Evaluating Seasonal Forecast Models for Cambodia’s Northern Tonle Sap Basin

Abstract

Accurate seasonal climate forecasts are vital for regions like Cambodia's Northern Tonle Sap Basin (NTSB), where agriculture is closely tied to rainfall patterns. While most studies have focused on the TSB, the northern areas, crucial contributors to Cambodia's national food basket, have remained largely unstudied. Here, this gap is addressed by evaluating the performance of 8 state-of-the-art seasonal forecast models from the Copernicus Climate Change Service (C3S) over a 24-year hindcast period (1993–2016). The evaluation is bolstered by ground-based data from 38 agrometeorological stations. Among the models, the Ensemble, the Japan Meteorological Agency (JMA) model, and the European Centre for Medium-Range Weather Forecasts (ECMWF) model emerged as top performers, with the Ensemble particularly excelling in replicating both temporal and spatial precipitation patterns, making it invaluable for agrometeorological applications. The Ensemble demonstrates particularly strong performance in regions such as western Oddar Meanchey and eastern Preah Vihear, where biases are less than 5%. To tailor the Ensemble to the specific climatic and geographic context of the NTSB, we refined it using the Delta Change technique, and this reduced biases even further to < 1%. Our study not only contributes to improving the precision of agrometeorological advisories in a key, but under-researched region, but also sets a precedent for how regional climate forecasting can be enhanced through context-specific model evaluations and corrections. These findings provide a practical framework for supporting resilient agricultural strategies in areas vulnerable to climate change, bridging a critical gap between climate science and agricultural practice.