README.md

Climate Analysis

Overview

This package provides tools for analyzing and validating high-resolution climate data, including:

• Data loading from NetCDF files
• Preprocessing and aggregation
• Regridding and subsetting
• Optional DEM (digital elevation model) for altitude-based subsetting and binning
• Bias metric calculations
• Temporal aggregation (daily, monthly, yearly, seasonally)
• Weather type filtering (optional),
• Configurable CLI interface to orchestrate all steps.

Project Structure

• cli.py
  Main entry point, using argparse for command-line options. Pulls in configuration from a YAML file plus CLI overrides, then calls run_analysis.
• scripts/
  Contains modular Python scripts for subsetting, regridding, loading data, computing metrics, etc.
• tests/
  Pytest-based tests verifying the loading, subsetting, filtering, and bias calculations.
  • test_cli_ens_selection.py verifies ensemble member selection.
  • test_cli_weathertypes.py verifies weather-type filtering.

Installation

pip install .

Command-Line Usage

Run a complete analysis using a configuration file:

python cli/cli.py --config configs/config.yaml

Example Analysis Workflow

A typical analysis involves:

1. Loading ensemble and reference datasets
2. Subsetting by space and time
3. Altitude Binning

• We introduced a DEM-based workflow that attaches an "altitude" variable to both ensemble and reference data.
• bin_by_altitude function creates subsets of the data for user-defined altitude ranges.
• We can compute bias metrics (RMSE, MAE, ME) per altitude bin to study error dependence on terrain height.

4. Filtering by Weather Type using a CSV file that defines weather types for each date. This feature now supports automatically filtering both ensemble and reference datasets via the CLI (--weather-type-file, --include-weather-types 1 2 3, etc.).
5. Regridding ensemble data to match reference grid
6. Computing daily, monthly, seasonal, ... precipitation sums/means
7. Calculating bias metrics (e.g., RMSE)
8. Saving outputs (NetCDF, plots, statistics)

Configuration File

Users define parameters in a config.yaml file:

input:
  ensemble_pattern: "../data/total_precipitation_2017*.nc"
  reference_file: "../data/INCA_RR_20171*.nc"
subset:
  lat_bounds: [46, 48]
  lon_bounds: [10, 13]
  start_time: "2017-10-01T00:00:00"
  end_time: "2017-10-02T23:00:00"
output:
  save_netcdf: "output/bias_metrics.nc"
  save_plot: "output/rmse_map.png"

Requirements

• Python 3.7+
• numpy
• xarray
• matplotlib
• scipy
• dask
• netCDF4
• h5py
• pyyaml
• pandas

License

MIT License