Accessible Statistical Model for Long-Range ENSO Forecasting

Forecasting the El Niño–Southern Oscillation (ENSO) remains a central challenge in seasonal climate prediction, with far‑reaching consequences for agriculture, water resources, and disaster preparedness. While dynamical models provide skillful forecasts, their reliance on supercomputing limits accessibility in many regions. This study introduces IndOzy‑LR, a parsimonious linear regression model that employs five lagged predictors of Niño 3.4 sea surface temperature anomalies to generate forecasts up to 11 seasonal leads. Results show that IndOzy‑LR achieves high accuracy at short horizons and retains useful skill through lead 8, performing competitively with a state‑of‑the‑art deep learning benchmark. The model further anticipates a neutral ENSO phase for 2025–2026, consistent with ensemble dynamical and statistical outlooks from the International Research Institute (IRI). These findings highlight the potential of low‑cost statistical approaches to extend forecast horizons, democratize access to seasonal climate prediction, and strengthen resilience in regions with limited forecasting infrastructure.