Online Catalog

GCRIO Home ->arrow Library ->arrow Consequences -> arrow Vol. 5, No. 2, 1999 -> arrow The Application of Climate Information Search
U.S. Global Change Research Information Office logo and link to home
Updated 17 November 2004
Consequences (title)
Consequences Vol. 5, No. 2, 1999
 
 

 

See also: article overview

 

 

 

 

 

The Application of Climate Information

BY E. S.SARACHIK

We have long been accustomed to the many benefits of daily weather forecasts, made several days in advance. But only in the last few years has another door begun to open, with even greater promises in store: the possibility of forecasting the climate, seasons or years in advance.

We learned of the advent of the 1997-98 El Niño several months before it appeared and, on both an individual and societal level, learned to prepare for its effects. There are now possibilities that climate forecasts made even farther in advance, will also be developed, and in not that many years. Armed with a new technology that promises longer and more accurate glimpses into the future, we might rightfully ask: If we have a reliable climate forecast, what should we do with it? What if we issued a climate forecast, and nobody cared?

The concept of end-to-end forecasting

Ultimately, climate forecasts are made to be used. A framework for thinking about the use of climate forecasting is called end-to-end forecasting, to embrace all the steps that go into making, disseminating, using, and evaluating the results of the forecast. Some typical steps are listed below.

  • Data relevant to the forecast are collected.
  • A forecast is made.
  • The forecast is used in a region for which there is a likely impact of the predicted variations.
  • The forecast is downscaled to the particularities of the specific region.
  • The accuracy and uncertainties of the regional forecast are assessed.
  • The system to which the forecast information will be applied is analyzed in order to cast the information in the most useful form.
  • A forecast of appropriate quantities is disseminated in understandable terms and to the right recipients.
    and
  • The forecast information is considered and acted upon, and the results of the application are evaluated.

Although our bulleted list makes it look that way, the end-to-end forecasting process is not linear, like a set of sequential steps for assembling a piece of lawn furniture. There are many interconnections among the various stages. For examples: the accuracy and specificity of the forecast will influence the data to be collected; the application will shape the forecast required; the regional culture and industrial base will help determine the information to be disseminated and the method of dissemination; and the applications will depend on the nature and directness of the climate impacts.

The concept of end-to-end forecasting is useful in that it provides a language for speaking and thinking about the applications of climate information. It also provides a research focus for understanding the connections between the physical, social, and institutional dynamics that are very much involved in the successful use of climate forecasts.

As with any new technology, the paths toward further development and implementation are by no means straight, and the implications are not yet clear. The federal government of our own country has made a major investment over the past ten years in examining the probable nature and implications of longer-term global greenhouse warming, but it is not organized to make the sustained climate observations that are essential for shorter-term climate forecasts. Nor is it equipped to apply predictions on seasonal-to-interannual time scales.

The close relationships that will be needed to bridge the gap between a public climate forecasting system, supported by the government, and a private industry, which shapes and applies the forecasts for the needs of specific customers, have not yet taken form. Thus the enterprise is in an experimental and demonstration phase. Both the U.S. government and non-governmental institutions (such as the National Centers for Environmental Prediction [NCEP] and the European Center of Medium Range Weather Forecasts, ECMWF) are making experimental forecasts. Other private and public institutions (such as the International Research Institute for Climate Prediction, IRI) are endeavoring to show that the forecasts of the climate can be put to work for the public good, by testing and demonstrating their use in real applications.

Here we briefly introduce the notion of climate information, discuss the concept of an application of that information, and look at some anecdotal examples in both the public and the private sector. We then discuss some of the obstacles and barriers to the successful applications of climate forecasts, and offer some possible guidelines for the future. Throughout, we shall concentrate on the applications of seasonal-to-interannual climate forecasts, as opposed to those that address climatic changes of decadal and longer periods of time.

