Applications of interannual-to-decadal climate prediction: An exploratory discussion on rainfall in the Sahel region of Africa
This perspective explores how climate services may potentially incorporate information emerging from the new science of interannual-to-decadal (I2D) climate prediction in the Sahel region of West Africa.
Over the last decade, a number of international climate modelling centres have been developing new forecasts targeting a timescale up to about 10 years into the future, although most of the evaluation effort to date has tended to focus on forecasts for one-year-ahead, and forecasts for the period 2–5 years ahead. The source of the new skill lies in longer-lead climate model forecasts for interannual timescales (such as related to El Niño) and improved representation of key processes that drive some of the observed multi-year climate anomalies, such as the Atlantic Multidecadal Oscillation. A WMO Lead Centre for Annual-to-Decadal Climate Prediction has been designated, responsible for collecting and providing hindcasts, forecasts and verification data from contributing centres worldwide (https://bit.ly/2MUGJAA).
Our focus here on the Sahel region of West Africa (boreal summer monsoon rainfall) has two practical advantages for exploring the potential of climate services to draw on the new science. First, there are clearly demonstrated prediction advances for Sahel rainfall on the I2D timescale, which provides for discussion of applications in the context of clearly presented forecasting science. Second, the Sahel region is well-known to be one of the most vulnerable regions in the world to I2D climate fluctuations, such that modestly skillful information can be expected to be of interest to at least some stakeholders.
In terms of potential uptake of the information, the new longer-lead interannual forecasts have substantially similar issues to those of the existing seasonal forecasts: the new information is, in large part, simply an extension of lead-time. However, there are many novel technical and application dimensions that challenge the product design and user uptake of the emerging multi-year forecast information on 2–5 year timescales.
Stakeholder responses (explored during a national workshop in Sudan) suggest some interest in experimentation with operational probability forecasts of tercile-category (or other similar) summaries for regional indices of multi-year rainfall totals to explore interest in the topic, along the lines done initially with short-lead time seasonal forecasts. In addition, both national climate capacity and stakeholders emphasized the value of information in the dry season as well, especially temperature, but also notably wind speed. At least for temperature, good skill is anticipated, and so experimental tercile forecasts for temperature should be assessed. Longer-lead seasonal rainfall forecast potential for agriculture was clearly recognized: information available in March opens an array of new early-season management possibilities. However, the experimental and modest levels of skill suggest use may at this stage be best explored through national/regional strategies of stakeholders such as in the case of Sudan the World Food Programme. Potential applications of the new longer-term (multi-year) forecasts were especially identified in the water sector for effective infrastructure and operational management strategies at regional scales. Assessment of the planning applications for new forecast information on 2–5 year timescales is in its infancy and in locations where skill exists careful consideration of potential by user groups will be required. In terms of furthering discussions with stakeholders in Sudan, the potential of incorporation into indices for food security safety nets provided an example of a possible avenue.
Importantly, however, stakeholders noted that the current experimental multi-year product, a 2–5 year average forecast of seasonal rainfall (and temperature), could be misleading, and not give users what they need because there is more interest in information about individual years. This motivates approaches to temporally downscale the information, to generate information on, for example, the risk of an extreme drought, or run of droughts, within the forecast 2–5 year period. The level of skill of such temporally downscaled information is at this point not known and represents a topic for applied research. In some ways, this is analogous to the early challenge issued to seasonal forecasts on the need to downscale seasonal information to the statistics of the daily timescale that most impacted such activities as rain-fed crop production.
The demanding technical requirements and their experimental nature reinforce the need to consider the capacity of National Meteorological Agencies (NMAs) in many low income countries, including Sudan, to engage in this emerging area of forecasting. While infrastructure and capacity gaps and financial precarity are known to be important for many NMAs, recent extensive funding and initiatives are going some way to address these concerns, although progress varies considerably between countries. Beyond the technical issues are the demands on staff to co-produce products to make the most of their potential by adopting user-centred climate service design and recognizing the importance of interactions with wider stakeholder groups like non-state actors.
Among the technical challenges, a key concern was the importance of attribution of recent multi-year anomalies in the regional climate system, as a starting point for interpretation of multi-year forecasts. This was emphasized in the stakeholder discussions, and was expressed by the national meteorological community in Sudan as something that, with capacity support (to make progress on attribution), would assist greatly in their interaction with climate service customers. In terms of fully effective climate service development, this analysis in Sudan is an exploratory phase, drawing on research outputs still in experimental stages. It motivates assessment of multi-year climate predictability in terms closer to those needed for applications, and co-production research into the implications and potential for climate services drawing on the multi-year information.
Neil Ward, Declan Conway, Published in: Climate Services
Volume 18, April 2020.