One objective of the European PROVOST experiment (PRediction Of climate Variations On Seasonal and interannual Timescales) is to estimate potential dynamical seasonal predictability given ideal surface boundary conditions on a global scale. To this end three European models (the UKMO Unified Model at climate resolution, the ECMWF T63 model and the ARPEGE model run at T42 by Meteo-France and T63 by EDF) have been integrated in 9-member ensembles initialized at 24-hour intervals for four months for each season over 15 years from 1979 to 1993. Common initial conditions and verifying analyses obtained from the ECMWF reanalysis as well as common SST anomalies from the UKMO GISST and Reynold's OI data sets were used in all experiments. All initializations were at 0000Z finishing on the day prior to the start of the season. Useful levels of skill appear to exist in these PROVOST simulations for most of the region over Africa between the Equator and 20 N and in this paper a real-time seasonal forecast with the UKMO Unified Model (UM) is provided at one month lead for the region based on these assessments.

An updated version of the Unified Model has been used in producing this year's forecast. Full evaluations of the skill (on seasonal time-scales) and climatology of the new model are currently being produced. However, in this paper, model climatology used to generate anomalies is obtained over the period 1979 to 1993 from an AMIP (Atmospheric Model Intercomparison Project) integration of the updated model.

Within the area of Africa bound by the Equator and the 20 N line of latitude, area correlations from the PROVOST data over the 15 years between ensemble-mean rainfall anomalies and anomaly values obtained from the gridded observed land-surface rainfall data set of Hulme (1994) vary between 0.18 and 0.67, with lowest correlations for the area of northern Kenya and Uganda and highest correlations over eastern Niger, western Chad and central Sudan (Fig. 1). The observed data set is gridded to the same resolution as the model (2.5 x 3.75 degrees), but in order to reduce noise 4x4 blocks have been joined together to produce Figure 1; only blocks with adequate data were retained. Time series of ensemble mean rainfall and the Hulme data for selected regions illustrate the fact that the dynamic model, although able to capture the interannual variability reasonably well, has insufficient variability (Fig. 2). Hence a variance inflation has been calculated using both ensemble means and ensemble members for each gridded area and applied appropriately to the PROVOST simulations from both the members and the ensemble means. Note that observed anomalies frequently lie within the ranges of the inflated ensembles or are close outliers to those ranges; capture rates vary between 5 (Area A in Figure 1) and 15 (Area C).

Correlations have also been calculated for three standard areas of the region (Fig. 3) as used in past UKMO predictions (regions 2, 3 and 4 in Colman et al., 1997 and this volume). The regions are defined in terms of the model grid and observed values have been generated from the Hulme data set. Correlations are lowest over the northern Sahel region (0.51), but exceed 0.6 in the central Soudan and southern Guinea Coast regions. Capture rates achieved after ensemble variance inflation (Fig. 2) vary between 10 (Guinea Coast) and 15 (Soudan).

Forecasts as produced for the 1998 July to September season are derived from nine-member ensemble runs, but with the difference from the PROVOST runs is that persisted SST anomalies (from May), rather than observed values, are used throughout. It is unlikely that the use of persisted anomalies will have a significant negative impact if results obtained from experiments for twelve winter and spring seasons, carried out with persisted anomalies, can be extrapolated to the summer. While there is some inevitable loss of predictability associated with the use of persisted anomalies this appears to be minimal in areas of relatively high predictability such as considered here, and certainly does not eliminate predictability in terms of the levels normally associated with seasonal forecasts (see Evans et al., 1998). Use of persisted anomalies fails, of course, during seasons in which there is a substantial readjustment of SST anomalies over ocean areas related to a given region's rainfall; experience has been gained of such failed forecasts for the Sahel in preliminary work with the model. Currently there is no solution to this problem of rapid intraseasonal SST anomaly distribution changes: the forecasts given below are conditional on the continuity of the May anomalies.

The ensemble mean provides a consistent prediction of above-average rainfall during July to September 1998 across the south of the region and near-normal in the north (Fig. 4 and Table 1). An increase in convection along the ITCZ from its model climatological average is suggested. Ensemble mean, together with maximum and minimum, rainfall anomalies, for each of the gridded areas depicted in Figure 1, plus for the three standard African rainfall indices, are listed in Table 1. There is agreement between ensemble and statistical model predictions for above-normal rainfall in the Guinea coast region but disagreement in the Soudan and Sahel regions where statistical methods predict below normal rainfall (relative to 1971-90 climatology), whereas the ensemble gives near normal (cf. Colman et al., 1998).

References:

Colman, A. et al., 1997: Multiple regression and discriminant analysis predictions of Jul- Aug-Sep 1997 rainfall in the Sahel and other tropical North African regions, Experimental Long-Lead Bulletin, NOAA, June 1997, 33-35.

Colman, A. et al., 1998: Multiple regression and discriminant analysis predictions of Jul- Aug-Sep 1998 rainfall in the Sahel and other tropical North African regions, Experimental Long-Lead Forecast Bulletin, COLA, June 1998, this volume.

Evans, A. et al., 1998: A dynamical one-month lead seasonal rainfall prediction for March to May 1997 for the north-eastern area of South America, Experimental Long-Lead Forecast Bulletin, COLA, March 1998.

Hulme, M., 1994: Validation of large-scale precipitation fields in general circulation models. Global Precipitation and Climate Change, M. Desbois and F. Desalmand, Eds., NATO ASI Series, Vol. 23, Springer-Verlag, 387-406.

Figure captions:

Figure 1. Correlations for simulations over July to September 1979-1993 between ensemble mean rainfall and the Hulme gridded rainfall set over 10 x 15 degree blocks.

Figure 2. Time series (area D, area F, area 2, area 4) of Ensemble Mean (EM) rainfall (as % of normal) pre- and post-inflation of ensembles created using observed SST and of the Hulme dataset for selected representative areas (see Figs 1 and 3). Bars indicate inflated range of ensembles and the median. Capture rate (CR) indicates the number of years out of 15 when the observation lies within the range of the inflated ensemble.

Figure 3. Correlations for simulations over July to September 1979-93 between ensemble mean rainfall and the Hulme gridded rainfall set over the areas shown for Sahel (labelled 2), Soudan (labelled 3) and Guinea coast (labelled 4).

Figure 4. Non-inflated ensemble mean precipitation forecast anomalies for July to September 1998. Negative contours dashed. Variable contour interval in mm/day.