The IRI Becomes The International Research Institute for Climate and Society

In October 2005, the Trustees of Columbia University changed the name of the International Research Institute for Climate Prediction to better reflect the work of the Institute. The IRI develops and implements strategies to reduce society’s vulnerability to climate risk. It partners with local institutions to best understand needs, risks and possibilities. The IRI supports sustainable development by bringing the best science to bear on managing climate risks in such sectors as agriculture, food security, water resources, and health. By providing practical advancements that enable better management of climate related risks and opportunities in the present, Columbia’s IRI creates solutions that will help to increase adaptability to long term climate change.

“The IRI was created from within the climate community roughly a decade ago,” Director-General Steve Zebiak explains. “We all wanted better climate information, but we felt also that this knowledge had to connect to society, and we really didn’t have much of a clue about how that was going to get done.” The process of transforming a fledgling IRI, which was heavy on climate, to a more mature one that is actually knowledgeable in the social sciences and the other dimensions of climate impact, meant engaging and training a cadre of international scientists empowered with a wide range of scientific expertise, much of which, at first glance, might seem only vaguely connected to climate prediction. The main areas of concentration that IRI follows today are prediction, monitoring and the impact of climate on society, leading to decision-making and policies more in tune with climate realities and geared to take advantage of current climate science. Operations are concentrated in Asia and the Pacific, Latin America and the Caribbean, and Africa.

Possibly the most revolutionary idea the IRI is trying to get across is that probabilistic information on the state of future seasonal climate is better information than deterministic forecasts. The initial reaction to seasonal climate forecasts is to seek deterministic predictions; government policy planners and farmers want definite answers: either it will rain or it won’t. Although such predictions can be given, they are often wrong. “The IRI approach is closer to the odds in a Las Vegas poker game,” says Tony Barnston, who heads up forecast operations at the IRI. “Certainty to a climate scientist is wooly uncertainty to anyone else,” counters Madeleine Thomson, who directs the Impacts research program at IRI, mapping climate effects on life in climate-affected areas. Instead of a “yes” or “no” approach to critical weather questions, IRI proposes a calculated “maybe.” In other words, there may be a 70% chance that drought conditions will prevail in the first few weeks of an upcoming planting season, but there is still a 30% chance that there will be rain. Policy makers need to hedge their bets by being aware of both eventualities. In a typical example, a farmer, not knowing what to expect, might plant seeds after the first seasonal rainfall, only to have the plants dry up and die when no subsequent rains follow. With probabilistic information, a farmer would have the option of planting only part of his seeds after the first rainfall, or spending the extra money to invest in seeds that take longer to mature and need less water. The farmer would in a sense be gambling, and he would still need to prepare for the unexpected. Hedging one’s bets is the key to the climate game.

Once the notion of probability has been inculcated, IRI has developed, or is in the process of developing a series of innovative tools designed to enable government policy planners and decision makers to make educated guesses about the future. These tools are works in progress, constantly being improved to aid policy planners and irrigation and agricultural organizations to visualize scenarios that are likely to follow specific actions in different climate situations.

The Philippine’s Angat reservoir is the focus of one of the exciting projects currently being developed. This single reservoir provides 97% of Manila’s drinking water and 65% of the hydroelectric power for Luzon, as well as irrigation for over 30,000 hectares of land. During the 1997-98 El Niño, rainfall was reduced so drastically that irrigation had to be stopped and the hydroelectric turbines had to be shut down. The Philippine government desperately scrambled to buy coal at elevated prices from Australia and Indonesia at the last minute in order to keep generating electric power. To make matters worse, in late September 1998, the El Niño gave way to La Niña conditions resulting in above normal rainfall and the reservoir level rose so rapidly that water had to be released to save the dam. Lowland farmers who had been starved for irrigation throughout 1998, stood by and watched as their crops washed away in the flood. “It was a shock to the system, and it perhaps could have been avoided if they had been using available climate information,” says Shiv Someshwar, who heads IRI operations in the Asia and Pacific region, and also directs the research program on Institutions and Public Policy.