Useful Climate Information

To the atmospheric scientist, weather deals with the detailed state of the atmosphere at a given moment on a specific day, while climate is an averaged and therefore more general descriptor that applies to a thicker slice of time. Averaged summer rainfall or temperature, the number of days with high precipitation in a season, the average number of freezing days in winter, the number of cloudless days in a season, all are climatic quantities. While it is not theoretically possible to forecast the weather more than perhaps ten days in advance, it is currently possible to forecast some aspects of the statistics of weather (and thus the climate) a year or more ahead, for some places on the globe. Any long-term average (or other statistic) of any atmospheric quantity is part of climate, and there are many of these. Here we limit our discussion to climatic variables that are known to affect daily life and commerce — such as temperature and winds and precipitation — and to information about them.

To be of any use, climate information must pertain ultimately to the future. It would indeed be interesting to know whether an El Niño or perhaps a La Niña was in progress when Ferdinand Magellan first sailed across the wide Pacific. But the real value of climate information is in what we will do with it — whether it tells of the past or the present or the future.

The most common type of useful climate information is the seasonally averaged temperature or rainfall in a given region. A knowledge of the past record of rainfall and temperature statistics is essential for all climate dependent activities. On the basis of past information, the farmer knows which crops will thrive; the city manager knows how much water capacity she must have to satisfy the needs of her constituents; the contractor knows how much insulation and how big a heating or cooling system to add to new construction; a resident knows whether or not to invest in an air conditioner. We have come to expect the local climate to vary within a certain range, but in the absence of other information, we base our decisions on the assumption that the climate in Omaha or wherever we live will behave, in the future, as it has in the past.

Information about the past, the present, and the future

Past climate information is useful to the extent that it describes a useful variable and is accurate, and to the extent that the existing record is long enough to have captured the full range of possible variations. It is made less useful when new, and therefore unexpected, conditions arise. This can happen, in the most trivial example, when what we know of river conditions is limited to a 200-year record, and a once-in-500-year flood comes along. It can also happen when conditions that shape the local climate are altered, driven perhaps by human activity, or by lasting changes in ocean currents, or other elements within or outside the atmosphere.

Present climate information is useful for knowing the current phase — in the simplest case: at a peak, or in a valley, or somewhere in between — in an assumed climate variation. With this knowledge, some predictive value obtains from the known past history of the variation. In a more typical example, knowing that sea surface temperatures in the eastern and central tropical Pacific are above normal in October indicates that El Niño conditions apply. This, and knowledge of how ENSO events have evolved in the past, allows us to infer, with some degree of confidence, the coming of warm conditions in the Pacific Northwest in the following winter, and cold conditions in the Southeast.

Future climate information (in the form of climate forecasts) gives us a detailed basis for planning. While past climate information tells us the range of possibilities, and present climate information tells us where we are within that range, future climate information provides a projection of what will likely happen and when, allowing us to make detailed preparations for the future. Because forecasts are not exact, the information about future climate will be less than 100 percent certain. The successful use of such probabilistic information is a skill that must be learned, to take into account both the known and the unknown uncertainties in the forecast.

Space and Time

How big an area should climate forecasts cover, to be most valuable?

While current climate forecasts pertain to relatively large areas — typically a square 1000 kilometers on a side, equal to all of California and Nevada — life and commerce take place at far more local scales, and so do many atmospheric effects. In the Midwestern part of the U.S., the land is flat and comparatively uniform so there is little orographic (land-surface-determined) effect on climate. In California, however, there is considerable orographic variation, and as a result, many different climate regimes. The coastal climate may be temperate, with high rainfall over the western slopes of the Sierras, and rainshadows on the eastern, lee slopes of the same mountains. The great California valleys, running north and south, are dry and hot during the summer and rely on irrigation for productive agriculture. The spatial scale of the forecast must be appropriate to local conditions, which generally means scales of 100 km or less, or about the size of Los Angeles. Since the basic forecast will generally be made for a much larger, averaged area, scaling it down to more local scales to add specificity will come at the cost of overall forecast skill.

