This example demonstrates how to run precipitation PDF (Probability Density Function) diagnostics comparing model output against observational datasets.

The precip_pdf diagnostic set:

• Calculates frequency and amount PDFs of daily precipitation
• Compares model output against observational datasets (GPCP, IMERG)
• Analyzes multiple regions: TROPICS, CONUS, and global
• Supports seasonal subsetting (ANN, DJF, MAM, JJA, SON)
• Generates interactive HTML viewer with detailed plots

• ex10.py: Main run script with parameter configuration
• diags.cfg: Configuration file specifying regions and reference datasets

• reference_data_path: Location of observational daily precipitation data
• test_data_path: Location of model daily time series output

• test_start_yr, test_end_yr: Years to analyze from model data
• ref_start_yr, ref_end_yr: Years to analyze from observational data

• season_subset: When True, analyzes all seasons (DJF, MAM, JJA, SON) plus annual
• save_netcdf: When True, saves calculated PDFs to NetCDF for later use
• regions: List of regions to analyze ("global", "TROPICS", "CONUS")
• ref_name: Reference dataset(s) - can be single string or list

# Allocate a node (or use batch job)
salloc --nodes 1 --qos interactive --time 01:00:00 --constraint cpu --account=e3sm

# Load E3SM Unified environment
source /global/common/software/e3sm/anaconda_envs/load_latest_e3sm_unified_pm-cpu.sh

# Run the diagnostics
cd examples/ex10-precip-pdf
python ex10.py -d diags.cfg --multiprocessing --num_workers=32

# Adjust permissions to view results
chmod -R 755 <your web directory>

e3sm_diags precip_pdf --no_viewer \
  --reference_data_path '/global/cfs/cdirs/e3sm/diagnostics/observations/Atm/time-series/' \
  --test_data_path '/global/cfs/cdirs/e3sm/chengzhu/tests/zppy_example_v3/v3.LR.amip_0101/post/atm/180x360_aave/ts/daily/10yr' \
  --results_dir '/global/cfs/cdirs/e3sm/www/chengzhu/tests/precip_pdf_test_seasons_multi_ref' \
  --case_id 'precipitation-pdf' \
  --sets 'precip_pdf' \
  --variables 'PRECT' \
  --regions 'TROPICS' \
  --save_netcdf \
  --test_name 'E3SMv3.LR.amip_0101' \
  --ref_name 'GPCPDAY_3.3' 'IMERG_Daily' \
  --season_subset \
  --test_start_yr 1995 \
  --test_end_yr 2004 \
  --ref_start_yr 2001 \
  --ref_end_yr 2010

The diagnostics generate:

• Plots: PNG files showing frequency and amount PDFs for each region/season
• NetCDF files: Cached global gridded PDFs (if save_netcdf=True)
• HTML Viewer: Interactive viewer at <results_dir>/viewer/precip_pdf/index.html

• Single "Precipitation PDF" group containing all results
• Each row represents a unique region + reference dataset combination
• Detail pages for each season showing:
  • Frequency PDF (df/dlog(P))
  • Amount PDF (dA/dlog(P))
  • Links to NetCDF data files
  • Metadata and command to recreate

• GPCPDAY_3.3: GPCP 1-Degree Daily v3.3 (1996-2023)
• IMERG_Daily: GPM IMERG Daily (2001-2021)

• global: Global mean
• TROPICS: 30°S - 30°N
• CONUS: Continental United States

• Daily precipitation data is required (not monthly climatologies)
• Model variable should be PRECT (total precipitation rate)
• Reference data is automatically subset to match test data time range
• PDFs are calculated on exponentially-spaced bins (0.1 to ~900 mm/day)
• The viewer fix (v3.1.0+) ensures that all regions generate distinct detail HTML pages