Detectable Climate

Scripts and notebooks related to the manuscript titled: "Improving Climate Reproducibility Testing with False Discovery Rate Correction".

Submitted to Platform for Advanced Scientific Computing (PASC) Conference

About

This investigation was set to improve the usability of non bit-for-bit testing in the E3SM Earth system model primarily by reducing the number of false positive tests (e.g. a statstical type I error where the test claims two ensembles have different simulated climates when in fact they are not)

Status

Package directories

• The run_scripts directory contains scripts nessecary for setup and execution of E3SM (specific to the LCRC cluster "chrysalis")

• The detclim directory contains analysis scripts

• The detclim/notebooks directory contains exploratory notebooks

Requirements

As described in pyproject.toml:

• setuptools
• numpy
• scipy
• pandas>=2.0.1
• matplotlib
• statsmodels
• xarray
• cartopy
• jupyter
• cftime
• dask
• nco
• nc-time-axis
• dask-mpi
• seaborn
• hvplot
• bokeh
• holoviews