A Dynamical One-Month Lead Seasonal Rainfall Prediction for October to December 1998 Short Rains over East Africa

contributed by Tony Evans¹, Richard Graham¹, Mike Harrison¹, Mike Davey²

and Andrew Colman²

¹NWP Division, ²Ocean Applications Branch UK Meteorological Office, Bracknell, United Kingdom.

A real-time seasonal forecast at one-month lead is provided for the East African region. This was a contributing forecast to the Greater Horn of Africa Regional Climate Outlook Forum (GHARCOF) held in September 1998, convened to formulate consensus guidance for the Short Rains season. Further information on GHARCOF is available on the World Wide Web from the Drought Monitoring Centre, Nairobi, Kenya at: http://www.meteo.go.ke/dmc/forum98/mombasa/ and the International Research Institute at: http://iri.ucsd.edu/forecast/sup/text/sep98/Mombasa.html

A verification of the UKMO forecast for October to December 1997 issued in last September's Bulletin is also presented. 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 East Africa between 10^oN and 10^oS.

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.

Correlations between October to December ensemble mean rainfall anomalies and observed anomaly values over 15 years, obtained from the gridded observed land-surface rainfall data set of Hulme (1994), have been calculated for three 10^o latitude bands between 10^oN and 10^oS, bound by longitudes 28^oE and 43^oE (Fig. 1). The highest correlation (0.52) occurs for Band (B), centred on Tanzania. For the northern Band (A), covering southern Sudan and Ethiopia, northern Kenya and Uganda and the north-eastern Democratic Republic of Congo, the correlation is 0.32. The lowest correlation of 0.12 is indicated for southernmost Band (C). The observed data set is gridded to the same resolution as the model (2.5^olatitude x 3.75^o longitude), but in order to reduce noise 4^ox4^o blocks have been joined together to produce Fig.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 dynamical 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. In most years, the observed anomaly lies within (or is captured by) the range of the inflated ensemble, or is a close outlier to the range (Fig. 2). The Capture Rates (CR), out of 15, for areas A,B and C are 13,11 and 12 respectively, as against the a priori expected value of 12 for a correctly-formulated ensemble.

In a first practical test of the level of forecast skill available, a verification is presented of the October to December 1997 rainfall forecast, which used the same methodology as this year's forecast and was issued in last September's Long Lead Bulletin (Graham et al., 1997). Observed anomalies are obtained from the CPC Merged Analysis of Precipitation (CMAP - Xie and Arkin, 1997) dataset, a combination of satellite- and raingauge-derived data. The ensemble mean forecast and observed anomalies are indicated in Figure 3 (a) and (b). The ensemble mean provided a reasonable prediction of general wet conditions in the north and near-normal conditions in the south, however the western parts of regions (A) and (B) were drier than predicted. The area-averaged observations for regions (A) and (C) is captured within the inflated range of the ensemble and is a close outlier for area (B) (Fig. 2). Futhermore, areas (A) and (B) were forecast (with relatively high confidence) to have a wet season, and observed anomalies were 166 and 173 % of normal. In area (C), the area of lowest skill, the forecast was for dry or normal conditions (with low confidence) and indeed observations were close to normal.

Forecasts for the 1998 October to December season are derived from nine-member ensemble runs, but with the difference from the PROVOST runs that persisted SST anomalies (from August), rather than observed values, are used throughout. It is thought 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 autumn. 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. 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 August anomalies. In summary, there is likely to be below-average rainfall during October to December 1998 across the south of the region and near-normal in the north (Fig. 4 a) and b) for probabilities and ensemble mean respectively). The prediction for dry conditions is supported by a statistical forecast produced by Mutai, Ward and Colman (1998). Ensemble mean, together with maximum and minimum, rainfall anomalies, for each of the gridded areas depicted in Fig.1, are listed in Table 1 and included at the extreme right of Figs 2 a)-c). All nine members indicate dry conditions in regions B and C, with six out of nine predicting dry conditions in region (A).

References

Evans, A. et al., 1998: A dynamical one-month lead seasonal rainfall prediction for March to May 1998 for the north-eastern area of South America, Experimental Long-Lead Bulletin, COLA,7(1), March 1998.

Graham, R et al., 1997: A dynamical One-Month Lead Seasonal Rainfall Prediction for Oct-Nov-Dec 1997 Short Rains over Equatorial East Africa, Experimental Long-Lead Bulletin, NOAA, 6(3), September 1997.

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.

Mutai, C.C., Ward, M.N., and Colman, A.W., 1998: Towards the Prediction of the East African Short Rains Based On Sea-Surface Temperature-Atmosphere Coupling. International J. of Climatology Vol. 18, 975-997.

Xie, P. and Arkin, P.A., 1997. Global Precipitation: a 17-year monthly analysis based on gauge observations, satellite estimates and numerical model outputs. Bulletin of the AmericanMeteorological Society Vol 78, 2539-2558.

Forecast %	A	B	C
E Mean	98	87	86
I Mean	95	72	33
I Wettest	112	63	64
I Driest	73	45	14
Number of runs <100%	6	9	9

Table 1. October to December 1998 seasonal forecast rainfall percentages of normal for the unmodified Ensemble Mean (E Mean - with respect to the model 1979-1993 climate), inflated ensemble mean (I Mean), the highest and lowest (inflated) ensemble members, and the number of members (out of 9) giving less than 100% of normal rainfall for each of the 3 areas depicted in Fig. 1.

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

Figure 2a, 2b, 2c. Time series of Ensemble Mean (EM) rainfall (as % of normal) for October to December 1979-1993 post-inflation of ensembles created using observed SST and of the Hulme dataset for selected representative are as (see Fig. 1). 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, and CORR indicates correlations between ensemble mean forecast and observed anomalies. The forecasts for 1997 and 1998 and observation for 1997 are also indicated. The "observation" (a combination of satellite-and rainguage-derived data) is obtained from the CPC Merged analysis of precipitation (CMAP-Xie and Arkin, 1997) relative to 1979-1993 base period.

Figure 3. (A) Non-inflated ensemble mean precipitation anomaly forecast and (b) Observed CPC Merged Analyses of Precipitation anomalies for October to December 1997. Negative contours dashed. Variable Contour interval in mm/day.

Figure 4. October to December 1998 forecast for: a) Probabilities of rainfall exceeding normals (in%), calculated from the proportion of ensemble members with above average predictions; and, b) Non-inflated ensemble mean forecast rainfall anomalies. Variable contour interval in mm/day.