A Quarterly Publication
Issue 9/January 1998
The ENSO Signal provides updates on follow-on activities and serves as a medium of dialogue on issues stemming from the International Forum on Forecasting El Niño: Launching an International Research Institute, held 6-8 November 1995 in Washington, DC
The International Forum on Forecasting El Niño (November 1995) and the time since have focused attention on the concept of a network of distributed climate forecast centers, one of which, the International Research Institute (IRI) for climate prediction, would emerge as the core. It is planned that the IRI Core will centralize functions associated with the regular production, dissemination, and improvement of experimental seasonal-to-interannual climate forecast information and guidance.
The IRI entered a stage of rapid multilateral development in 1997. A series of four IRI "Core Group" meetings were held among a number of interested countries and agencies to advance interest in long-term sponsorship, governance, and management of the IRI Core. The IRI Core is currently hosted in the United States by Columbia University/Lamont-Doherty Earth Observatory (LDEO), and Scripps Institution of Oceanography (SIO), University of California-San Diego through a cooperative agreement with the US National Oceanic and Atmospheric Administration (NOAA).
The first IRI Core Group Meeting was held January 16-17, 1997 in Vancouver, British Columbia, Canada. The second meeting was convened April 29-30, 1997, at the Instituto Nacional de Pesquisas Espaciais (INPE), Sao Jose dos Campos, SP, Brazil. The third met July 30-31, 1997 in Washington, DC USA. The fourth and most recent meeting gathered participants October 30-31, 1997 in Taipei, Taiwan.
First Core Group Meeting:
The principal topic of the first IRI Core Group Meeting was to explore the multilateral dimension of IRI "core" activities. In response to Forum recommendations regarding a planning process for the IRI, this particular meeting was intended to bring together a select group to begin a realistic assessment of the prospects for multilateral support of those core functions. IRI Core Group participants were asked to help ensure the transition of the IRI from a unilateral activity at LDEO and Scripps, to a fully multilateral entity that is implemented, owned and operated by the international community.
Key issues emerging from this first meeting, included a suite of options for concrete sponsorship of the IRI Core in both direct and in-kind contributions, the significance of integrating IRI with on-going international activities and institutions, the importance of international visiting scientists to IRI Core staffing, and the added value of forecast dissemination, interpretation, and applications that IRI would provide to the world.
Second Core Group Meeting:
The purpose of the Second Core Group meeting was to continue discussion about multilateral support for the IRI Core, and to offer participants an opportunity to begin shaping the Institute's institutional structure while considering sponsorship of its activities.
Discussions at the second meeting centered around criteria for membership and representation on the future IRI governance and policy bodies which would be based on a minimum threshold level of direct cash or in-kind contributions. The support of international Visiting Scientists at the IRI Core from Core Group members was strongly encouraged as a means of in-kind contribution to the staffing of the IRI, as well as a basis for regional capacity building in climate forecasting and applications. It was decided that early contributors to the IRI Core would be known as "Founding Members" with due credit and privileges commensurate with their initiative and leadership. A time-table for IRI Core development was proposed beginning with an interim transition period moving toward fully multilateral status around the year 2000.
Third Core Group Meeting:
The third meeting effectively melded the IRI Core Group into an interim coalition of multilateral "Founding Members" who agreed to define and implement cooperative bilateral mechanisms to financially sustain, govern, and manage the near-term development of the IRI Core Facility over the three-year transitional period. The interim sponsors at the same time provisionally committed to pursue more permanent multilateral arrangements defining IRI Core sponsorship post-2000. A draft model for a reliable yet flexible IRI Core governance and financial management structure was discussed, along with the presentation of a draft hiring plan and operating budget.
Dr. John H. Gibbons, Director, Office of Science and Technology Policy (OSTP), and Dr. D. James Baker, Administrator, National Oceanic and Atmospheric Administration (NOAA), attended the fourth meeting on behalf of the White House and NOAA. As Host and Chairman respectively of the November 1995 El Niño Forum, both commended the efforts of the IRI Core Group in advancing the IRI toward fully multilateral status.
