WAS-NextGen: An Objective Multi-Method Framework for Seasonal Climate Forecasting in West Africa

Seasonal forecasting
Machine learning
Climate services
West Africa & Sahel
Open source
WMO

Invited talk in the WMO Artificial Intelligence Webinar Series — “Good practices of KNUST and AGRHYMET”.

Authors
Affiliation

Coovi Mahuwètin Mandela Houngnibo

AGRHYMET Regional Climate Centre – West Africa & Sahel (RCC-WAS), Niamey, Niger

Abdou Ali

AGRHYMET Regional Climate Centre – West Africa & Sahel (RCC-WAS), Niamey, Niger

Published

April 27, 2026

About this talk

Presented in the WMO Artificial Intelligence Webinar Series (session on the good practices of KNUST and AGRHYMET), this talk introduces WAS-NextGen, an objective, reproducible framework for seasonal climate forecasting in West Africa and the Sahel, implemented through the open-source Python package wass2s.

Watch the recording

The WAS-NextGen segment begins at 29:38 (the KNUST presentation opens the session).

Slides

Download the slides (PDF)

Abstract

Seasonal climate forecasting in West Africa has traditionally relied on consensus-based methods that lack reproducibility and high-resolution detail. This presentation introduces WAS-NextGen, an automated, multi-method framework that integrates machine-learning-calibrated multi-model ensembles (SV–ML–CMME), statistical–dynamical CCA calibration, lagged-predictor components, and analogue-year methods. Implemented via the open-source Python package wass2s, the system ensures a fully reproducible workflow from data acquisition to probabilistic mapping.

Highlights

  • From subjective to objective. Replaces consensus-based forecasting with an automated, traceable, skill-assessed pipeline, aligned with WMO guidance for objective and operational seasonal forecasting.
  • One reproducible workflow. End-to-end and modular: data acquisition → preprocessing and bias correction → model training with leakage-free cross-validation → verification → multi-model ensemble → probabilistic tercile maps.
  • Multiple forecasting engines, combined objectively. ML-calibrated multi-model ensembles (SV–ML–CMME), statistical–dynamical CCA calibration, lagged-predictor models, and analogue-year methods — behind a common interface.
  • Open source and operational. Built on the wass2s Python package, supporting national meteorological and hydrological services and capacity building across the region.

Resources

How to cite

@misc{houngnibo_wass2s,
  author = {Houngnibo, Coovi Mahuw{\`e}tin Mandela and Segnon, Alcade
            and Tonle, Franck and Kiema, Ars{\`e}ne W. and Ali, Abdou
            and Sounouke, Valerie H. and Zougmor{\'e}, Robert},
  title  = {{wass2s}: An Open-Source Python Tool for Objective Seasonal
            Climate Forecasting in West Africa and the Sahel},
  year   = {2026},
  note   = {SSRN preprint},
  url    = {https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6569116}
}

Acknowledgements

Developed at the AGRHYMET Regional Climate Centre for West Africa and the Sahel (RCC-WAS), with support from the Accelerating Impacts of CGIAR Climate Research for Africa (AICCRA) project.

Back to top