NCAR --> SERE --> ISSE --> La Niña Summit
Review the Causes and Consequences of Cold Events
A United Nations University / NCAR / UNEP Activity
Summary Report Update of current La Niña event Table of Contents
Introduction The workshop on "A Review of the Causes and Consequences of Cold Events: A La Niña Summit" was convened in Boulder, Colorado (USA) on 15-17 July 1998. Michael Glantz, Senior Scientist at the National Center for Atmospheric Research (NCAR) and convenor of the workshop, opened the La Niña Summit. He welcomed the 80 participants from 14 countries on behalf of the United Nations University (UNU), the UN Environment Programme (UNEP), NCAR, and the US National Science Foundation (NSF). He noted that the discussions of the La Niña Summit would be carried real-time in audio on the Internet by the Exploratorium (San Francisco, California) at www.exploratorium.edu/la_Niña/ He introduced Dr. Richard Anthes, President of UCAR (University Corporation for Atmospheric Research), who welcomed the participants to the first workshop focused on La Niña, cold events in the tropical Pacific Ocean. Anthes commented on the importance of La Niña and El Niño research and its potential benefits to societies around the globe. Peter E.O. Usher, Head of UNEP's Atmosphere Unit (Nairobi, Kenya), greeted the participants on behalf of the UN sponsoring agencies, the UNU and UNEP. He recounted the history of UNEP's concern with the ENSO (El Niño-Southern Oscillation) cycle since the mid-1980s, when UNEP in cooperation with NCAR formed the UNEP Working Group on the Socioeconomic Impacts Associated with ENSO. He mentioned the numerous workshops and publications carried out by the Working Group. He also noted that ENSO research served as a bridge between the climate change and the climate variability research communities. Glantz then highlighted to the participants the reasons for convening the La Niña Summit. He noted that the overriding goal of the Summit was to encourage a review of the causes and the physical and societal consequences of La Niña (cold) events and to identify what was known and not known by researchers about La Niña and to identify what societies need to know about the phenomenon. The reason to know more about La Niña (and El Niño) is to improve the reliability and credibility of ENSO-related forecasts so that societies could better prepare themselves for their adverse, as well as beneficial, impacts. It was a first attempt to have scientists focus on La Niña, through their discussions with forecasters, social scientists, climate impacts and forecast application researchers and to help to identify and lay out what potential users of La Niña information (including La Niña forecasts) might be able to consider as reliable information about cold events. This workshop was NOT an attempt to forecast whether the recent strong 1997-98 El Niño would be replaced by a La Niña, when that replacement might be expected to take place, the magnitude of the potential La Niña, or what La Niña's impacts on particular societies and ecosystems might be. The idea to convene such a workshop was developed very early in 1998 and was in direct response to the media coverage (some say hype) of the 1997-98 El Niño event, one of the biggest in a century, and to the beginning of talk of the eventual development of a strong La Niña event later in the year. It had become clear from media coverage and interviews with scientists that much less was known about La Niña and its possible societal impacts in comparison to El Niño. For whatever reason, it was a phenomenon that had not received much scientific or media attention in the past couple of decades. Another workshop objective was to bring together as many as possible of the key El Niño researchers and representatives of the climate-information-user community to discuss what was known about La Niña's scientific dimensions and societal impacts. In addition, representatives of the media were invited to the workshop as participants as well as potential reporters of its proceedings. Media interest in the summit was surprisingly high. With respect to educating the public about La Niña (or cold events), we (i.e., physical, biological, and social scientists and the media) are in a good position to identify and avoid some of the false expectations that would likely accompany the misperceptions and misinformation about La Niña. As has been suggested in the popular literature, the extremes of the ENSO (warm event/cold event) process in the equatorial Pacific are among the major climate-related disrupters of human activities, bringing droughts, floods, and other severe meteorological events to various parts of the globe. The more that societies learn about the entire ENSO process, the better prepared they can become to cope with the regional extreme events that tend to accompany it. The following sections of this report present information from each of the workshop sessions in the order in which they appeared in the agenda: A Review of the 1997-98 El Niño Event, Definition(s) of La Niña, What Constitutes Normal, La Niña Teleconnections, The Impacts of La Niña on Specific Countries and Sectors, Climate Change and the ENSO Process, Symmetry Issues, Attribution of Impacts to ENSO Extremes, Media Panel, Forecasting the 1997-98 El Niño, Forecasting the Onset of a La Niña in 1998-99, Differences in Forecasting El Niño and La Niña, and Monitoring La Niña. It is important to note that the information presented under each topic was not mutually exclusive and was frequently referred to in other sessions as well. Thus, similar issues were raised in more than one section. El Niño, La Niña, and the Climate Swings of 1997-98: A Review Dr. Michael McPhaden (National Oceanic and Atmospheric Administration/Pacific Marine Environmental Laboratories) provided an overview of recent developments in the tropical Pacific Ocean. He noted that the 1997-98 El Niño has been called "The Climate Event of the Century". It was one of the strongest El Niños on record, with spectacular impacts on global weather variability and Pacific marine ecosystems. Successes in observing the evolution of the El Niño, and in forecasting its impacts, were in striking contrast to the situation only 15 years ago when the last major El Niño in 1982-83 was not even detected until nearly at its peak in the fall of 1982. These successes were in large part the result of (1) the development of an observing system of satellites and ocean measurements, anchored by a buoy network spanning the equatorial Pacific (Figure 1) providing data in real-time (within hours of collection); and (2) the development of computer forecast models for predictions of the El Niño/Southern Oscillation (ENSO) cycle of both warm (El Niño) and cold (La Niña) events. The onset of the 1997-98 El Niño was heralded by warm sea surface temperature (SST) anomalies (i.e., deviations from normal) erupting in the tropical eastern Pacific during April-June 1997 (Figure 2c). By July 1997, SST anomalies were the warmest observed in the past hundred years in this region. By the end of the year SSTs exceeded 5°C, higher than observed any time during 1982-83 (Figure 3), which up to that time had been the strongest of the century. As the warm SST anomalies began to wane in early 1998, prediction models were largely consistent among themselves in forecasting an end to El Niño, and the development of a La Niña, sometime during the second half of 1998. The surprise to El Niño researchers, though, was how quickly the tropical Pacific switched from warm to cold conditions. McPhaden commented that the system of weather buoys that was successful in tracking the evolution of the 1997-98 El Niño through its many phases detected an unprecedented 8°C drop in sea surface temperature at one of the buoy locations in the central Pacific from early May to early June 1998 (Figure 3). Such a temperature drop over a relatively brief 30-day span had never been observed before, making the termination of the El Niño as stunning as its onset. He suggested that the seeds for El Niño's demise, and for the birth of La Niña, were to be found below the surface. In late 1997, the buoy array detected a cold mass of water, situated at 100-150 m depth west of the date line, beginning to expand eastward along the equator (Figure 4). This cold water progressively underrode the layer of El Niño-warmed surface waters throughout the first part of 1998. However, because the trade winds remained weak during early 1998 in the eastern and central Pacific, this cold subsurface water could not be upwelled and mixed with the warmer surface water. By late April 1998, a thin surface veneer of unusually warm water capped a thick layer of unusually cold water just below across a wide swath of longitudes. When in early May 1998 the east-to-west trade winds finally picked up to near normal strength in the eastern and central Pacific (Figure 2a), the cold water hovering just below the surface was drawn up to cool the surface at a record pace. Since then, the patch of cold water that developed at the surface has continued to expand westward and to cool slightly (Figure 4). Interestingly, unusually warm SSTs still lingered for several hundred kilometers off the west coast of South America in a region where the trade winds had as of mid-August still not fully returned to normal. The slow development of the subsurface cold water anomalies along the equator was in effect a manifestation of thermocline shoaling. (The thermocline is the sharp boundary separating