West African Seasonal Climate Variability and Predictability.

摘要

The emerging hypotheses over the last two decades relating to the complex ocean-atmosphere coupled systems dynamics over the Atlantic and their teleconnectivity have been a focal point in continued efforts to elucidate further the interactions governing the high West African Sahel climate variability. Pursuant to this, a polymorphous ocean-atmosphere phenomenological model has been developed to investigate the Western Sahel seasonal (July-September; JAS) climate variability in 2008 using predictors over the period 1950-2008, primarily along three lines---causality inference, intermediary pathways and the generalized circulation patterns associated with them, but with respect to an extratropical North Atlantic Oscillation (NAO) hypothesis. It is prototypic in the sense that the integrated approach provides a solid diagnosis in its optimal detection, discrimination, extraction and combination of key ocean-atmosphere drivers of the region's climate operating under three independent timescales---monthly, bimonthly, and seasonal, into a consolidated form that enables a complete overhauling of the region's climate. Generally, it is more robust than the unconsolidated models, adjudged by the R² and Environmental Causality Impact (ECI) score statistics, using heterogeneous datasets that include NCEP/NCAR reanalysis, NOAA extended reconstructed sea surface temperature (ERSST) and CRU precipitation. The model comprises three main architectural components---Least Absolute Shrinkage and Selection Operator (LASSO) model and ECI Analysis Model (ECIAM), and Complex Coupled Association Rules Model (CCARM). While the study does not rule out anthropogenic forcing, vegetation dynamics, and various forms of feedbacks, it suggests important clues for addressing the dichotomous Sahel climate change projected by numerical climate models. This drive is investigated with the hypothesis that the disharmony between the models may be linked to their inability to capture the correct sign, or path, of the NAO forcing, as it modulates the low-level westerlies (LLWs) in the vicinity of the equatorial Atlantic modes. The driving hypothesis is validated based on five canonical metrics---support, confidence, p-value, ECI and R² statistics, as well as on dynamical consistency of the plausible pathways, comparing the Sahel Region to the Western Sahel. The standardized, generalized 200 hPa circulation, SST and precipitation regression composites over the Sahel Region and Western Sahel relative to the bimonthly (Dec/Jan) and monthly (Apr) NAO detection in the annual cycle for the former and latter but over the same timeframe are distinct, characterized by anomalous wet and dry conditions, respectively. The impacts associated with their pathways drop from R² skill scores of 0.67 and 0.71 over the Sahel Region and Western Sahel, respectively, with the NAO, to 0.50 and 0.65, without it, substantiating its extratropical forcing. Perhaps, this mechanism may be part of the bunch of disparities confronting the numerical models---one set captures a different NAO forcing over the Sahel Region, and another captures a different NAO forcing over the Western Sahel and represents it for the Sahel Region, producing bifurcated climates.;The model has catalogued the best set of predictors from an initial pool of 73 for investigating the predictability of the region's climate. Multi-year surface relative humidity (RH)---a meningitis-proxy, ensemble forecasts for April 1987-2010 over Ghana have subsequently been produced using just the monthly timescale predictors, up to a 12-month lead time---based on one, five, and ten-year moving window forecasts, using the FORECASTER Model. The results show that the mean accuracy ranges between 0.6-0.8, with the five and ten-year forecasts generally being more skillful than the one year. Physically, this may be ascribed to better predictability inherent in the semi-decadal to decadal predictors than in the interannual predictors. Strategic logistical planning may be based on these outcomes.