In fact, there is a close correlation between the El Niño and the flow of water into the Angat basin. 50% of the reservoir’s water is supplied by rains during October, November and December. Angat, which Someshwar describes as a very well-run reservoir, had tried to base its water strategy by averaging out historical records, which were unable to predict the unexpected variations likely to result from an unexpectedly strong El Niño. In contrast, if the reservoir management knows in advance that rainfall is going to be heavy, it can release water for irrigation early on.

Someshwar stresses that IRI makes its contributions through partnerships with established regional institutions, such as the Philippine National Water Resource Board and the Philippine Atmospheric, Geophysical and Astronomical Services Administration in the case of Angat. Someshwar notes that it was necessary early on to caution against illusions about infallibility in climate prediction. “You can’t put all your eggs in one basket,” he says. “What we do not want is for them to use the seasonal forecast in a deterministic way. Even in 1998, there was a 30% chance that the weather could have gone the other way. The last thing you want to do is to oversell the power of science, without a full understanding of the limitations.”

A large part of Someshwar’s job in Asia has been to find converts to climate prediction who are ready to become champions of the idea within central and local administrations. Getting officials used to making decisions based on probability was less than evident. “Most government institutions around the world are not structured to take risks,” says Someshwar. But, in the Philippines, where detailed records have been kept for decades, a convincing argument is provided by “hind casting.” Historical data going back 30 years or more can be run through computer models and officials can see what would have happened if earlier planners had access to modern climate tools. In some cases, it is simply a question of not failing to take advantage of an excellent harvest year. “It’s not only managing negatives,” Someshwar explains, “It is also capitalizing on and managing the positive.” Besides the Angat project, IRI is also developing reservoir management software for Kenya’s Tana River basin, which provides 70% of Kenya’s electricity, and in northeastern Brazil’s generally arid state of Cear”.

Now that the powers of available climate information are beginning to make an impact, an increased emphasis is being placed on developing sensitivity to the needs of users in the field who can employ the information to actually make a difference.

In Latin America, IRI learned that the climate data needed by large ranches in Argentina and Uruguay might be quite different from that required by potato farmers in Peru. More important, it is not enough to provide generalities such as a broad announcement that rainfall will likely be reduced across all of South America. To be useful, climate predictions have to take into account local exceptions. Farmers want to know what will happen in their specific region. This process of tailoring computer models to include local variations is called “downscaling” and it has become a major focus at IRI. Regional variations derived from studying past histories are worked into the overall computer model until more and more accurate regional predictions become possible. In 1999, a strong La Niña led to predictions that there was a strong risk of reduced rainfall in Southeast South America. That October, IRI predicted reduced seasonal rainfall, and collaborators in Uruguay also produced satellite maps showing where the vegetation was being hit the hardest. The maps made it possible for government policy planners to track the process of the drought on a weekly basis and eventually to distribute aid to the farmers who needed it the most. “At the end of the season we received a letter from Uruguay’s Minister of Agriculture, telling us that this was the first time that relief could be distributed on a needs basis,” says Walter Baethgen who manages Latin American and Caribbean operations at the IRI. In previous emergencies, aid money had simply gone to whoever had lobbied the most for it.

Baethgen also considers the notion of probability to be one of the most difficult ideas to get across. At this point climate models can generally favor one of three possibilities: that there will be more rain than normal, less rain than normal, or that the rainfall is likely to be “normal” for the season under consideration. The added climate information means that a blind guess that has just a 33% chance of being right, might be replaced by an educated guess with a 60% chance of being right, for example. But there is still a lingering possibility that the actual rainfall may fall in a smaller chance category. In one instance, Baethgen says, IRI made a three month prediction and it turned out that the actual rainfall fell into the highest chance category. Policy makers thought they had found the magic bullet. “Everyone thought that we had the problem solved,” says Baethgen, “and of course at the very next meeting, the exact opposite happened – the actual rainfall fell into a lower chance category, shattering user confidence. The prediction wasn’t wrong, but the way the information was being used wasn’t right either.” Over a difficult period, the users went from being extremely confident, to thinking that climate prediction was worthless to being somewhere in the middle. “Knowing something is much better than not knowing anything,” Baethgen says, “but we do not know everything. Your decision-making process will be much stronger, but there will always be uncertainties.”