the warm surface layer from the cold interior ocean.) Thermocline depth variability is in turn largely controlled by changing large-scale wind patterns. As the trade winds weakened at the onset of the El Niño, wind-forced equatorial Rossby waves radiated westward causing the thermocline to shoal (upwell), and equatorial Kelvin waves radiated eastward causing the thermocline to deepen (downwell) (Figure 4). As the event continued to evolve, the westward radiating upwelling Rossby waves bounced off the land masses bordering the western Pacific and came back along the equator as upwelling Kelvin waves. Also, the trade winds in the western Pacific actually became stronger than normal in early 1998 (Figure 2a), tending to excite upwelling, eastward-radiating Kelvin waves. The net result of these wind-forced wave processes was to cause a slow thermocline shoaling to progress from west to east along the equator during the first several months of 1998. It is these slow ocean dynamics involving wind-forced variations in the thermocline that preconditioned the equatorial Pacific to switch from El Niño toward La Niña in 1998. Many of the ENSO prediction models that use equations of motion to represent the dynamics of the atmosphere and ocean successfully forecast the eventual movement toward La Niña, in part because they were initialized with wind and ocean temperature data that carried the information about these relatively slow time-scale physical processes. However, none of the computer forecast models predicted exactly when, or how rapidly, the switch would occur from El Niño to La Niña. The transition was triggered by a rapid and relatively unpredictable strengthening of the trade winds over a very short period of time in the eastern and central equatorial Pacific. In summary, McPhaden noted that during the past two years we have witnessed a spectacular display of climatic variability. The first part of the display, El Niño, is over and the second, La Niña, is under way. A newly completed observing system designed specifically for ENSO detection and forecasting provided essential information at a level of detail never before possible. Recently developed ENSO forecast models, initialized with these new data sets, were correct in most cases in predicting that 1997 would be warm and the latter half of 1998 would be cold in the tropical Pacific. As a result, our climate crystal ball was sufficiently clear in 1997 to motivate nations and governments to understand disaster preparedness, mitigation efforts, and other responses to developing El Niño conditions on an unprecedented scale. We can expect that, as current cold conditions continue to develop in the tropical Pacific, climate analysis and forecast information will be likewise valuable for planning purposes, allowing us to translate scientific progress in climate research into societal benefits worldwide. Key Points:
Definition(s) of La Niña James O'Brien (COAPS) opened up the discussion on the definition(s) of La Niña. Cold events in the central equatorial Pacific have been referred to in several ways: one of the most popular has been La Niña. Other ways include but are not limited to the following phrases, which have been taken from the peer-reviewed literature and the popular media: a cold episode, El Viejo, a cold phase of ENSO, El Niño's counterpart, a cold counterpart of El Niño, the inverse of El Niño, a cold weather version of El Niño, a cold phase, El Niño's sister, anti-El Niño, seasons with cold sea surface temperatures, an anomalous cooling, the opposite of El Niño, a sisterly event, El Niño's cold water opposite, the flip side of El Niño, abnormally cold, El Niño's lesser known twin, a periodic abnormally cold sea surface current, the girl child, the other extreme of the ENSO cycle, a mature cold episode, ENSO's lesser known twin, non-El Niño, and so forth. Each of these descriptions has actually been used by one scientist or another and the media. This particular set of synonyms suggests that the reference point for La Niña is El Niño, and not average or "normal" sea surface temperatures in the central equatorial Pacific. Most articles for that matter that refer to La Niña put it in the context of the El Niño phenomenon. However, most articles on El Niño, until recently, have not referred to La Niña. Thus, unlike for La Niña, it appears that the reference point for an El Niño event has been average or normal conditions and not its counterpart, La Niña. In a strict sense, La Niña is an extreme cooling and not just any cooling of sea surface temperatures in the central equatorial Pacific. Some people have suggested that for many parts of the globe La Niña can be viewed as an extreme case of normal. This point generated considerable discussion at the workshop without developing a consensus on the resolution of this issue. Definitions are arbitrarily true. As long as one makes explicit his/her definition, others can understand what s/he is talking about. However, when there are several definitions of the same phenomenon, it becomes difficult to compare analyses and findings. With regards to El Niño and La Niña, various researchers, groups of researchers, and research centers prefer one definition (theirs) over the others. As a result, they produce, according to their definitions, lists of years during which cold events occurred. A problem is that these lists do not necessarily agree from one researcher to the next, making it difficult to correlate cold events reliably with, for example, crop yields or fish landings. This problem becomes more acute because we are dealing with a relatively small number of La Niña events (qualitatively defined as unusually cold sea surface temperatures for an extended period of time in the central and eastern equatorial Pacific Ocean). According to one definition, there have been four La Niña events since the early 1970s. By another more strict definition of La Niña, there were only two La Niñas in this period. Since the early 1970s, scientists note that there have been twice as many El Niño events. Thus, changing the quantitative aspects of the La Niña definition changes the number (up or down) of events that one would include in an assessment of the phenomenon or its impacts. The less strict the definition, the more La Niña events that can be counted. As Barnston noted in his "Thoughts About La Niña" (see Barnston in Appendix), "when the limit [or threshold set for defining a La Niña] is too strict, the events included are strong, but the sample size is small. When the limit is more lenient, a larger sample can be included, which is good for circumventing flukes but runs the risk of including La Niñas of very different strengths and flavors in the same classification." Thus, in the absence of an agreed-upon definition, it becomes more difficult for climate impact researchers to identify with some degree of confidence important La Niña-related climate anomalies and their societal and environmental impacts in tropical regions, let alone the extratropical impacts that might be linked to La Niña. During discussions, the idea was raised that perhaps La Niña could be defined by the location of its worldwide teleconnections and societal impacts. However, various participants pointed out problems with such an approach: (1) given the increasing reliability and credibility of forecasts, proactive preventive or mitigative actions can be taken to reduce potential impacts, thereby reducing the severity of first-order impacts (e.g., in anticipation of forest fires, underbrush can be cleared away); (2) During the 1997-98 El Niño, The Indian monsoon did not fail as had been expected. Likewise, the Australian winter wheat crop was successful despite drought conditions; (3) During weak La Niña events, the possible impacts in distant locations (teleconnections) are greatly weakened and, as a result, local and regional climate conditions are likely to be determined more by local factors than by relatively small sea surface temperature anomalies in the distant equatorial Pacific. It was also suggested that La Niña could be defined by various 'users' of information, as dictated by their needs. This may be appealing to knowledgeable users of La Niña (and El Niño) information, because they are likely to be in a better position to identify and use the time series that best suits their needs (e.g., SST, SOI, OLR) for various parts of the equatorial Pacific (e.g., Niño1, Niño2, Niño3, Niño3.4, Niño4, or Niño C [see Wang in Appendix]). This approach to defining a La Niña event may be of little value to other potential users of such information who are less sophisticated in the science of air-sea interactions. Potential users are those who have not yet realized the value to their decision making processes of La Niña information, including La Niña forecasts. Converting potential users to actual ones is an important but difficult challenge for the ENSO forecast applications and impacts communities. For those concerned about the correlations of cold events with various human activities (agricultural production, fish catches, damages associated with extreme events such as droughts, floods, hurricanes, etc.), too many definitions can be confusing to the unsuspecting public. In addition, the use of any single, potentially incomplete, or unreliable list of La Niña years will likely produce statistically misleading results about impacts, applications and forecast value. Finally, like El Niño events, La Niña can be weak, moderate, strong, very strong, and