Regional Drought Forecasting in CARICOM
A statistical downscaling climate prediction software package, Mason (2011) has been used to provide seasonal tercile rainfall forecasts since 2012. A factor limiting the effective utilization of tercile-based forecast information as a prediction component in drought early warning is that precipitation forecasts are expressed in probabilities of normal, above-normal, and below-nor- mal rainfall, which may be difficult to effectively integrate in sector-based decision-making processes. Although these forecasts can provide an initial indication of pending periods of unusual dryness or wetness, they can neither distinguish unusual from extreme rainfall totals nor ensure sufficient certainty on an extreme outcome, such as rainfall deficits serious enough to cause severe drought.
The first regional drought forecasting system that extended the capabilities of the CPDMN and utilized the CPT was presented at the Wet Season CariCOF held in Jamaica in May 2014 (http://rcc.cimh.edu.bb/long-range- forecasts/caricof-climate-outlooks/). Using this system, forecasts are updated each month for moving 6-month periods (Figure 20.3a), as well as for 12-month periods (Figure 20.3b) ending at the end of the wet season or dry season that the CariCOF calls the hydrological year. Through this
(a) Drought forecast covering the 6-month period December 2015 to May 2016 and (b) the drought forecast for the hydrological year (June 2015 to May 2016) released at the end of February 2015. The forecasts indicate the impact that is likely to be felt from any deficits in rainfall for the time periods considered.
approach, which adds observed precipitation totals at the start of the period to more uncertain, probabilistic precipitation forecasts for the remainder of the period, the CariCOF drought alert outlooks provide more confidence than the probabilistic precipitation forecasts themselves. This approach enables the identification of 88 percent of observed rainfall deficit levels consistent with severe long-term hydrological drought by the end of the dry season, for example, as early as November (i.e., with a 6-month lead time).
The forecast information included in the CariCOF drought outlooks contributes to a seasonal decision-support system based on alert levels that are tied to increasing probabilities of crossing specific SPI thresholds—with specific but different SPI thresholds being established for the wet and dry seasons. This approach is used to construct drought alert maps with actions corresponding to each alert level following extensive sectoral stakeholder engagement.
Communicating early warning information from the operational drought monitors and forecasts to sectoral stakeholders occurs through packaged products disseminated via the monthly Caribbean drought bulletin (http:// rcc.cimh.edu.bb/climate-bulletins/drought-bulletin/) and discussed at the CariCOF.