Fourth Core Group Meeting:
The parties present at the fourth meeting agreed to form an interim policy board to officially charter, govern, and support the IRI during the interim period toward fully multilateral status around the year 2000. As agreed to at the third meeting, sponsorship of the IRI Core will be managed through a series of interim bilateral agreements through the year 2000. The Core Group also consented to help recruit other potential sponsors for IRI Core Founding Membership in the coming months.
Following goals established during the November 1995 El Niño Forum, the IRI Core Group has undertaken in 1997 to establish a strong basis for the transition of the IRI Core from a unilateral activity to a fully multilateral institution for global climate research and forecast applications. This groundwork has provided the IRI with the opportunity to actually begin issuing experimental forecast products and guidance worldwide in the face of the strong 1997-98 El Niño event. The current IRI Core Group participants plan to sign interim bilateral agreements for sponsorship of the IRI in early 1998, around the time of groundbreaking for the IRI's new building at Columbia University, and in conjunction with the first meeting of the interim policy board that will charter and guide the IRI during the transition period.
Southern Africa Regional Climate Outlook Forum by Macol Stewart, NOAA Office of Global Programs
The first meeting of the Southern Africa Regional Climate Outlook Forum convened in Kadoma, Zimbabwe, 8-12 September 1997, to formulate consensus guidance for the 1997-98 rainfall season in southern Africa. This pivotal meeting was sponsored by the International Research Institute (IRI) for Climate Prediction, the European Network for Research in Global Change (ENRICH), the Southern African Transport and Communications Commission (SATCC), the U.S. National Oceanic and Atmospheric Administration (NOAA), and the World Meteorological Organization (WMO), and hosted by the Zimbabwe Meteorological Service. Together, Meteorological Services from SADC countries and climate scientists from universities and international research institutes reviewed the state of the global climate system and its implications for southern Africa, taking into account the major El Niño event currently on-going in the tropical Pacific Ocean. Previous El Niño occurrences such as in 1982-83, 1991-92 and 1994-95 resulted in low rainfall across much of southern Africa south of 10 degrees South and disrupted climate patterns around the globe.
The Outlook produced at the Pre-Season Forum divides the 1997-98 season into two periods (October-December and December-March) and provides probability distributions to indicate the likelihood of below-, near- or above-normal rainfall for a series of sub-regions. This Outlook was updated at the Mid-Season meeting, hosted by the Namibia Meteorological Service, 18-19 December 1997 in Windhoek, Namibia. As a follow-up to the successful Pre-Season Outlook Meeting, the focus of the Mid-Season Forum was to assess the Pre-Season consensus guidance based on climatic conditions and user community feed-back, to adjust this guidance according to updated model results, and to work towards a consensus Outlook for the remainder of the rainy season.
The Mid-Season Outlook Forum was open to and enthusiastically attended by members of the user community. Close partnership and interaction with the user community has been a distinguishing feature of the Outlook Forum since its conception at the Workshop on Reducing Climate-Related Vulnerability in Southern Africa (October 1996, Victoria Falls, Zimbabwe). This initial model, designed by the scientists, practitioners, and decision-makers who will be among the principal producers and users of climate forecasts in southern Africa, has been so successful that already it is being replicated in other regions of Africa, Asia, and the Americas.
The Post-Season Assessment Forum, to be held in South Africa, 12-15 May 1998, will evaluate the success of previous Outlooks, both scientifically and from the point of view of the user community by providing an opportunity for feed-back regarding content, format, lead-time, delivery, distribution, and other relevant issues. Additional information regarding upcoming meetings of the Southern Africa Regional Climate Outlook Forum and the consensus climate guidance produced at the Pre-Season and Mid-Season Meetings is available on the internet at http://www.ogp.noaa.gov/enso/africa_outlook.html or via e-mail at sarcof@ogp.noaa.gov.