extraordinary. However, it appears that what constitutes the quantitative differences among these levels of "strength" of a La Niña event has yet to be agreed upon. For example, there are differences of opinion, as to whether the 1988-89 La Niña was a moderate or strong event, although most observers seem to agree that the 1995-96 La Niña was a weak event. Several participants underscored the fact that the climate is not stationary but is constantly changing on a variety of time scales from seasonal, to interannual to decadal. Today, there is considerable speculation about how the ENSO "cycle" might be affected by the human-induced global warming of the atmosphere. That there have been twice as many El Niño as La Niña events since the early 1970s has been cited by some researchers as an early impact and sign of global warming. However, according to one of the participants, the reconstruction of the SST time series back to 1867 indicated that over this period of time there has been about an equal number of these changes in SST in the equatorial Pacific. To some of the scientists at the workshop a precise definition of a natural phenomenon, such as La Niña or El Niño or more generally the ENSO "cycle," was not viewed as an important part of the problem associated with forecasting and forecast application. They argued that more knowledgeable users were not likely to focus on lists of La Niña or El Niño years but, as noted earlier, are likely to use relevant time series. Others, however, pointed out that most users would rely (at least initially) on a definition of La Niña. Therefore, they would have to learn how to interpret the various time series related to the ENSO cycle (SSTs, SOI, etc.). Key Points:
What Constitutes "Normal"? Joseph Tribbia (NCAR) was asked to start the discussion on what constitutes normal in the context of the ENSO cycle. The discussion was focused less on what constitutes the climatological normal and more on what one thinks of as normal, or is perceived to be normal. For discussion purposes he presented the following four questions: 1. "Normal is what we expect" Expectations about the state of sea surface temperatures in the tropical Pacific Ocean, and about the climate-related impacts that changes in that state bring about, do not necessarily match well with what actually takes place. However, actions taken that are based on one's perceptions of expected changes in oceanic or atmospheric conditions, expectations prompted by forecasts of El Niño, will be real and those actions will, therefore, have real impacts on and consequences for society. Normal does not mean that the tropical ocean is in a fixed state (considered as the statistically calculated average) most of the time. Variations in SSTs are to be expected and in fact are part of the normal process. However, societies often tend to minimize (i.e., discount) the possible occurrence of what they consider, rightly or wrongly, to be rare or unusual events. Instead, they tend to consider only the middle section of the bell-shaped distribution curve as normal, thereby omitting totally their consideration of the "wings" of the bell-shaped curve. Hence, they set themselves up to be surprised when an ENSO event does occur. It was suggested that there was a number of surprises related to the ENSO cycle of warm and cold events that ought to have been expected, given the knowledge about ENSO that existed at the time. 2. "Normal might not be likely" Looking at SST anomalies in the tropical Pacific one might get the impression that there is a normal (i.e., average) condition to which the temperature state of the ocean returns. However, if one uses the average SST as an indicator of normal, one could argue that the tropical ocean's sea surface temperature is seldom normal, because it is most often in a transitory state passing from warm to cold or cold to warm conditions. If we expand the definition of what constitutes normal SST from the mathematical average to a range and consider that range as normal, the possibility has increased that SSTs are in that range. A definition of what constitutes normal with respect to SSTs in the central Pacific in general and El Niño and La Niña specifically is clearly warranted. 3. "Symmetric variations in SSTs do not mean symmetric anomalous impacts" Typical variations around the norm in tropical Pacific SSTs do not (in general) generate normal (i.e., bell-shaped) variations in impacts at distance locations. The influence of tropical SSTs on remote meteorological quantities such as precipitation and temperature is not linear. This means that the strength of their influence does not scale in a proportional fashion. Thus, the impacts can have large magnitude, even though the SST anomaly may only be of moderate magnitude. Furthermore, the societal and economic impacts of variations in precipitation in a given location are also frequently non-linear, making the impacts on key sectors of society, such as agriculture and fisheries, highly skewed with regard to benefits that might be accrued or losses that might be incurred. 4. "Extremes are part of normal" Even under average (i.e., normal) conditions in the central equatorial Pacific, extreme meteorological events will occur around the globe. Droughts tend to occur with a particular frequency, even when nothing unusual or anomalous is taking place with respect to tropical Pacific SSTs. With a normal forcing of the atmosphere, there will still be a wide range of variability in climate anomalies around the Pacific basin as well as in the tropics and extra-tropics. Thus, one ought to expect a certain amount of extreme meteorological activity to occur during "normal" times in the tropical Pacific. This, of course, creates a problem for those seeking to attribute (link in a cause-and-effect way) a distant climate anomaly or its socio-economic impacts to specific SST conditions in the tropical Pacific at a given point in time. In sum, extreme meteorological events can and do occur all the time during the full range of SSTs in the central Pacific. A participant pointed out that the short record of observations does not allow for a reliable definition of what constitutes normal SSTs in the tropical Pacific Ocean. With a short climate or SST record, expectations can be distorted by using only that portion of the record that we have observed directly. That short record does not capture the decadal-scale fluctuations that are known to occur in the ocean environment. Thus, normal fluctuations in meteorological terms do not translate easily (or directly) into societal impacts of importance. It was suggested that users of ENSO information should be taught not to seek only information on the average conditions but should also seek information on the standard deviation. The public can and must be taught how to understand and use probabilities. Deviations from normal conditions (however defined) can allow for prediction of potential impacts. However, it would be wrong to attribute every drop of rainfall that fell in a particular region in the 1997-98 period to the recent El Niño, even if there is a strong teleconnection to that region. Some rain would likely have fallen there anyway. However, above-normal precipitation in a particular season can likely be attributed to El Niño. A participant suggested that the issue of what constitutes normal is a psychological problem shared by individuals, sectors as well as societies. It is not a scientific one. Further, it was suggested that in the US, there is considerable interest in climatic conditions that are considered to be unexpected or abnormal. That is why the media reports and societies take seriously such statements as "this is the biggest rainfall in six years", or "the warmest winter in ten years". People also tend to discount the past; that is, to forget previous anomalous weather and climate conditions (even fairly recent ones). They are, thus, more susceptible to react to conditions they perceive to be abnormal but which are, in fact, part of what they may have already experienced. It was suggested that society will become more reliant (and dependent) on large-scale computer modeling activities to provide insights into what might be expected to result in and from air-sea interactions in the tropical Pacific. The limited period of observations and the shortness of truly reliable time series of SST, SOI, OLR, among other indicators, gives increasing importance to enhancing the reliability of coupled general circulation models. Key Points of "What Constitutes Normal"