For what period of time will climate forecasts prove most valuable? For the next season? The next year? The next decade? To answer this question we need to look very closely at the characteristics and capabilities of the region to which a forecast is applied.

Vulnerability, sensitivity, and adaptability

It is a well known human trait that most people are more concerned with avoiding damage than in accumulating benefits: few of us would forego homeowners’ insurance in order to invest in the stock market. The same notion of vulnerability applies to how we make choices with regard to climatic changes. Vulnerability can be thought of in terms of two factors: our sensitivity to unfavorable climate variations, and our ability to adapt to them. There are physical and sociological elements in both of these.

Thus, for example, our own country is sensitive to particular hurricanes, depending on their path and intensity, and certain of them can cause great local damage. Since the U.S. is an affluent nation with many tools to deal with disaster, both ex ante and ex post, the country as a whole is relatively invulnerable to the effects of any one of these storms. In contrast, Honduras is similarly sensitive but lacks the tools and resources to deal with such disasters, and is thus highly vulnerable, as was tragically demonstrated by hurricane Mitch in 1998.

Nature determines the course and timing of climatic variations, but our sensitivity to them is partly determined by human institutions, and particularly by the strength and quality of the built infrastructure. In the most extreme case, a region that chooses to build all houses underground, or otherwise hardens them, would be less sensitive and therefore less vulnerable to any and all climatic effects. Adaptability, however, is entirely a function of the ability and will of social and economic institutions and structures to respond. Wealth, education, and social development generally increase our ability to adapt, and thus potentially decrease vulnerability.

Ability to adapt is also affected by the relative size and population of the climatically impacted area in any country. The same hurricane that devastated a major portion of Honduras would have damaged a far smaller fraction of the U.S. When impacts of severe weather or changing climate affect a major part of the U.S. — as was the case in the 1930s Midwest droughts — the ability of the country to adapt indeed becomes an issue.

Adaptability also suffers when the duration of the damaging climate effects increases. A drought that lasts a week or two is obviously more easily adapted to than one that persists for a year, or a decade. At some point, the capacity to adapt to drought in even the most prosperous nations is strained, as in the 1930s in the U.S. When time scales of major climatic excursions approach a hundred years or more, a whole new class of challenges emerges.

Optimum scales of time

The time scale that is most important to address in climate forecasts will therefore differ from country to country and user to user. The vulnerability of some less developed nations to even a single year of adverse climate may mean that a forecast for next season or next year is the highest priority. Forecasts for longer periods, though important for any country, address issues that are not as pressing for they lie farther down the road. For the more developed nations, a one-year forecast is indeed valuable (both to avert damage and to take advantage of opportunities) but a forecast for a decade or longer may seem as important for decreasing vulnerability to adverse climate variations. While such forecasts are not presently made, the economic motivation for examining the predictability of climatic variations more than a year in advance is clearly compelling.

The only long-term climate forecasts that now exist offer climate information up to a year in advance. For the purpose of this review, therefore, we will concentrate on applications of seasonal and annual forecasts which can help nations by anticipating (and therefore preparing for and ameliorating) environmental risks, and/or by providing advance information that can help them economically. Sometimes neither can happen, no matter how accurate the forecast. It is sad but true that some regions of the world are unable to react to forecasts that already exist, due to limitations of their own resources, and insufficient interest on the part of the rest of the world.

Essential features of seasonal and annual forecasts

The preceding arguments could also have been made to justify the tremendous amount of activity and planning that have been invested in anticipating the response of the climate system to the build-up of carbon dioxide and other radiatively-active gases in the atmosphere. The mechanics of modeling the long-term response of the global climate system to changes in atmospheric chemistry are fundamentally different, however, from what goes into the more detailed predictions of climate that we are discussing here. Both are complex and highly sophisticated, and use up lots of computer time, but they serve different purposes.

