Skip to content

README.md

Latest commit

 

History

History
149 lines (115 loc) · 4.57 KB

README.md

File metadata and controls

149 lines (115 loc) · 4.57 KB
Raw

Scripts

High-level workflow scripts for common validation and analysis tasks.

extract_preprocessed.py

Extract annual means from pre-processed NetCDF files for all RECCAP2 regions and generate time series plots.

Usage

# Basic usage
python scripts/extract_preprocessed.py xqhuc

# With custom base directory
python scripts/extract_preprocessed.py xqhuc --base-dir ~/annual_mean

Requirements

  • Annual mean NetCDF files in ~/annual_mean/{expt}/ (or custom base directory)
    • {expt}_pa_annual_mean.nc (atmosphere)
    • {expt}_pt_annual_mean.nc (TRIFFID)
    • {expt}_pf_annual_mean.nc (ocean)

Outputs

Creates validation_outputs/single_val_{expt}/ containing:

validation_outputs/single_val_{expt}/
├── {expt}_extraction.csv           # Time-mean values for all variables and regions
└── plots/
    ├── allvars_global_{expt}_timeseries.png
    ├── allvars_Europe_{expt}_timeseries.png
    ├── allvars_North_America_{expt}_timeseries.png
    ├── allvars_South_America_{expt}_timeseries.png
    ├── allvars_Africa_{expt}_timeseries.png
    ├── allvars_North_Asia_{expt}_timeseries.png
    ├── allvars_Central_Asia_{expt}_timeseries.png
    ├── allvars_East_Asia_{expt}_timeseries.png
    ├── allvars_South_Asia_{expt}_timeseries.png
    ├── allvars_South_East_Asia_{expt}_timeseries.png
    └── allvars_Oceania_{expt}_timeseries.png

What it does

  1. Extracts data for all RECCAP2 regions (global + 10 regions)
  2. Saves to CSV with time-mean values in same format as observational data
  3. Generates time series plots for each region showing:
    • Carbon fluxes (GPP, NPP, Rh, fgco2)
    • Carbon stocks (CVeg, CSoil)
    • Climate variables (tas, pr)
    • PFT fractions (if available)
  4. Automatically skips regions with no data

CSV Format

{expt}_extraction.csv: Time-mean values in same format as validate_experiment.py output

,global,North_America,South_America,Europe,Africa,...
CSoil,1234.56,123.45,234.56,45.67,345.67,...
CVeg,567.89,56.78,78.90,12.34,89.01,...
GPP,134.81,18.40,31.65,6.11,30.02,...
NPP,68.29,10.36,14.45,3.42,14.45,...
Rh,65.52,7.99,17.20,2.69,15.57,...
...

Notes

  • Only variables successfully extracted are shown in plots
  • If many variables are missing, check:
    1. Annual mean files exist in base directory
    2. Files generated using annual_mean_cdo.sh
    3. STASH codes are correct in files

validate_experiment.py

Comprehensive validation of a single UM experiment against CMIP6 and RECCAP2 observations.

Usage

# Basic usage
python scripts/validate_experiment.py xqhuc

# With custom base directory
python scripts/validate_experiment.py --expt xqhuc --base-dir ~/annual_mean

Requirements

  • Annual mean NetCDF files in ~/annual_mean/{expt}/
  • Observational data (automatically loaded from package data)

Outputs

Creates validation_outputs/single_val_{expt}/ containing:

validation_outputs/single_val_{expt}/
├── {expt}_metrics.csv              # UM results (obs format)
├── {expt}_bias_vs_cmip6.csv        # Bias statistics vs CMIP6
├── {expt}_bias_vs_reccap2.csv      # Bias statistics vs RECCAP2
├── comparison_summary.txt          # Text summary with performance comparison
└── plots/
    ├── GPP_three_way.png           # Three-way comparison plots
    ├── NPP_three_way.png
    ├── CVeg_three_way.png
    ├── CSoil_three_way.png
    ├── Tau_three_way.png
    ├── bias_heatmap_vs_cmip6.png   # Regional bias heatmaps
    ├── bias_heatmap_vs_reccap2.png
    └── *_timeseries_global.png     # Time series plots

What it does

  1. Computes UM metrics for all RECCAP2 regions (global + 11 regions)
  2. Loads observational data (CMIP6 ensemble and RECCAP2)
  3. Computes bias statistics (bias, bias %, RMSE, within uncertainty)
  4. Exports to CSV in standardized format matching obs/ files
  5. Creates visualizations:
    • Three-way comparisons (UM vs CMIP6 vs RECCAP2)
    • Regional bias heatmaps
    • Time series with observational uncertainty
  6. Generates text summary comparing UM vs CMIP6 performance

CSV Format

{expt}_metrics.csv: UM results in same format as observational CSV files

,global,North_America,South_America,Europe,Africa,...
GPP,134.81,18.40,31.65,6.11,30.02,...
NPP,68.29,10.36,14.45,3.42,14.45,...
...

{expt}bias_vs*.csv: Detailed bias statistics

metric,region,um_mean,obs_mean,bias,bias_percent,rmse,within_uncertainty
GPP,global,134.81,124.04,10.77,8.68,10.94,False
...