La Niña Teleconnections George Kiladis (NOAA) opened the session on La Niña teleconnections by first loosely defining the term "teleconnections" as referring to "remote influences". More specifically, in the context of La Niña, it is the influence of SST variations in the tropical Pacific on regional and local climate regimes. He then provided a brief overview of the probable physical causes and effects of teleconnections during El Niño and La Niña (Figures 5, 6). Physical scientists are researching how events in the tropical Pacific transmit a signal through the atmosphere and ocean to distant places on the globe. Since tropical convection represents the primary "heat engine" for the global circulation, changes in the location of this convection alters the global circulation, which in turn is manifested as climate anomalies. For example, as convection shifts westward from the tropical Pacific into the Indian Ocean during La Niña, the jet stream over the Pacific is weakened, thereby affecting the downstream wave activity (high and low pressure weather systems) moving into North America. Kiladis provided a list of La Niña and El Niño years in order to identify statistically likely impacts of La Niña around the globe. This was done by the method of "compositing", or averaging the anomalous temperature and precipitation over all La Niña (or El Niño) events, regardless of their magnitude. In this way, the statistical likelihood of a particular teleconnection can be calculated, thereby giving a measure of the probable impact of a La Niña event where the signals are strongest. Kiladis first showed how islands in the equatorial Pacific Ocean are affected by changes in SSTs. During La Niña, these islands experience drier than normal conditions due to the stabilizing effect of cold SSTs. It was pointed out that these signals might not really be considered teleconnections in sensu strictu, since they are directly influenced by the local SST, rather than remote forcing. He then introduced one measure of the robustness of the observed signals, which is based only on the percentage of time a meteorological station had above or below normal temperature or precipitation, compared to the "expected" sign of the teleconnection signal according to the composites. It was demonstrated that the equatorial Pacific islands had very robust signals, with nearly all La Niña events drier and all El Niño events wetter than normal at these locations. Several maps were then presented depicting the global temperature and precipitation composites for both La Niña and El Niño events. It was suggested that there was some degree of linearity between El Niño and La Niña impacts over many regions, meaning that cold events in several locations produced the opposite climate anomalies to those occurring during warm events. However, the reliability of the teleconnection signals becomes less as one moves farther away from the tropical Pacific, the "center of action" of ENSO. Thus, even though a given teleconnection might still be defined as "statistically significant", and almost certainly related to the entire population of warm and cold events, the probability of that signal occurring during any one event may not necessarily be very high due to the large amount of climate "noise", or random fluctuations in the atmosphere. This is especially true in the extratropics, where large "internal climate variability" is dominant, as opposed to the tropics, which is to a much larger degree determined by SST. In addition, weak and moderate La Niña events might not be strong enough to generate climate anomalies in distant locations. Kiladis then discussed what he considered to be among the more robust La Niña teleconnections. These included a tendency for wetter than normal conditions, with a risk for flooding, in southern Africa and the monsoon regions of India, Indonesia, and northern Australia, and drier than normal conditions, sometimes leading to drought, over eastern Africa, the western equatorial Indian Ocean, southern South America, and the southern Plains and southeastern portions of the U.S.. In general, tropical surface temperatures tend to be below normal, with robust signals even as far away from the tropical Pacific as Africa. The most pronounced extratropical temperature signals during La Niña are seen over North America, where there is a pronounced tendency for colder than normal conditions over Alaska, western Canada, and the central Plains of midwestern Canada and the northern United States, and warmer than normal tendencies over the southeastern United States. Discussion was initiated concerning the tropical temperature signal. Kiladis showed that far-field tropical SST anomalies of the same sign as those in the Pacific developed in the Indian and Atlantic Oceans three to six months following the onset of both warm and cold event conditions. These remote SST anomalies were in phase with the observed tropical surface temperature anomalies, even over the tropical continents, which also lagged the equatorial Pacific SST by the same amount of time. Kiladis pointed out that, if one could explain the far-field SST signal, one could then also account for the tropical temperature signal, since over the ocean surface air temperature follows the SST very closely. He suggested that anomalous surface solar heating because of changes in cloudiness during warm and cold events was a probable factor. Peter Webster (University of Colorado, PAOS) then commented that the monsoon circulation was also a likely player, through observed changes in the intensity of its circulation during warm and cold events and the impact of these changes on the heat budget of the ocean, at least in the Indian sector. It was agreed that this was an important area for future research. As a cautionary note, Kiladis cited a nearly 20-year-old article by Colin Ramage entitled "Teleconnections and the Siege of Time". Ramage referred to the fact that many of the time series used in teleconnection analyses are of relatively short duration, and that teleconnections identified as being robust during one epoch may fail completely during a later epoch. While some of this might be attributed to the statistical fragility of using short samples, there could also be long-term changes in the climate system itself which could alter the response of the atmosphere to SST anomalies. Thus, forecasts based on established La Niña teleconnections, even those considered highly statistically significant, could fail or even reverse sign in the future due to decadal time scale climate variability. Finally, Kiladis also noted that the linearity, or the reversal in the sign of anomalies in the same location between La Niña and El Niño, exists to some extent on the large spatial scale of his maps, but would likely break down with increased spatial resolution in many regions. This would result from, for example, the effect of mountains or of proximity to the ocean on the local climatic response to global-scale atmospheric circulation changes.
Key Points about Teleconnections:
The Impacts of La Niña on Specific Countries and Sectors Several participants were asked to present their thoughts in brief discussion papers that focused on country- and sector-specific information about La Niña and the ENSO cycle in general. The papers in this report are as follows: Climate Change and the ENSO Cycle: El Niño, La Niña and Normal Kevin Trenberth opened the session on climate change and the ENSO cycle. He noted that there is considerable (and still growing) scientific and public interest in and concern about the possible impacts of global warming on El Niño events, and vice versa. Some studies have suggested that (1) the unusual behavior of the tropical Pacific's sea surface temperatures in the first half of the 1990s (considered by some observers to have been one continuous El Niño from 1991 to 1995) and (2) the apparent climate regime shift after the mid-1970s are the result of influences of human-induced global warming of the atmosphere. Others have challenged that view, arguing that there is not enough evidence as yet to identify with certainty the linkages between these two processes, one natural and one possibly human-induced. Because ENSO is involved with the movement of heat, it is conceptually easy to see how increased heating from the buildup of greenhouse gases can "interfere." Trenberth and Hoar (1996, 1997) wrote articles suggesting that El Niño, a natural process, was being influenced by climate change, and possibly human-induced global warming. Their research suggested that something was happening to El Niño. According to Trenberth, "What no one, including us, has been able to say is how global warming is influencing El Niño. . . . While we can say we have detected a change, we cannot do the attribution, we cannot say which part of what we are seeing is due to global warming, although there is a very strong case that some part is. One reason is that the models do not agree and do not adequately simulate El Niño. There is no clear signature that we know to look for." A Greenpeace-sponsored review of this issue, prepared for the Kyoto Conference of Parties to the Climate Convention in December 1997 could not draw a conclusion about the links between global warming and El Niño events. However, the review could not rule out such a linkage. It is highly likely that Greenpeace had hoped to be able to find conclusively scientific consensus on a positive linkage between these two processes, but was unable to do so after reviewing the existing scientific literature and interviewing members of the ENSO and climate change research communities. These issues are as yet unresolved. Different views exist over the possible connections between El Niño events (their characteristics and their impacts) and human-induced global warming. In recent times there have been a few reasons for this: as noted earlier, in 1996 and 1997, Trenberth and Hoar published articles suggesting that the recent changes in ENSO behavior were not expected given the behavior in the 100 years from 1880 to 1980. Then in October 1997 an El Niño Summit was convened in California a few months in advance of the Kyoto Conference of Parties to the Climate Convention. During the El Niño Summit, U.S. Vice President Gore spoke of the likely impacts of global warming on El Niño. He suggested that El Niño's characteristics would strengthen and impacts would worsen with global warming. Given the heated political disagreement over the climate change issue, the