Global climate models are designed to tell us the end effects of a predicated change in the climate system, such as doubling the amount of CO2 in the air, or turning up the solar radiation by 1 percent. The answers, in whatever form, respond to the questions that were asked: for example, what happens to surface temperature, over the globe, or precipitation, or wind patterns, or soil moisture?

In contrast, seasonal to annual predictions are generated by mathematical models of the atmosphere that begin at a precise point in time, with highly specific initial descriptions of the atmosphere at that moment. In the words of the forecaster, they are initialized. The model is then run to project the state of the atmosphere (for us, the weather and climate) forward in time for up to a year, with step-by-step reports of the state of the atmosphere along the way. Greenhouse warming models are not initialized, and are therefore in some regards less specific. They can tell us that seventy-five years from now the summers in Patagonia are likely to be warmer than now, but not whether a particular summer — say that of 2075 — is likely to be particularly warm or cold there.

Finally, we should probably point out that there are types of climate information that are not useful. Inaccurate, incomplete or otherwise corrupt information is clearly not. But so is information used unwisely or incorrectly. As we shall see below, the proper regard for the production and dissemination of predictive information, with due respect to its uncertainties, is one of the keys to successful applications.

Applying Climate Information

An application of climate information is the use of that information to change or influence a decision regarding future actions. Climate information could influence as simple a decision as an individual postponing a vacation or as complex as the redesign of the operating rules for a series of dams on a major river. In the former case, the climate information could be a forecast of a rainier than usual summer and, in the latter, a forecast of decreased rainfall for the next decade.

The decision may be made by a variety of actors and take a wide range of forms: for example, the passage of legislation (e.g. agricultural subsidies, flood insurance, regulatory actions); the buying of financial instruments (futures, credit, hedges, insurance); the hiring or firing of additional workers; the stockpiling of materials; the relocation of people; the shipping of materials; the changing of operating rules; the hardening of existing structures; or the building of additional infrastructure such as roads, bridges, aqueducts, reservoirs, and dams. Two actors, given the same information, may take different actions depending on their understanding of and belief in the information.

When and how to make the decision; what decision to make; what action to take; when and how best to take the resultant action; and how to judge whether or not to take a similar action next time: these options are the essence of climate applications and will differ from actor to actor and from case to case. The decisions that are made will depend in part on the receipt of the information in good time; on the accuracy, specificity, and credibility of the information; on the freedom of the actors involved to make such a decision and subsequently take the proper action; on the perceived value that the decision would bring, versus the perceived costs of a wrong decision; and on the outcome of such decisions in similar circumstances in the past.

In considering these questions, we need to remember that choices that affect the future are constantly being made based on very limited information. Until a decade ago, the only climatic information available to decision makers was information about past conditions: yet decisions that depended on future climate were constantly being made. Until now the most likely climate information for future decisions has been no climate information, and it takes little imagination to realize that a small amount of information about the future, however probabilistic, is better than no information at all.

Decisions based on probabilities

We can never predict the future climate with absolute certainty. For this reason, predictions are and always will be expressed in terms of probability of occurrence — much as the chances of rain, in today’s weather forecasts, are expressed in terms of percentage probability. As with any probabilistic scheme, significant benefits can be realized only over a long sequence of trials. The need to think and act in terms of probabilistic strategy is one of the greatest obstacles to the applications of forecast information. It takes a long sequence of forecasts to establish the specificity, uncertainty, and skill of the climate information, and a long series of applications to learn how to get the most from it.

