Evaluation of SEAS5 Precipitation Forecasts in the Central American Dry Corridor

Seasonal forecasts provide an opportunity to enhance drought preparedness in the Central American Dry Corridor (CADC). Evaluation of seasonal precipitation predictions within the CADC is important because rainfall affects many local livelihoods, and rainfall predictability in this region may be low due to its complex climate and terrain. In this presentation, SEAS5, a leading seasonal forecasting system produced by the European Centre for Medium-Range Weather Forecasting, is evaluated for the accuracy of its precipitation predictions across the CADC relative to the GPCC gridded precipitation dataset derived from rain-gauge data. A few studies have assessed its predecessor (System 4) in Central America but none have assessed SEAS5 in the CADC. SEAS5 predictions of rainfall mean, variability, and extremes are evaluated using one- to seven- month lead-times over 1981-2016 and for known historical droughts. Results show that SEAS5 precipitation forecasts often perform best during July and August, two important months for crop growth because they occur during the mid-summer dry period, which separates the wet season into distinct phases. Elevation seems to have an influence, although alone it does not explain variations in forecast skill across the region. SEAS5 overpredicts rainfall in greater quantities at high elevation. This analysis showcases promising forecast skill of relevance to agricultural forecasting and could be expanded on in future work by evaluating skill of other drought indicators.