linkage between El Niño and global warming had to some extent become a politicized issue as much as a scientific one. Kevin Trenberth opened discussion with a brief presentation on the climate change/El Niño/La Niña issue. He raised the following set of questions: How have global temperatures changed? How has El Niño changed? What is the role of El Niño in global temperature changes? Can we distinguish between decadal scale variability and anthropogenic effects on global temperatures? How helpful are the paleo records? How useful are the general circulation models? He noted that 1997 had been the warmest year on record and that now there has been a discernible human influence of global climate, citing the 1995 IPCC report. He then noted that the first six months of this year (with El Niño peaking and winding down) have been warmer than any comparable period in the climate record. He suggested that something unusual was going on and that some of this year's global temperature increase could be related to El Niño, because following El Niño's peak there is a "mini-global warming" that takes place. This effect is partly due to the cooling process that goes on in the tropical Pacific during El Niño's decay phase during which much of the heat lost from the ocean goes into the atmosphere. A look back at the 1982-83 events shows that an even bigger warming followed the decay of that major event. Trenberth then discussed atmospheric temperature trends in the past several decades comparing trends in the 1982-97 period with those of the 1951-80 period. He noted that El Niño events in the recent past have been bigger and more frequent. He also noted that there had been fewer La Niña events in this period. Trenberth commented on the reliability of SST data. He suggested that perhaps researchers should be using SST changes in the Niño3.4 region, as they appear to be more closely correlated with Southern Oscillation Index (SOI). However, much of the Pacific Ocean data had been taken from ship tracks, and before 1950 such recordings were relatively less reliable. Thus, there has been a problem getting good data prior to 1950, but what is needed for ENSO research is a reliable homogeneous time series. He also noted that using the SOI alone as an indicator of change is not sufficient to provide information about changes in SSTs in the different Niño regions or throughout the tropics. Another approach is to use models. In this case sampling is not a problem but the verisimilitude of the simulation of the mean and variability in the Tropical Pacific is critical. At this point, no global climate model is able to simulate El Niño events as realistically as desired to build confidence in the results. Several models, but not all, suggest increased warming in the eastern Pacific, so the mean climate becomes a bit more El Niño-like. Some models suggest stronger El Niño-La Niña swings may occur. So while climate models certainly show changes with global warming, but none simulate El Niño with sufficient fidelity to have confidence in the results. How clouds might change, especially the brightness of convective clouds, is especially uncertain and can influence the outcome. So the question of how El Niño may change with global warming is very much a research topic. The instrumental observational record is not really long enough (about 120 years for the Southern Oscillation Index) to sample all of nature's variety, especially with regard to decadal variations. One approach to this then is to use proxy data, such as from tree rings, and the annual layers of coral or tropical glaciers. Indeed some reconstructions have been made using these methods, and they show El Niño variations quite nicely. However, their decadal variability is compromised by non-climatic effects such as growth of trees, or coral colonies, and biological influences, and so there is a need for multiple reconstructions using nearby cores to discern the common signal, which is likely to be from climate, from the spurious component related to the individual core. Another problem with proxy data, is that the actual climate is changing, or instance the Little Ice Age occurred, and so the results do not simply give the variability that occurs in an unchanging climate. He summed up by noting that there was a very low probability, according to his calculations and methods of analysis, that global warming was not influencing El Niño. Because ENSO is involved with movement of heat, it is conceptually easy to see how increased heating from the build up of greenhouse gases can interfere. One reason for the unusual behavior of El Niño in the past twenty years may be that the Warm Pool in the tropical western Pacific is becoming larger, the recharge phase of El Niño is apt to be faster and/or the heat loss phase is less efficient, all of which would suggest more El Niño events. With greater warming in the upper layers of the ocean, the vertical temperature gradients in the thermocline could sharpen, potentially increasing the magnitude of ENSO events. He concluded by suggesting that we should expect big changes in El Niño's behavior but that we do not know what those changes will be or how they will affect worldwide atmospheric teleconnections. In the discussion a question was raised about whether, with La Niña, there might occur a "mini-global cooling," paralleling the mini-global warming noted earlier. Trenberth said yes but that it would have a minor (about a tenth of a degree) and short-lived effect on global atmospheric temperatures. Concern was raised about global warming attribution: how could one rule out that the global warming we are currently witnessing was not due to natural variability (on decadal scales) as opposed to increased emissions of greenhouse gases? Trenberth suggested that detection studies (one by him and his colleagues) have ruled out that all of the changes in the time series can be attributed to natural variability. This point remains controversial, however, as other participants contended that the use of other methods of analysis might not yield the same finding. There was considerable discussion about the movement of heat in the atmosphere and in the ocean, as an El Niño event decays and a La Niña event emerges. It was suggested that there was a need to expand the TAO array to include coverage of areas outside the near-equatorial region (i.e., the wave guide). Key Points about Climate Change, El Niño, La Niña, and Normal
The Symmetry Issue Martin Hoerling (NOAA - University of Colorado, CIRES Climate Diagnostic Center) identified that the question of symmetry in the physical processes behind El Niño and La Niña events is both interesting and important. One sometimes hears the question "Are there only two states in the tropical Pacific Ocean: El Niño and non-El Niño (e.g., climatic conditions perceived to be normal)?". Or are there three states, namely El Niño and La Niña, and a so-called normal condition? Scientific evidence is convincing in demonstrating that the tropical Pacific Ocean undergoes an oscillation, albeit irregular, between warm El Niño conditions an cold La Niña conditions. This variability occurs relative to a climatological mean, or normal ocean state. Sometimes, this normal state is little more than nature's brief respite between transitions from El Niño to La Niña. The normal state is, however, the most common occurrence for the sea surface temperatures (SSTs) in the sense that a frequency diagram of the SST anomalies shows the most likely occurrence to have near-zero anomalies. The overall distribution resembles a bell curve, or Gaussian, shape, and the El Niño and La Niña events reside in the less-frequented tails of that curve. But is that distribution a perfect bell shape, or is it somehow skewed? For example, can the sea surface temperatures become anomalously colder by as many degrees as they can become anomalously warm? Furthermore, does the atmosphere, by which we refer to the rainfall, pressure and wind systems, respond symmetrically with respect to El Niño and La Niña? In other words, is the atmospheric response to El Niño merely the mirror image of its response to La Niña? Does the atmosphere react twice as strongly if the ocean temperature anomalies are doubled? To the extent that answers to the above are in the affirmative, one would say that there exists a linear relation between the SST anomalies and their climatic impact. This, of course, would greatly simplify our ability to predict the expected effect of tropical Pacific SST anomalies on climate, but the problem is somewhat more complicated than this simple linear symmetric view of the world. Hoerling then presented an overview of the issue of symmetry between El Niño and La Niña, using as a tool the results from several sophisticated numerical models of the atmosphere and their sensitivity to ENSO extremes. He noted that George Kiladis, in the Teleconnections session, had summarized 120 years of climate data, of which nearly 50 were classified either as El Niño or La Niña years. His point was that the results from such an empirical analysis would tend to yield a linear view of the atmosphere's response by default, since the data record was virtually split into two equal parts. His purpose was to study the atmosphere's sensitivity in the extreme tails of the ENSO bell-shaped distribution, since it is the extreme events that undoubtedly exert the largest impact, and for which any departures from symmetry would be critical to understand. Needless to say, the observational data offers