Year-end Sale % OFF 
Abstract
Seasonal-to-interannual hindcasts (re-forecasts) for December-January-February (DJF) produced at a 1-month lead-time by the ECHAM4.5 atmospheric general circulation model (AGCM) are verified after calibrating model output to DJF rainfall at 94 districts across South Africa. The AGCM is forced with SST forecasts produced by (i) statistically predicted SSTs, and (ii) predicted SSTs from a dynamically coupled ocean-atmosphere model. The latter SST forecasts in turn consist of an ensemble mean of SST forecasts, and also by considering the individual ensemble members of the SST forecasts. Probabilistic hindcasts produced for two separate category thresholds are verified over a 24-year test period from 1978/79 to 2001/02 by investigating the various AGCM configurations’ attributes of discrimination (whether the forecasts are discernibly different given different outcomes) and reliability (whether the confidence communicated in the forecasts is appropriate). Deterministic hindcast skill is additionally calculated through a range of correlation estimates between hindcast and observed DJF rainfall. For both probabilistic and deterministic verification the hindcasts produced by forcing the AGCM with dynamically predicted SSTs attain higher skill levels than the AGCM forced with statistical SSTs. Moreover, ensemble mean SST forecasts lead to improved skill over forecasts that considered an ensemble distribution of SST forecasts.Keywords: AGCM, SST predictions, seasonal forecasting, South Africa
Highlights
The ECHAM4.5 forecast ensembles used in this study are obtained from the Data Library of the International Research Institute for Climate and Society (IRI) and not from the archives of South African Weather Service (SAWS), since the IRI archived data set of this atmospheric general circulation model (AGCM) is more comprehensive
Since AGCMs do not require the same amount of computing resources as coupled models, higher resolution, larger ensembles and longer lead-time forecasts can be produced by an AGCM
The generation of hindcasts to assess model performance and to calibrate model output are more achieved with an AGCM (Troccoli et al, 2008)
Summary
South Africa’s seasonal rainfall variability is associated with different levels of predictability and is dependent on the time of the year: Spring (September-October-November) season rainfall totals are for the most part not predicted with high confidence owing to the fact that this season is mostly influenced by transient weather systems, while the best forecast skill has been demonstrated during mid-summer (Landman et al, 2005; Landman et al, 2012) when the tropical atmosphere starts to dominate the atmospheric circulation over South Africa (e.g. Mason et al, 1996; Landman and Mason, 1999). Skill levels of the retro-active downscaled probabilistic forecasts of the three AGCM-MOS models are shown in Figs 2 and 3 (ROC scores and reliability diagrams, respectively).
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
