Special Report: El Nino and Indian Ocean Dipole conditions likely into early 2007, with drought and flooding implications for Southern and Eastern Africa
Introduction to El Niño and the Indian Ocean Dipole: What are they doing and why are they important?
Over the past several months most model forecasts have trended towards warmer conditions in the tropical Pacific through October and November. The National Center for Environmental Prediction (NCEP) predicts El Niño conditions for the remainder of 2006 and into January and February of 2007. Typical El Niño effects are likely during November to March, including drier-than-average conditions over southeastern Africa, and wetter-than-average conditions over East Africa. In addition to El Niño related warming, the Indian Ocean is also undergoing a moderate positive Indian Ocean dipole event, with warmer-than-normal waters in the west and colder-than-normal waters in the east. This pattern is also associated with more rain in eastern Africa and less rain in the south.
Implications of El Niño and the Indian Ocean Dipole
While a poor 2006-2007 rainy season in southern Africa is not definite, the development of an El Niño increases the potential for poor rains and drought development over the region during this period. Areas most likely to experience drought under these conditions are central and southern Zimbabwe, adjacent parts of South Africa, Mozambique and Botswana, as well as northern Namibia. These areas may start the season with average to above-average rains, only to see them temporarily stop, or become erratic later in the season. Therefore, month-to-month variability of the rains will be important, and rainfall distribution as well as rainfall totals need to be monitored closely as the season progresses. Unfortunately, the state of the science of climate prediction does not allow us to confidently forecast the month-to-month evolution of the upcoming season, and the presence of a drought may not become known until late December or mid-January. In East Africa, the development of an El Niño increases the odds for above-normal rainfall around the Lake Victoria Basin. This would favor crop and pasture development, but can also cause flooding.
In the Indian Ocean, water temperatures are cooler than normal off the west coasts of Sumatra and Java, and are warmer than normal across the central and western parts of the basin. This phenomenon, known as the Indian Ocean Dipole (or IOD), occurs less frequently than El Nino, and is associated with wet conditions across East Africa during the October to December rainy season. Already, normal to above-normal rains have fallen over the pastoral areas of Africa’s Greater Horn. These rains have boosted moisture supplies for crops and pastures, but have also triggered significant floods in portions of southeastern Ethiopia and Somalia, as well as localized flooding in Kenya. The current state of the El Niño and the IOD increases the odds for continued wet conditions during the next few months in eastern Africa. However, continued wet conditions are not guaranteed. Local dryness and drought also remain possible.
Enhancing Statistical Forecasts
To address the need for more specific climate forecasts, FEWS NET scientists have derived statistical forecasts based on October climate conditions over the Indian Ocean, Africa and the tropical Pacific. These forecasts employ the same technique that identified emerging drought conditions in southern Africa in November of 2002. A technical description of the data and methodology has been posted as an extended abstract on the American Meteorological Society website(1).
These forecasts use statistical relationships between October climate fields and mid-season rainfall to make predictions. These predictions are expressed as standardized precipitation index (SPI) values. The models have standard errors of between 0.6 and 0.7 SPI. While useful, this level of accuracy still encompasses a broad range of potential outcomes. The November-December-January 2006/07 forecasts are shown in Fig. 1.
The forecasts are probabilistic, describing a range of outcomes. The central panel shows the most likely SPI distribution. The left and right panels display likely dry (left panel) and wet (right panel) outcomes. There is a 68 percent chance that the observed SPI values will fall between these results. There is a 32 percent chance the observed SPI values will exceed the left and right quantities. Blue shades denote above normal rainfall conditions. Red hues imply rainfall deficits.
For the Greater Horn, normal to above-normal conditions appear likely, while normal to below-normal conditions seem likely for southern Africa, with northeastern South Africa, eastern Botswana and southern Zimbabwe being the most likely to see below normal rains.
Figure 2 shows the observed and forecast regional November-December-January time series SPI for eastern southern Africa and the northeastern Greater Horn. Model fidelity is reasonable over the 1979-2005 test period, with R2 values of 0.71 and 0.76. The gray boxes cover the 68 percent confidence interval for the 2006/07 forecasts. The forecast for southern Africa is -0.8 ± 0.6 SPI for eastern southern Africa and +0.5±0.5 SPI for the northeastern Horn. Note that there is 1 in 3 (32 percent) chance that observed rainfall totals will fall outside these ranges. Also shown are the co-occurring El Niño-IOD events (nidos) in 1982/83 and 1997/98. The range of outcomes for these years converges on a similar range of scenarios: normal to dry for southern Africa and normal to wet for the Greater Horn.
While cautious use should be made of this information, there does appear to be solid reasons for drought concerns in Southern Africa, especially within the red regions of the right panel (wet scenario) of Figure 1. Extreme rainfall in eastern Africa is also possible, although probably harder to predict, and less certain given the forecast results.
Note: (1) http://ams.confex.com/ams/pdfpapers/119819.pdf, A PowerPoint with individual climate fields and correlations for Southern Africa WRSI is at: ftp://hollywood.geog.ucsb.edu/pub/EvolvingOctoberAtNov5.ppt