only a few such cases, though the 1997-98 and the 1982-83 northern winters were during the strongest warm events in the instrumental record. The last extreme La Niña was during the winter of 1988-89, and before then 1973-74. Hoerling provided some examples from the observational data alone: if you separate out the events according to their varying intensities, one finds that the weaker events tend to behave in a linear fashion. as suggested in the overall analysis of Kiladis. It is with the larger ENSO events that one tends to see asymmetry or nonlinear behavior in atmospheric and oceanic anomalies, although those assertions were derived from Hoerling's analysis of only the 4 cases noted above. His main point was that in several parts of the globe, including North America and even the tropical Pacific, the wintertime (defined as December, January and February) climate anomalies associated with the extreme La Niña events were weaker than their El Niño counterparts. Also, the spatial patterns were not mirror images of each other, but instead the centers of maximum anomalies over the Pacific-North American region were shifted several thousand kilometers with respect to one another. These empirical results cast some doubts over the notion of simple symmetry although, as Hoerling noted, only a few cases could not provide statistically significant information, although they suggested a course of numerical experimentation the highlights from which will be discussed shortly. To the media and to the public, it appears that what transpired in the most recent major events can serve as the baseline as to what one should expect to happen in the next event. This is a problem of "discounting the past". That is, when people tend to weight more heavily the events that occurred more recently than those that occurred further back in time. Another way to look at this problem relates to the use of analogues. Researchers, and then the media, stated that the 1997-98 El Niño was like the 1982-83 El Niño. Thus, by inference we are to expect that the behavior of the tropical Pacific Ocean and its distant impacts around the globe will likely be similar to those that had occurred at that time. This could be referred to as a form of anchoring, where our perceptions and expectations are tied to a previous notable event, (in this case the 1982-83 event). In future years it is likely that such anchoring will shift to the most recent extraordinary 1997-98 event in lieu of the 1982-83 event. With regard to the La Niña event that has been forecast for winter 1998-99, researchers and the media have apparently anchored the public's memories to the La Niña that occurred in 1988-89. There is also the underlying notion of linearity, and so we are likewise witnessing anchoring to the 1997-98 event, but the expectation is that impacts will assume the opposite sign. In both cases, anchoring can produce very misleading results. There have been relatively few La Niña events in the past few decades and the impacts of the few events that we have observed are not necessarily reliable indicators of what might be the meteorological impacts that accompany a future La Niña event. Thus, anchoring can yield inappropriate responses to La Niña also. Hoerling presented results from new general circulation modeling experiments related to the issue of ENSO symmetry. He focused on the strong El Niño event of 1982-83, and the major cold event of 1973-74. In each case, the observed tropical Pacific sea surface temperatures were specified as a global boundary condition, and the atmospheric model was run in order to study climate sensitivity. In order to overcome the problem of statistical sampling which plagued the observational work, the experiments were repeated 10 times using the same SSTs, but each run begun from a different initial atmospheric condition. A unique aspect to the study was the execution of a parallel suite of model runs. In these, the signs of the 1982-83 and 1973-74 SST anomalies were reversed in the tropical Pacific and, once again, ten simulations of the atmospheric model were conducted for each. Thus, a perfect mirror image of the SST anomalies was created for the two ENSO extremes, and a precise analysis of atmospheric symmetry was undertaken within the idealized scenario that the SST anomalies were themselves perfectly symmetrical in sign. The results tended to confirm the impression gathered from the observational analysis. In particular, the atmospheric response to strong El Niño forcing is greater than it is to equally strong La Niña forcing, and their respective spatial patterns of atmospheric responses are shifted. In sum, Hoerling pointed to several physical factors that could account for this asymmetric behavior: 1) Under normal sea surface temperature conditions in winter, extreme cold water exists in the central and eastern equatorial Pacific Ocean, while the pool of warm surface water is confined to the western Pacific. Owing to surface warmth in the western Pacific, the surface trade winds tend to converge there and fuel an abundance of convective activity, whereas the cold tongue area is a comparative dry zone. During an El Niño, warm water covers a large expanse of the equatorial Pacific, and during an extreme event such as 1982-83 enhanced convective activity (as measured by outgoing longwave radiation [OLR]) occurs over the central and eastern Pacific. This has a major impact on the location of storm tracks across the North Pacific and North America which are instrumental in determining temperature and rainfall regimes during winter. During La Niña, an anomalous cooling of the cold tongue region only achieves a modest further reduction of rainfall in the already arid zone of the eastern Pacific. Further cooling during extreme events yields no further affect on the local convective activity (which by then has already been virtually shut down), and thus there is an obvious saturation effect of the climate response to La Niña which does not occur in response to El Niño. 2) There is a remote influence of these disruptions in the equatorial Pacific rainfall regimes that is apparent in the middle troposphere, for example the flow of air on the 500-mb pressure surface. Hoerling showed that there is roughly a linear increase in amplitude on the 500-mb response over the Pacific North American area for successively larger-amplitude positive sea surface temperature anomalies, consistently low pressure over the central North Pacific, and North American high pressure. He also showed evidence for a modest eastward shift of this wave pattern for increasingly larger positive SST anomalies. Thus, the North Pacific low tended to be positioned closer to the Pacific west coast for stronger El Niño events. On the other hand, stronger La Niña SST forcing did not lead to a further response of the 500-mb response. There is an initial linear response, and then the relationship becomes insensitive, consistent with the saturation of the tropical rainfall response. It was noted also that the climate system was much more deterministic with respect to tropical Pacific rainfall: a single event produces a reliably detectable signal across the tropical Pacific. Over North America, however, one needs to witness several El Niño events in order to get a statistical sampling of the climate signal related to the SST forcing, and to be able to distinguish that from the the range of climate states that occur naturally in the absence of El Niño conditions. Many areas of the middle and high latitudes were shown to have little or no ENSO teleconnection signal. Questions were raised about the influence of temperature changes in other oceans, especially those of the Indian Ocean. For example, the influence of ocean temperature anomalies along the East African coast and in the vicinity of the Maritime Continent may be at least of local relevance, as they are believed to be the effects of anomalies adjacent to eastern South American coasts. These have some relation to the typical life cycle of ENSO itself, though not all SST variations owe their origins to ENSO. SST changes in the Indian and Atlantic oceans, and convection processes in the maritime region of Indonesia and Malaysia have been less well studied than those in the tropical Pacific. One participant pointed out that what was very important was the shutdown of convection in Southeast Asia, asserting that we need to better understand the mechanisms that bring droughts to the tropics. While the computer models are good at simulating ocean temperatures in the eastern Pacific, and the enhancement of rain in that area during El Niño, they are less skillful in simulating the impact on the rainfall systems over the Maritime Continent. It was also suggested that there is a need for improvement to model SST changes in the western Pacific, though it is unclear to what extent seasonal rainfall anomalies in the west Pacific are forced by local SSTs, or remotely driven by east Pacific SST anomalies. It was also noted that the activity, strength and location of the ITCZ (Inter-Tropical Convergence Zone) and SPCZ (South Pacific Convergence Zone) have important influences on global climate and, therefore, on the societal impacts of La Niña and El Niño. The ITCZ was more active during the cold event of 1988, and it remains to be determined what, if any, effect that may have had on the US Midwest drought in the spring and summer of 1988. Participants also referred to the PNA circulation pattern that consists of alternating high and low pressure arching across North America, and its sensitivity to ENSO. This pattern is an important determinant of US weather and climate, though it was pointed out that its sources of excitation are more varied that just El Niño. Key Points:
