README.md

AA triggers selection using forecast verification

The following Jupyter notebooks and Python scripts need to be executed sequentially:

1. Notebook src/01-input-spi-seas51: Contains methods for processing SEAS51 data and calculating the Standardized Precipitation Index (SPI).

2. Notebook src/02-input-spi-chrips: Contains methods for processing CHIRPS data and calculating SPI.

3. Python Script src/run_kmj.py: Performs 2D and 1D forecast verification, calculating metrics such as AUROC using bootstrapping. It subsets and saves trigger data and metrics as CSV files.

4. Python Script src/run_map.py: Generates stamp plot maps of SPI for forecast ensemble members, CHIRPS observations, and threshold exceedance empirical probabilities from NetCDF files.

5. Python Script src/run_bar.py: After manually selecting trigger values from the CSV file in step 3, this script creates bar plots of selected triggers, comparing 1D analyses of observations and forecasts.

6. Python Script src/run_heatmap.py: Generates a heatmap of decisive triggers, displaying False Alarm Rate (FAR) and Hit Rate (HR).

7. Python Script src/run_latex_table.py: Creates a LaTeX longtable for inclusion in the analysis report, listing trigger values with AUROC scores greater than 0.5 and highlighting manually selected triggers.

To compile the report, use the LaTeX template provided in the doc/ folder.