Pyenspp

A Python package for ensemble precipitation forecast post processing.

• Created by Fuxuan Jiang
• Free software: MIT License
• Current Version: v0.2.0 (Updated May 2026)

Features

• Pyenspp is a Python package for post-processing of ensemble precipitation forecast.
• Pyenspp implements several calibration methods, including Quantile Mapping (QM), KAN-CSGD, and Censored Shifted Gamma Ensemble Model Output Statistics (CSG-EMOS).
• The verification module provides more than 20 metrics including probabilistic (ensemble) metrics and deterministic metrics.
• The package was evaluated using the ECMWF Sub-seasonal to Seasonal (S2S) dataset over the North River catchment. The case study demonstrates that the KAN–CSGD model implemented in Pyenspp significantly improves forecast reliability and accuracy.
• Example data and scripts are provided in the examples/ directory to demonstrate the basic usage of Pyenspp.
• Designed with modularity in mind, Pyenspp provides a flexible foundation for future enhancements, and we welcome community-driven improvements.

Development

To set up for local development:

# Clone your fork
git clone https://github.com/Curallin/Pyenspp.git
cd pyenspp

# Install in editable mode with live updates
pip install -e .

This installs the CLI globally but with live updates - any changes you make to the source code are immediately available when you run pyenspp.

Version

1. first public release (v0.1.0, Feb 2026)

Pyenspp (v0.1.0) provides the core framework for ensemble precipitation forecast post-processing, including preprocessing, calibration, dependence reconstruction, and verification.

As the initial release, the package focuses on establishing a modular and extensible foundation for future development.

2. second public release (0.2.0, May 2026)

Added a CSG-EMOS calibration method with SCE-UA for parameter optimization, together with sample data and reproducible code examples in examples/6.csgemos_usage.ipynb.

Roadmap

Future versions of Pyenspp are expected to include:

• Additional post-processing models such as meta-Gaussian model (MGB) and distributional regression networks (DRNs).
• Enhanced visualization and diagnostic tools for forecast assessment.

Author

Pyenspp was created in 2026 by Fuxuan Jiang.