Symmetry in La Niña's Societal and Environmental Impacts Statements about symmetric changes in the societal and ecological conditions between El Niño and La Niña must be challenged. The impact of drought is quite different from the impact of flood, for example, and each yields its own unique problems concerning human health. Each of these extreme climate-related events prompts different national responses. Furthermore, the timing of such climate extremes during different times of year, and their relation to the local rainy season are important issues in planning and mitigation. As a first approximation, the notion of symmetry with respect to ENSO's extreme phases provides the public with information regard to possible impacts. However, the science and the expectations of society have matured to the point where more scrutiny of the notion of symmetry in ENSO impacts is both feasible and necessary. It is necessary and more useful to talk about societal impacts in terms of tendencies and probabilities of occurrence. Several analyses of impacts based purely on historical observations were reviewed at the meeting. They depicted El Niño and La Niña impacts around the globe as being of opposite sign. But have there been enough La Niña events to make such an assertion? In fact, there have been relatively few La Niña events since 1976. Besides its purely scientific merits, determining the level of symmetry in the behavior and impacts of ENSO will enable decision makers to better prepare for and mitigate the negative impacts of ENSO, and also to capitalize on the positive effects that such phenomena have on society and economies. Glantz opened the session on symmetry in La Niña impacts by drawing attention to the sharp increase in El Niño related websites on the Internet. He suggested that there was a need to develop a way to sort out these sites, as there is no way for "surfers" of the Internet to distinguish among these sites with regard to their reliability of information on ENSO. This issue led to debate, with some participants loudly opposing any form of "sorting out" the websites according to someone's set of criteria. He noted that there were some websites that wrote about El Niño and non-El Niño as if there was no need to distinguish between El Niño and La Niña (or extreme cold) events. The reason for this most likely stems from the fact that in some locations around the globe the impacts of a cold event are perceived to be no more than a severe case of what might occur in a normal year. Factors that cast a shadow on the claim of symmetry include, but are not limited to, the following: scientists do not agree on a precise list of La Niña or El Niño years; composite maps are misleading in that they average out the impacts of weak events with the strong ones; a reliance on a previous El Niño event as an analogue to identify or forecast impacts is also misleading because the set of impacts from one event to another will likely differ, depending on the influences of regional and local climate processes around the globe; and the number of observed events is too small as yet for reliable scientific projections. However, there are some locations around the globe where the societal impacts of El Niño and La Niña appear in general to exhibit some "usable" degree of symmetry: for example, Uruguay, Argentina, Brazil, Australia, Indonesia, Papua New Guinea, Peru, Ecuador, Chile, the US Gulf states, Central America, western Canada, and southern Africa. Glantz pointed out another problem that can confound an evaluation of symmetry in societal and environmental impacts: there are some conflicts, as noted earlier, among researchers as to which years were cold event years or which were warm events years. When seeking to determine whether there are symmetric environmental and societal impacts around the globe between warm and cold events, it is imperative to know this information. Otherwise, there may be no way to reliably address the issue of symmetry of societal impacts between the two extremes of the ENSO cycle. Many articles in the scientific and popular literature rely on the use of composite maps to identify areas that are likely to be affected by warm and cold events. However, many of those maps are not independently derived but have been reproduced from the maps first produced in 1987 and modified in the early 1990s by Ropelewski and Halpert (1987). The original maps are now more than a decade old and, as one would expect, considerable data needs to be added, given the occurrence of La Niña and El Niño events in this period. A close look at these maps shows that the impacts of cold events were considered to be of the opposite sign of the warm events; warm regions in El Niño get cold and wet regions get dry during La Niña (Figure 5). Many of the maps showing impacts are composites and, as noted earlier by Hoerling, and composites hide detail. They combine all magnitudes of ENSO events into one composite and what one sees is an average of impacts associated with La Niña (or with El Niño)(Figure 7). Thus, those concerned about the possible impacts of El Niño or La Niña in southern California for the 1997-98 El Niño event might not have paid much attention to early warnings of severe flooding in southern California, if all they had to rely on were these composite impact maps. The impact maps are characterizations of La Niña or El Niño impacts; they are suggestive and illustrative (but not definitive) of potential impacts. They should serve only as a starting point for impact assessments. Forecasting the impacts of La Niña or El Niño would be much simpler, if symmetry dominated the effects of ENSO -- but "it just ain't so." The composite maps of ENSO's global impacts should carry a "buyer beware" label. Perhaps one can speak of symmetry between the societal and environmental impacts of El Niño and La Niña in much the same way that one refers to the greenhouse analogy when discussing the greenhouse effect with regard to the atmosphere. While the analogy of the atmosphere as a greenhouse has its beneficial uses (e.g., educating the public), it is inaccurate and misleading if carried too far. The same applies to the notion of the symmetry of ENSO impacts on environment and society. Key Points:
Attribution of Societal and Environmental Impacts to Specific La Niña and El Niño Events Just about everything that happened between September 1997 and May 1998 that was either unwanted or unexpected was blamed either directly or indirectly on El Niño. The ice storm (Figure 8) in the northeast of North America in January 1998, the killer tornadoes in Florida and the scores of human deaths that ensued, the flooding in southern California, and the drought in Texas were linked by one observer or another to the 1997-98 El Niño. However, while few challenge the linkage between seasonal anomalies in climate in these regions and the ENSO cycle, many of the claims that the impacts of specific anomalous events can be attributed to a particular El Niño or La Niña must be challenged. The major Midwest drought in 1988, a drought whose impact was estimated at about $40 billion, has been attributed without much question to the 1988-89 La Niña. The 1988 Midwest drought-La Niña connection is now being challenged, as researchers identify other plausible climate scenarios that could produce droughts of similar magnitude in the region. There are several efforts under way by numerous organizations (NOAA, WMO, IDNDR, International Red Cross, CARE, etc.) to calculate the "cost" of the 1997-98 El Niño event. To do so, however, a rigorous method must be developed in order to discriminate among impacts clearly associated with El Niño and those that fall into a grey area. This would give credibility to the dollar numbers generated about the cost of an El Niño event. The same logic and problems exist with regard to costing out the positive and negative impacts of La Niña. Gerald Meehl (CGD, NCAR) opened discussion on ENSO attributions by identifying two major issues. The first type of attribution is to determine the physical mechanisms in the air-sea interaction that cause La Niña events. This involves a search among the varied mechanisms of Pacific ocean-atmosphere coupling, and those mechanisms are captured to various degrees in the coupled climate models used to forecast El Niño/La Niña events. The physical attribution of cause and effect was left to other discussion sessions. Instead, the focus was on a second type of attribution. This concerns the type of climate anomalies or impacts that could be attributed to La Niña. Here we separate the manifestations of occurrence of El Niño/La Niña, from the climatic impacts of individual events. For the former, a number of studies have documented low frequency decadal timescale variability in the tropical Pacific that could affect the manifestation of El Niño and La Niña impacts. Thus, the record-setting El Niño of 1997-98 could be characterized as having a contribution from 1) coupled processes on the interannual timescale, 2) a warm phase in the decadal oscillation, and 3) a longer term warming trend in the eastern Pacific. These effects, superimposed one on another, could combine to produce the extraordinary El Niño that occurred in 1997-98. Similarly, these factors could affect manifestations of