ENSO and the Asian Monsoon by Dr. N. Sen Roy, India Meteorological Department
El Niño, as we all know, is the anomalous rise in sea surface temperature above its normal values off the coast of Peru, which can spread to the middle of the Pacific Ocean along the equator. A seesaw relationship between sea-level pressures over the eastern South Pacific Ocean and the Indian Ocean closely linked with the El Niño phenomenon is called the Southern Oscillation (SO). Gilbert Walker, who discovered SO around 1924, observed that "there is a tendency of pressure at stations in the Pacific and rainfall in India to increase while pressure in the region of the Indian Ocean decreases". It is now well established that these two are the atmospheric and oceanic manifestations of a single phenomenon. Extreme phases of El Niño and SO occur together although there is no one-to-one relationship between the two in their evolutionary cycle. Hence, these linked phenomena are often referred to as "El Niño-Southern Oscillation" or "ENSO".
El Niño proves to be an important heat source for the atmosphere, due to the Pacific Ocean being a vast body of water spread over a large part of the globe. It influences the climate of the planet, especially the monsoon of India and other Asian countries. Since its impact is often adverse, its periodic occurrence leads to social and economic dislocation over many parts of the world. Hence, there is considerable global effort in recent years to have a predictive understanding of this interesting phenomenon.
It is now well understood that ENSO arises primarily from mechanical and thermodynamical coupling of the atmosphere and the ocean. This has opened up the possibility of its dynamical simulation and prediction with the help of coupled ocean-atmosphere models. The growing capability of the models to predict an ENSO signal represents hope for climate predictions and seasonal forecasts in Asian countries where economies are greatly dependent on monsoon rains. In this article, we shall confine ourselves to its impact on Asian monsoon only.
Asian Monsoon
The Asian monsoon broadly consists of two distinct components: the Indian monsoon and the East Asian monsoon; and is characterized by the seasonal reversal of wind. Thus, in most areas we get two monsoons: one in the summer and another in the winter. However, it is the summer monsoon that is the main rainy season for a greater part of the area. It brings the usual summer rains in India, which is about 80 percent of the country's annual rainfall.
The East Asian Monsoon has different names as it progresses northward: "Mei-yu" in China, "Baiu" in Japan and "Chang Mau" in Korea. These rains start as early as April over southeast China and shifts progressively northeastward in the following months, reaching northeast China by August. It touches the southern tip of India by June 1st and covers the entire country by July 15th. The advance of the summer monsoon over Asia has a close association with the seasonal northward movement of the sub-tropical anticyclone in the western Pacific. The monsoon withdraws from the continent in September.
Monsoon - ENSO Relationship
The Asian Monsoon region is linked with other parts of the tropics including the Pacific and subtropics by the Hadley and Walker circulations. In spite of the early association of ENSO research with the Indian monsoon, the focus of several recent studies has been on the Pacific region. In some studies, reduced summer rainfall over the semi-arid region of northern China has been found to occur during the El Niño years, whereas enhanced rainfall tends to occur during the La Niña years. In the Yangtze valley Mei-yu is often seen to depend on the onset of an El Niño, and rainfall is often greater than normal when the onset is in fall or winter. It is less than normal when the onset is in summer or spring. In Japan, El Niño events are generally associated with mild winter and cool summers and the Baiu (rainy season) is relatively prolonged. Changes in the frequency and location of the genesis of typhoons during El Niño years have also been reported.
Research on operational long-range forecasting of the Indian summer monsoon has continued to explore the relationship between ENSO and the monsoon. From 1875 to 1995, India experienced 19 drought years when rainfall was 10 percent or more below normal. Out of these 19 drought years, 11 years were associated with ENSO, while the remaining nine were non-ENSO years. Similarly, from 1901-1995 there were 17 ENSO events out of which nine were associated with below normal (-10% or less) rains, while on eight occasions rainfall was normal (+ 10%). Thus, there is no one-to-one relationship between monsoon rains in India and ENSO. This appears to indicate that there are other countervailing factors that neutralize or even reverse the effect of ENSO. It is significant to note, however, that in none of the ENSO years, was monsoon rainfall excessive (i.e., more than 110%).
This explains the rationale of using 16 predictors, including five ENSO related parameters, in the monsoon seasonal rainfall forecast model being used successfully by the India Meteorological Department at present.