As a simple illustration, consider the tossing of a slightly dishonest coin — say one with a 55 percent probability of landing face up. A long series of tosses would be required to verify that the coin was indeed dishonest, and, having ascertained that fact, a gambler would need a strategy in order to profit from this information. For example, he would probably not risk everything on a single bet on heads, for the probability of tails is 45 percent, and this is too risky. A succession of a few large bets on heads could also lead to the rapid loss of the initial stake. A better strategy would be to bet a small amount of money on heads for as many tosses as possible, to get the most from the statistics of probabilities. For an initial stake of $1000, for example, a dollar bet on heads each time for 1000 tosses will yield an expected gain of $100 — as compared to zero for an honest coin.

In this case, as with other probabilistic forecasts,

  • a strategy is needed for using the predictions;
  • the strategy should depend on the level of uncertainty;
    and
  • the benefits can be realized only over many trials.

In the case of climate forecasts, this means that many forecasts will be required in order to ascertain the skill of the forecast system. It will also take many applications to determine whether the advantages of the forecasts are being realized.

Public and Private Applications

In describing the applications of climate information, we need to distinguish between applications taken by the public sector and those made for more commercial purposes in the private sector.

The essential difference in this case is one of goals: private companies exist to make profit and their motives and actions revolve around this unambiguous purpose. Public organizations, including governments or foundations and non-profit organizations, have goals that are generally more diffuse, since the concept of public good and the desirability of public action change constantly with time and circumstance.

In most cases, the success of climate applications taken by the private sector can be judged by a single criterion: whether or not they were profitable for the entities involved. The success of public applications is more difficult to evaluate, given the multiplicity and nature of public goals, and the fact that they are ever changing and often contentious.

Some public sector applications can benefit certain members of the public while harming others. Further, since public goals may be multifaceted, it is often not clear what is being optimized by the application of climate information, and for what subset of the public. Ready examples can be found in the management of public water resources, where priorities of water quality, recreational use, the desirability of avoiding floods, and the needs of agriculture may come in conflict with each other, and often do. A public water manager can hardly manage for the success of a single specific goal. The success of his or her tenure depends mainly on how well conflicting priorities are balanced.

Examples of Applications

Private applications

Applications in the private sector involve profits, and competition among companies is often involved. Thus, the relative advantage that one company may gain over another, on the basis of climate information, is not always divulged. This may explain why there are so few confirmed studies of the private use of climate information. There are anecdotal examples, nonetheless, that illustrate some typical cases.

Many of us have gone to a department store in the spring of the year to look for a sweater and instead found only bathing suits. Similarly, when looking for a mower to deal with late summer lawns, it is frustrating to find only snow blowers in hardware stores. The correct placement of seasonal merchandise is worth a great deal to retail businesses, in that merchandise in stock during the wrong season either does not sell, and therefore must be warehoused until the next season, or must be heavily discounted to make room for the next season’s stock. In the absence of reliable climate information, stock is simply shipped to the merchandiser by certain canonical dates to assure delivery and display by the beginning of the season.

The application of more specific climate information in this case might unfold as follows. Suppose an El Niño, the warm phase of ENSO, has been predicted for the coming winter. The probability of a warm Pacific Northwest and a cold Southeast will be highly likely, based on that forecast. If additional information is available, such as the phase and intensity of the Pacific Decadal Oscillation (an oscillation similar to ENSO, and which interacts with it on longer scales of time), a reasonable probability could also be assigned to the expected rainfall in the two regions. Based on these probabilities, a certain amount of merchandise more suited to a warmer-than-average winter could be shifted to the Northwest, while a similar amount of colder-winter merchandise could be diverted to the Southeast. The merchandisers involved would have bet on the odds; and in the long run, if there were any skill at all to the forecast, they would benefit in doing so.

Some of the other private areas in which applications of short range climate predictions can be made include agriculture, horticulture, hydroelectric energy generation, other energy production and distribution, tourism, construction, fisheries, aquaculture, siting, trucking, shipping, and insurance.

Public applications

Climate is global but applications are invariably local. Even when aided by outside institutions, like FEMA or UNICEF or perhaps the IRI, the ultimate responsibility for making the applications falls on those who live in the region concerned, and who are most likely to be affected by the consequences of the applications.