Another difficulty in South America is establishing credibility with an indigenous population. In Northeast Brazil, farmers tend to listen to “rain prophets,” who predict rain based on the flight of birds and other traditional criteria. The gap, IRI decided, was largely due to the way scientific information was presented. A possible solution emerged through work with an anthropologist, who observed that in Northeast Brazil information is often communicated through music. One option under consideration there is local hiring of wandering minstrels to sing ballads that explain the impact of El Niño in terms that the villagers can relate to. “That can have more of an impact than any number of power point presentations,” Baethgen explains.

While many national economies and societies around the world are developing, poverty remains widespread in much of Africa and is getting worse. Part of the problem lies with the large number of rural people dependent on subsistence agriculture for their livelihoods. Too many people have inadequate land with poor soils and marginal climates, and consequently their livelihoods are extremely vulnerable to climatic fluctuations. Furthermore, the national economies of some African countries are highly dependent on seasonal climate variability, and may suffer repeated “shocks” from drought. A number of significant new initiatives are being launched to address entrenched problems. The Millennium Development Goals (MDGs) provide a common focus for development in Africa. But climate also has a confounding influence on many development outcomes. “Attention to climate variability is essential for measuring real progress toward the MDGs,” observes Madeleine Thomson, currently overseeing the Africa operations, in addition to leading the Impacts research program at IRI.

Three critically important sectors in the development agenda are particularly sensitive to climate variability — agriculture, health, and water resources — and IRI researchers are working with counterparts located in Greater Horn of Africa, Southern Africa and Sahelian West Africa toward best practices for managing climate risk in these areas. In particular, enhanced early warning systems to prevent climate-related disasters, drought and food security crises, disease epidemics and pest outbreaks are critical. Contributions of IRI scientists in these areas have led to the establishment of IRI as a WHO Collaborating Centre on early warning systems for climate-sensitive diseases with current projects focusing on malaria. Malaria is widely appreciated as the most important of the climate-sensitive diseases. It is seen as a major impediment to socio-economic development particularly in Africa where 90% of the 1-3 million deaths it causes each year occur.

There is considerable evidence that many infections diseases such as malaria occur within a climate “envelope” and vary between years in intensity as a result of inter-annual variation in climate and vulnerability. “Information on the seasonality of climate and its variability must be taken into account when planning and implementing routine health campaigns and epidemic preparedness,” says Stephen Connor, research scientist and Director of the WHO Collaborating Centre at IRI. An integrated framework for Malaria Early Warning Systems (MEWS) in Africa has been developed with WHO and many partners. “This framework features increasingly in national and regional malaria control strategies for epidemic prone areas as a result of new resources, knowledge, and changing policies across the continent,” Connor adds.

The broad social analysis which is an integral part of the short-term climate prediction that IRI is carrying out now may have long range implications for longer range climate changes that many scientists believe are being caused by global warming. The incidence of climate-related natural disasters appears to be increasing on a yearly basis. While the kind of climate science being developed at IRI doesn’t mitigate long term climate change, it does explore the best possible solutions for adapting to it.

“Most of the climate related events that affect societies play out over a short period of time, a given year, one rainy season, a cyclone or a flood,” Says IRI’s Director General, Steve Zebiak. “What we are really trying to do is to manage the impact. Building resilience to these kinds of problems now may be the best way to develop resilience to global climate change later on. The kinds of things that happen will be the same, there may just be more of them, and they may be happening more often. On the adaptation side, this is a way of worrying about now and the future.”