Influence of Monsoon on ENSO
Many recent studies suggest that variations in monsoon circulation often precede variations in Southern Oscillation Index (SOI). This points to the possibility that the monsoon itself influences ENSO through a complex interactive mechanism. It is of interest to mention here that such possibility was foreseen as early as 1953 by Normand when he concluded that the Indian monsoon stands out as an active, not passive, feature in world weather, and proves to be more efficient as a broadcasting tool than an event to be forecast. Complexity, however, arises because of the fact that such a two-way relationship goes through strong and weak phases.
El Niño Prediction
In view of the global impact of El Niño, intensive effort is being made in predicting El Niño events over varying time-scales. The successful completion of the TOGA (Tropical Ocean-Global Atmosphere) program (1985 - 1994) has helped in this endeavor through the use of coupled ocean-atmosphere models, canonical correlation analysis (CCA) and linear inverse modelling techniques. However, only modest skill has been achieved so far. For example, the skill of CCA forecasts routinely issued by the NOAA (National Oceanic and Atmospheric Administration) Climate Prediction Center in Camp Springs, MD,USA shows a clear annual cycle. This skill is derived from cross-correlation tests from 1956 to the present.
However, as we have seen earlier, the exact relationship between El Niño and the monsoon is still elusive, while there are still other forcings to reckon with. An example is the slowly varying boundary conditions provided by the snow cover over the Himalayas and the Eurasian landmass. Therefore, even after the skill in El Niño forecasting improves, translation of such skill into the prediction of a highly complex phenomenon like monsoon rainfall will still be a formidable scientific dilemma.
ENSO, Tropical Cyclones, and the Insurance Industry by Dr. David Malmquist, Bermuda Biological Station
One of the most consequential of ENSO's economic impacts is its effect on tropical cyclone activity. When Hurricane Hugo struck South Carolina in 1989, it caused $7 billion in damages, nearly equal to all the damage wreaked in the US by the 1982-83 El Niño. Given the impacts of tropical cyclones on insured property, and the link between these storms and ENSO, it is not surprising that insurers have begun to look for ways to use ENSO and tropical cyclone forecasts in their business. This article explores the skill and value of tropical cyclone forecasts as they relate to ENSO and the property/catastrophe insurance industry. Evaluation of these forecast parameters in terms of the insurance industry stems from a dialog between climate scientists and insurers via the Risk Prediction Initiative (RPI). The RPI is an industry-funded partnership operated through the Bermuda Biological Station for Research.
The link between the state of ENSO and global tropical cyclone activity is well documented (RPI 1997, Landsea in press). In the Atlantic, a strong correlation exists between El Niño, increased vertical wind shear, and fewer hurricanes (Gray 1984, Goldenberg and Shapiro 1996). The correlation between ENSO and tropical cyclone frequency permits forecasters to use the state and trend of ENSO as one predictor of seasonal tropical cyclone activity, both in the Atlantic (Gray et al. 1992, 1993, 1994; Lehmiller et al. 1997) and western Pacific (Chan 1985, Nicholls 1992).
Forecast skill and value
Forecast skill measures the performance of a forecast compared to a standard, such as climatology or persistence (Murphy 1993). Table 1 shows the skill of Gray et al.'s Atlantic seasonal hurricane forecasts. Note that nearly all the June and August predictions show significant skill, whereas the December forecast currently shows no significant skill in real-time predictions.
Forecast value measures the economic benefits realized by decision-makers using a forecast. So how valuable are tropical cyclone and ENSO forecasts to property/catastrophe insurers? And how can they be made more valuable? Answering these questions requires an understanding of the skill and timing of current forecasts; the impact of tropical cyclones and ENSO on the insurance industry; and the way in which insurers conduct their catastrophe business (Roth 1997, RPI 1997).