In some cases, the capacity to use the climate information already exists in terms of a local understanding of weather and climate variations and the local capacity to understand forecasts. In some cases it doesn’t, but these abilities can be developed through education and training. In others, acquiring the needed capacity may be hindered by institutional, social, or political rigidities. Once the capacity to understand and apply the information exists, there needs to be an organized infrastructure to effect the application, gather feedback on its success or failure, and make adjustments in the light of lessons learned.

An example from northeastern Brazil

One of the few documented examples of the public use of advance climate information pertains to the Nordeste region of Brazil. This northeastern region is composed of nine states, covering about a fifth of the land area of Brazil and including about one third of its population. Except near the coast, much of the Nordeste is semi-arid and not well suited for agriculture, yet 20 million people live in these marginal regions. The social structure of the Nordeste is highly stratified in that a relatively few large landholders own most of the arable land. The tenant farmers and small independent farmers have relatively little political or economic power and are in many ways beholden to the large landholders. In this regard, little has changed in the Nordeste for a matter of centuries. Poverty is endemic, and the land itself is not a rich resource.

This region is particularly vulnerable to droughts, and for generations, multi-year droughts in the Nordeste have been followed by large-scale migrations of impoverished agricultural workers to the large cities of Brazil, and especially to overburdened Saõ Paulo. More recently, multi-year droughts have led the federal government of Brazil to make drought relief funds available, in part to assuage suffering, but also to keep the workers on the land and thus stem the flow of migrants which affect the remainder of the country. These large outlays demonstrate an obvious interest on the part of the federal government to improve agricultural output in the Nordeste in drought-beleaguered years.

The normal rainy season in the Nordeste is fixed by a feature of atmospheric circulation called the intertropical convergence zone, or ITCZ. Rain is most plentiful when this meandering element of the climate system reaches what is normally its southernmost position, sometime from February through May, and touches the northeastern corner of Brazil. Prediction schemes for rainfall in the Nordeste are based on sea surface temperature in the eastern tropical Pacific and in the tropical Atlantic. Rainfall drops below normal in the Nordeste during El Niños and at times when the Atlantic ITCZ moves northward. Over the last twenty years, statistical predictions of seasonal rainfall have been developed for the Nordeste, and they exhibit some skill.

The story of Ceará

In the 1980s, an institution of the state government of Ceará, one of the states of the Nordeste, decided to take advantage of the newly developed ability to predict climatic variations in rainfall. Toward this end, Ceara’s Foundation for Meteorology and Hydrological Resources (FUNCEME) converted its primary mission from cloud seeding to the applications of rainfall prediction. With the advice and help of the Brazilian Space Agency (INPE), satellite links were established, data were collected, and predictions were made, beginning at the time of the 1992 El Niño.

The Governor of Ceará was a strong supporter of the activity. Since his office controlled the distribution of drought resistant seed to small farmers, and because he could also influence the broadcast media, it was easier to elicit direct action in response to the projected deficits in rainfall. Forecasts were widely disseminated and the media were enlisted to spread the word of impending drought. No indications of uncertainty were included with the forecasts, however, and the authority of science was relied upon to encourage farmers to take adaptive measures.

The result, compared with an earlier El Niño, is shown in Figure 1.

While these results are impressive, there are as yet no studies of who benefited from the forecast, nor precisely what suite of actions were decided upon or how successfully they were carried out. Moreover, the publicity that accompanied the much-advertised forecast for 1992 led in time to grave problems for FUNCEME. Several years later, in December of 1994, a preliminary forecast of dry conditions for the 1995 rainy season was announced by the Director of FUNCEME. The early warning, which was coaxed out of the director by an anxious press, was less well founded than that for 1992, and based on preliminary data. A revised and more accurate forecast for normal rainfall was issued a month later, in January 1995, but the damage had already been done, for actions had already been taken on the strength of the preliminary forecast. The Director lost the confidence of the Governor of Ceará and was replaced.

