles can be used. The plotting is done using plotly and the kaleido engine. Example scripts are provided for both the computational (vertical_coherence.py) and plotting (vertical_coherence_plot.py) part.
These functions contain code to compute column saturation fraction (CSF) from reanalysis or model output and bin CSF and precipitation against each other (moisture_convection_coupling.py). Routines for plotting the circulation in preciptiation - CSF space and the distributions are contained in moisture_convection_coupling_plot.py. The example script precip_csf_coupling.py computes CSF from model output, bins TRMM precipitation against model CSF and plots the results. B_L_analysis.py shows how to run the B_L analysis scripts and plots results.
Scripts containing example use cases. These scripts read in data, compute diagnostics and plot the results. This is a good place to start when first using the diagnostics. The user will need to supply their own data and edit these examples to get them to work.
Download or clone the source code from github. Make sure all the required packages are available. If working on a machine that requires python environments to install packages create a conda environment first:
MyEnv = your chosen name of the environment
conda create --name MyEnv
Install the required packages:
conda install -n MyEnv scipy
conda install -n MyEnv xarray
conda install -n MyEnv numpy
conda install -n MyEnv netCDF4
conda install -n MyEnv pyngl
Activate the conda environment:
conda activate MyEnv
Change directory to top level package directory tropical_diagnostics and install package using pip.
cd tropical_diagnostics/
pip install./
To run scripts from the examples directory copy the script into a directory where you would like to run. Make
sure to activate the conda environment with conda activate MyEnv. Adjust all path and filenames if necessary.