idownscale

Project Context

The IRISCC project provides high-quality, fine-resolution (10 km) climate projection data downscaled from GCM simulations. This repository contains the tools for preprocessing, bias correction, training, and evaluation of DL downscaling models.

---

📖 Documentation

Full documentation is available in the docs/ directory.

• Release Notes: Latest stabilization and scientific fixes.
• Getting Started: Installation and directory structure.
• Workflow Management: Running the run_exp5_full.sh pipeline with automated integrity checks.
• Evaluation: Detailed metrics and visualization protocols.

---

⚡ Quick Start

Installation & Workspace Setup

conda activate idownscale_env
./bin/utils/setup_workspace.sh # Configures absolute paths for your environment

Running the Experiment 5 Pipeline

The main entry point for the full automated workflow is run_exp5_full.sh.

# Run the full pipeline with structured logging and automated validation
./run_exp5_full.sh

All process logs are stored in logs/exp5/<TIMESTAMP>/. Final scientific validation artifacts (plots, CSVs, PDF reports) are consolidated in output/exp5/validation/.

---

🏗 Project Structure

• bin/: CLI scripts and automated validation utilities.
• docs/: Sphinx documentation (RST).
• iriscc/: Core library (100% compliant with Ruff and scientific standards).
• logs/: (Ignored) Structured execution logs for debugging.
• output/: Results and centralized validation artifacts.
• tests/: Automated unit tests (integrated with Pytest).