Other stories could be told that have happier endings. Some other public applications of seasonal-to-interannual forecast information include: staging for forest fires; staging and other preparations for food relief; infrastructure enhancements for rainy seasons, such as dam building and strengthening, reservoir lowering, road diversions, and dwelling hardening); legislation for agricultural subsidies and seed distributions; legislation for flood or other types of insurance; enhancement of broadcast media and other warning systems; imposition of fisheries and other resource regulations; staging and warnings for climate related disease and infections such malaria, cholera, dengue fever, and other water borne diseases.

Barriers and Obstacles to the Use of Climate Information

It may seem obvious that since climate information is potentially helpful, it will naturally be put to immediate use, like the gentle rain that falls on a parched land. Nonetheless, there have been a number of interviews, especially with water managers, showing that climate information is rarely used, even when it is readily available.

Part of the reluctance is probably the novelty of such information. Established institutions have well-worn ways of doing things. The introduction of new information that may appear subtle and difficult to interpret — and which can prove its worth only over the course of many years — is all too easily dismissed. Another part of the reluctance to use climate information that foretells conditions a year or more ahead may be a fear of the future, which all of us share and most of us deny. We live only and always in the present, where there is plenty to be done. Even the Bible tells us to take therefore no thought for the morrow.

Yet another reason for reluctance on the part of decision-makers may be the risks involved in taking novel actions that may not work out. The penalties that might accompany failure could seem to outweigh what might be gained by success. These barriers to the adoption of new technologies are common and will only be surmounted as predictions are demonstrated, rather than argued, to be useful.

As we saw in the 1992 Nordeste example, adding a semblance of certainty to a climate forecast will more readily mobilize people to act. When a forecast presented in this way proves correct, the exaggerated claim of greater certainty may indeed be responsible for added success in applications. But when wrong, mistrust and suspicion can poison the atmosphere and endanger future applications. It seems an odd circumstance that for end-to-end forecasting to succeed, the first shot fired must hit the mark. A first-game win builds public interest, and allows a novel concept to be taken seriously. At the same time, an early defeat cannot be a reason to quit trying, for the stakes are large and important. The applications of predictive information are a long-term venture that will inevitably include failures, or series of failures.

An incomplete understanding of the system involved is a common cause of failure. The system to which climate forecasts are applied typically includes physical, social, political, ecological, and commercial aspects. Within each of these are elements that by their neglect can doom the application.

In a well-known study made twenty years ago, Peruvian fishermen were asked what they would have done had they an accurate forecast of the 1972 El Niño that had so disrupted their livelihood. Half said they would have caught all the fish possible, in anticipation of the event, since the fish were going to die anyway. Another half said that they would have refrained from fishing altogether, to increase the number of surviving fish, and thereby help revive the industry after the El Niño had passed. In this example, an accurate forecast might not have helped the plight of fishermen and fisheries at all. Unless we understand the precise relationship between warm water, fish recruitment and survival, and fishing practices, we are unlikely to make successful applications.

The enterprise can also founder due to unintended consequences. It is undeniable that predictions will disproportionately favor the more technologically advanced — because of their ability to more quickly and easily respond and adapt — and those with easier access to information. An application designed to help the more disadvantaged may in the end help those who need, and perhaps deserve, no help at all. In the final analysis, outside perceptions of the relative success of the application of climate information may depend most of all on who benefited most.

Finally, there are untested barriers — and some would add, all too many barristers. The body of case law that has been developed for weather forecasts may or may not apply to climate forecasts. We cannot dismiss the possibility, in this litigious society of ours, that users of climate information who feel they were misled — however probabilistic the forecast — will sue for real or perceived damages.