Insurers tend to write annual contracts, and develop business plans and client histories over years to decades. Thus, insurers currently find multi-year forecasts more valuable than seasonal or real-time forecasts, as they can use them to help plan long-term investment strategies or portfolio changes. The value of seasonal forecasts depends on their lead time (Table 1). Catastrophe insurers write most of their US business for the June-November hurricane season in December of the preceding year. Thus post-December forecasts have reduced value, and the March-May barrier to ENSO forecasts (Wright et al. 1988) is an important limitation. About 35% of US catastrophe business is written to a July 1st deadline, which does allow reinsurers to consider using later forecasts to restructure their portfolio or hedge their risks to other carriers.
Enhancing seasonal forecast value for insurers entails increasing the lead time of skillful forecasts (Table 1), altering the timing of insurance decisions, or both. Real-time forecasts are valuable for assembling claims adjustment teams, and may soon become much more valuable given recent attempts to treat insured risks as tradeable commodities. Primary insurers in many countries are regulated by government, which further affects forecast value. For instance, even if a primary insurer receives a December forecast that a Category 4 storm will hit Miami on August 24 of the following year, regulations limit the actions they might take to reduce their losses. Reinsurers are less regulated and thus more flexible in their use of forecasts. Greater forecast value might be obtained if regulators and insurers were to mutually determine the proper role of climate forecasts in rate-setting.
The relationship between forecast location and the concentration of insured risks also affects forecast value. From an insurer's perspective, skillful forecasts of Bangladesh typhoons have little value, due to the area's paucity of insured property. This fact also limits the value of ENSO forecasts for many catastrophe insurers, as the most predictable consequences of ENSO occur in the developing nations of the tropics, where the value of insured property is relatively low. For areas with significantly insured property, catastrophe insurers must avoid aggregating risk geographically or via positive correlations between non-adjacent locations. Localized forecasts of tropical cyclone landfall are thus more valuable to insurers than basin-wide forecasts. Also valuable is knowledge of teleconnections among risks in widely separated regions (hence the value in knowing the simultaneous effects of ENSO on Atlantic hurricanes and western Pacific typhoons).
Forecasts that distinguish between storms of different intensity also add value for insurers. Analysis of tropical cyclone damage in the Atlantic between 1925-1995 shows that damage on the US mainland increases exponentially with increasing wind speed. Thus, a landfalling Category 4 hurricane causes up to 250 times the damage of a Category 1 storm (Landsea 1993, Pielke and Landsea in press). Hence, skillful forecasts of intense landfalling storms are of greater value to insurers than blanket forecasts covering storms of all intensity classes.
Another factor that controls forecast value is the nature of the predicted peril. In the US, most flood insurance (especially for homeowners) is carried by FEMA. This reduces the value of flood forecasts to insurers. Federally funded flood insurance also covers damage from hurricane-generated storm surge, whereas insurers tend to cover losses only due to "wind-driven rain". Yet insurers do find some value in storm-surge forecasts, as they may help claims adjusters distinguish between wind- and water-caused damage.
The value of climate forecasts depends as well on their treatment of uncertainty. Forecasters need to continually stress the probabilistic nature of their predictions, often in the face of media oversimplification. Consider El Niño forecasts. El Niño is a predictable feature of climate, yet this does not imply the skillful forecast of the onset, duration, and demise of every El Niño. Failing to stress the uncertainty of probabilistic forecasts invites the "overselling of science" (Pielke and Pielke 1997), with negative implications for users' confidence in future forecasts if currently forecasted events do not occur. Hurricane Andrew's landfall during an El Niño year of below average Atlantic hurricane activity underscores the uncertainty of ENSO and tropical cyclone forecasts.
Finally, for ENSO and tropical cyclone forecasts to be most valuable to the insurance industry, they should readily plug into the catastrophe models that insurers use to assess risk (Figure 1). This might be via a "tuning knob" that lets a model user alter tropical cyclone landfall probabilities based on the predicted state of ENSO. Full value here requires additional research to explicate the impact of ENSO on tropical cyclones both before, during, and after an event, and in terms of the time of year that an event occurs. Modelers must also provide mechanisms that allow model users to distinguish between forecast uncertainty and model uncertainty, and to compare their magnitudes.
