What If Numerical Forecast Has No Skill? Using Emerging Rainfall Anomaly to Forecast State of Water Resources During Cape Town "Day Zero" Drought.

The 2015-2017 drought that triggered the Cape Town "Day Zero" crisis has illustrated the deficiency of numerical seasonal climate forecasting in the Southern Africa's winter rainfall region. A number of numerical forecasting systems failed to provide reliable information on rainfall anomalies to support defensible decisions to alleviate the emerging water crisis. As an alternative, we present a statistical rainfall forecast that is based on a relationship between current and end-of season anomaly of accumulated rainfall. In that forecast, the anomaly of accumulated rainfall, which is tracked as the rainy season progresses, is used to derive probabilities of total season's rainfall terciles. We illustrate the application of the method to forecast water availability in the Western Cape Water Supply System and how the results were used to communicate the need to maintain low water demand in the water shortage-stricken city. Furthermore, we apply the forecast over Southern Africa, map lead time at which signal emerges, and disaggregate sources of variance in total seasonal anomaly into lead-time dependent factors: "committed" anomaly, persistent anomaly, and random noise. These provide a basis for informed use of the method within the region as a tool for drought early warning.