Looking to the Future

We saw that climate information is about the future and that its proper use must be learned over many sequences of predictions and ensuing applications. The analogous use of weather information has spawned a multi-billion dollar weather applications industry in the U.S. that has proven its value, over the tens of thousands of weather forecast-application cycles that have gone by since the advent of modern computer-aided weather forecasting. Climate forecasting is much newer, and by its nature, will mature more slowly. We can expect a climate applications industry to grow in time in this country and others — like the many interpreters of daily weather data that now serve newspapers and the broadcast media and other entities. But we are at such an early stage of development in climate forecasting that these and other applications are still being carried out by trial and error, to learn how best to do them.

Hope and help from the IRI

While most private applications are out of the public sight, public applications are relatively accessible. The International Research Institute for Climate Prediction (IRI — at http://iri.ldeo.columbia.edu) was designed to demonstrate the end-to-end concept and provide useful applications in the spirit of international partnership. There are good reasons to concentrate, initially, on demonstrating what can and cannot be done.

  • The Institute is a highly visible user of climate information, and by demonstrating successful applications, can strengthen the global consensus that is needed to sustain a climate observing system.

  • Applications thoughtfully conducted and honestly evaluated can help define research agendas for the physical, biological, social, and economic aspects of end-to-end forecasting.

  • Applications successfully executed can help many of the regions involved to help themselves, and serve as an example of unselfish collaboration between the developing and the more developed countries.

  • The concept of end-to-end forecasts, suitably demonstrated, can serve as a model of a different kind: for the interdisciplinary studies that are needed to understand the physical climate system, and its effects in natural and man-made systems. These studies include, among others, the natural behavior of complex systems; the impacts of climate variations on these complex systems; the economic and social value of climate impacts and applications; the optimum ways of disseminating climate forecasts; the political and social ramifications of applications of climate forecasts; and the diffusion of the new forecast technology into society.

Perhaps the greatest hope is that such an Institute may sensitize the citizens of the world into thinking about the future, and in the process, about making it a place where they would truly like to go.

For Further Reading

Currents of Change: El Niño's Impact on Climate and Society, by M. H. Glantz. Cambridge University Press, Cambridge, 1996. 206 pp.

El Niño and Climate Prediction, by E. S. Sarachik. Reports to the Nation on Our Changing Planet. NOAA Office of Global Programs and UCAR, 1998.

Learning to Predict Climate Variations Associated with El Niño and the Southern Oscillation: Accomplishments and Legacies of the TOGA Program. National Research Council, National Academy Press, Washington, D.C., 1996. 192 pp. Available for reading online at National Academy Press.

Making Climate Forecasts Matter, edited by P.C. Stern and W. E. Easterling, National Research Council, National Academy Press, Washington, DC, 1999. 164pp. Available for reading online at National Academy Press.

Socioeconomic Impacts of Climate Variations and Policy Responses in Brazil, edited by A Magalhães and M.H. Glantz. UNEP, New York, 1992. 155pp.

"The Brazilian Nordeste" by A. Magalhães and P. Magee. In Drought Follows the Plough, edited by M.H. Glantz. Cambridge University Press, Cambridge, 1994. 197 pp.

Figure Captions

Figure 1. Annual Rainfall (black, in mm) and grain production (blue, in percent of normal yield) in two dry El Niño years in the Nordeste, when seasonal precipitation dropped, as is typical, from an expected 1000 mm to 600-700 mm. In 1987, at the left, no El Niño-related action was taken, and grain production fell to less than 45 percent of the normal amount. Advance prediction of the 1992 El Niño, on the right, allowed drought resistant crops to be planted, and grain production dropped only to about 70 percent of the normal, wet-year harvest.

 

 

 

 


U.S. Global Change Research Information Office, Suite 250, 1717 Pennsylvania Ave, NW, Washington, DC 20006. Tel: +1 202 223 6262. Fax: +1 202 223 3065. Email: . Web: www.gcrio.org. Webmaster: .
U.S. Climate Change Technology Program Intranet Logo and link to Home