In assessing the risks posed by ENSO and tropical cyclones, climate scientists and insurers share many interests. Ultimately these interests also affect society at large. If insurers incorrectly assess their risks, they may face bankruptcy. If that occurs, property owners or government are forced to pay the costs. Unpaid claims following Hurricane Andrew totaled $400 million.
For society to fully utilize ENSO and tropical cyclone forecasts, scientists must develop strategies for improving forecast skill and value. Also, insurers must be prepared to deal with the limitations and realities of science in their decision making. The Risk Prediction Initiative has recently published a research agenda that identifies some of the steps required to increase the value of seasonal and multi-annual tropical cyclone predictions.
This document is available on request from the RPI (www.bbsr.edu/rpi/).
References
¥The National Oceanic and Atmospheric Administration (NOAA) recently released the publication entitled "Reports to the Nation on Our Changing Planet: Our Changing Climate" designed to raise public awareness of issues dealing with global environmental change. If you are interested in receiving a copy, please forward your request to Sondra Lancaster by internet at Lancaster@ogp.noaa.gov or in writing to:
¥The International Research Institute (IRI) for Climate Prediction is holding a 1997/1998 lecture series on the Application of Seasonal-to-Interannual Climate Predictions. All lectures will be held at 4:00pm in the Davis Auditorium of the Chapero Center for Engineering and Physical Science Research at Columbia University in Palisades, NY USA. (Unless otherwise noted)
The following speakers are scheduled to appear:
January 29, 1998. Frederick Semazzi, World Meteorological Organization (WMO), Geneva, Switzerland
"Stepping Forward: Implementation of the WMO Climate Information and Prediction Services project."
February 12, 1998. Daniel Suman, Rosenstiel School of Marine and Atmospheric Sciences at the University of Miami
"Political and Economic Adaptations of the Chilean Fishery Sector to ENSO climate variability."
Upcoming speakers include: Mickey Glantz (NCAR); Mohammed Sadek Boulahya (ACMAD); Jim Buizer (NOAA/OGP) and Bob Watson (World Bank)
For further information, please call (1-914) 365-8368 or visit the IRI homepage at http://www.iri.ldeo.columbia.edu
¥ Two International Symposia on Culture and Sustainable Development (CUSDEV '98) are scheduled to be held 8-29 January 1998 in Fatick (Senegal) and 8-29 March 1998 in Kaolack. Volunteer service and community dynamics will be explored through practical work camps including but not limited to cassava harvesting, natural dyes promotion for sustainable textile technology development, tree planting, medicinal plants use and local knowledge, village health center building, and several other cultural areas. The focal point of the Symposia will be an integrated farm whereby participants, while getting hands-on-training, will be in close contact with the social and multicultural dimensions of community dynamics. Formal and informal discussions on the impact of local legends on sustainable development will be conducted, and special emphasis will be placed on the economic empowerment policies for women underway worldwide.
The Symposia will be diagnosing and documenting the impact of community perspectives on sustainable development in the coming century. Because of the interactive orientation, students, volunteer services, NGOs active in community development issues, youth groups, and the general public are encouraged to participate. For more information or to register, contact Suzanne Seck, Registration Div., PO Box 10775, Dakar-Liberte, Senegal, West Africa. Tel: (221)820-38-35; Fax: (221)827-54-04.
¥ The International Symposium on Climate Variability: Impacts on Rural Activities, organized by the Department of Geography and Environmental Science, University of Newcastle, Australia, will be held 13-17 April 1998. The primary goals of the symposium are to explore the present state of knowledge in the broad area of climate variability and climate change, and to assess the impacts of climate variability and change on rural industries, activities, and lifestyles. Papers focused on the assessment and causes of climate variability and change, the environmental and social impacts, and methodologies used to cope with climatic change and variability will be welcome. For more information, contact Howard A. Bridgman, Organizer, International Symposium on Climate Variability, Dept. of Geography, University of Newcastle, NSW 2308, Australia or by e-mail at gghab@cc.newcastle.edu.au or by fax to (61) 49-21-5877.
Lisa Farrow (301) 427-2089 ext. 25