individual La Niña events. We could think of attributing El Niño/La Niña occurrences to a combination of processes acting on different time scales. In the latter category -- how we can attribute seasonal mean climate anomalies to individual El Niño or La Niña events -- the concept of a seasonal mean anomaly as a collection of meteorological events was developed. The quantification of this concept involves a shift of the probability distribution of certain conditions in a season in a certain region. For example, during a La Niña, there is a greater chance of dry conditions over the southeast U.S. during winter than during non-La Niña years. There is the strong, almost irresistible temptation to attribute individual meteorological events to El Niño or La Niña. Though this runs counter to the concept noted above of a shift in probabilities for seasonal climate anomalies, a forecast study has been performed to see if individual meteorological events could be attributed to the El Niño of 1997-98. Results from that study indicated, for example, that the 1998 ice storm in the northeast United States was probably intensified from atmospheric forcing from the tropical Pacific, that California's rains in the spring of 1998 could probably be attributed to forcing from the tropical Pacific, but that the cause of the Denver blizzard of October 1997 was inconclusive and could not be attributed with confidence to El Niño. Glantz commented on the chain of attributions linked to the ice storm in the northeastern part of North America in January 1998. Figure 8 was used to ask participants to decide when attribution of impacts to El Niño should no longer be considered reliable. There was some agreement that the best way to view attribution was to think of the forcing from the tropical Pacific as having the effect of shifting the probabilities of collections of meteorological events over the course of a season towards wetter or drier, warmer or colder. These effects are particularly attributable in certain seasons in certain regions identified from previous events as susceptible to being affected during either El Niño or La Niña events, (e.g., a shift toward wetter or drier than normal conditions). It is very important that attributions to La Niña events be reliable, because actions taken by societies will likely be taken based on the belief that La Niña events tend to spark certain kinds of adverse conditions, such as Midwestern drought conditions in the summer of 1988. If such an attribution proves to have been incorrect, societies will have wasted scarce resources to respond to unlikely adverse events. Key Points of Attribution:
The Role of the Media in ENSO Reporting: A Media Panel Participants on the Media Panel included Bill Meck (WTHR TV, Indianapolis, Indiana), Gary Robbins (Orange County Register, California), Joe Virengia (Associated Press, Denver Bureau), Madeline Nash (Time Magazine), Mary Miller (Exploratorium, San Francisco), and John Kermond (Office of Global Programs, NOAA, Washington, DC.). Media coverage of the 1997-98 El Niño is currently being reviewed and assessed by government agencies, researchers, media specialists and by the media themselves in several countries. One of the major problems has been a concern about the "hype" aspects of reports and especially of the headlines on El Niño that have been used to attract attention to specific articles in newspapers and magazines, or sound bites on TV. In one specific instance, for example, El Niño had been described by one organization on various occasions in November 1997 as growing, decaying, and growing again within a span of a few weeks. These rapid changes in forecasts of the fate of the 1997-98 El Niño were viewed in Peru on the Internet and caused considerable confusion. Sensational, often misleading, headlines created a "doom and gloom" perception of El Niño. It got to a point where newscasters had to report something -- anything -- that El Niño had done recently. El Niño didn't stand a chance; it was blamed for just about everything that took place while it was in progress, and beyond that time frame as well. It was pointed out that the media do not speak with one voice nor do they operate according to the same ethical standard. In fact, each story appearing in the media may be treated by a different set of rules. The result is that the media are varied in the way they report on the El Niño and La Niña phenomena and on their environmental and societal impacts. Editors differ as well in the way they treat the writings of their scientific writers. While some seek to present the ENSO situation so as to include the uncertainties of the science or the impacts, others tend to rely more on the sensational and speculative aspects. It was mentioned that the press covers the news, not necessarily good things. Thus, there has been a tendency to report El Niño (and now La Niña) in negative terms, e.g., to report about El Niño in terms of "doom and gloom." This can be seen from the headlines of various El Niño (Figure 9) articles and magazines. One participant noted that the headlines used to introduce El Niño stories in the printed media tended to track well the changes in the attitude of the press toward the phenomena, as it developed in 1997 to 1998, from the forecast of the onset of an El Niño to its demise and the possible onset of La Niña. To capture the broadest audience, the media tend to present scientific information that is geared to the 5th to 7th grade levels [in the US education system]. This can lead to misinformation and misunderstanding about ENSO in attempts to keep the science rather simple and the story readable if not sensational. The panelists identified several problems associated with reporting on La Niña and El Niño such as the following: A serious language (i.e., communication) problem exists between the media and the scientific community. This is because of several factors: very few science writers have a scientific background; there is a tendency of both the scientist and the media to overstate [the media tend to highlight extremes and shifts, while the scientists tend to present their efforts in a more positive light]; it was acknowledged that the media has a penchant for bad news and so there is a tendency for them to highlight adversities rather than to present scientific information they believe will not capture or hold the public's interest; time factors also provide pressure on the media to produce a story early if not first. Time is also a problem in the sense that it is difficult for the media to sustain interest in a scientifically based issue; the science reporters and writers are put in the position of writing about a natural process occurring over a relatively long period of time (12-18 months) (i.e., El Niño or La Niña) which the scientific community itself has different views on how to describe or to quantify it. According to media representatives, the scientific community is in a way passing this responsibility on to the media; the media have a problem with providing a balance of views in their story, and with regard to the sorting out of expertise: which "experts" should they trust; a problem exists with the need for the media to report minority views on a scientific issue that are not shared or championed by the scientific community. Media representatives contended that public interest in ENSO prompted continuous coverage of its development and likely impacts. This led to charges against the media that they were hyping El Niño [and were as of mid-July 1998 beginning to do the same with La Niña]. One participant suggested that for all practical purposes "El Niño might as well have been the Spanish word for hype." Several members of the media panel felt that only some scientists were good at dealing with the press and that a large part of the scientific community looked upon or treated the press in a condescending way. One panelist reminded the workshop that the reporter is not to be viewed as a friend to the scientist. S/he is on a mission to produce a news report and to meet a deadline, often set by someone else. If the scientist has a point to make it is up to her/him to get that point across clearly. They try to fact-check the content of their stories but often it is not possible, given the strict deadlines that they have to meet. Often, it is the sensational headline that gets quoted and referred to repeatedly. It is designed to capture the readers' attention even though it may prove to have little to do with the text or tone of the El Niño story. The reporter, however, is not the one who chooses those headlines; it is usually the editor who decides on that headline. The time factor is a very important aspect of producing a story on ENSO. The reporter often does not have time to hold a second interview with the scientist(s) that s/he interviewed at first. There is, thus, a tendency for a reporter to go back at first to the familiar sources that s/he trusts, even if that person is known not to be the best source of that particular piece of information. They do this in order to get started on the story. For some media, the reporter/writer is supposed to see and discuss any changes made to their story by others. This is not always possible however, and as one participant suggested "gremlins make changes in their stories", that is, errors have been inadvertently introduced into the final version by others. A common complaint by scientists with regard to media interviews is that they were either misquoted or that their statement(s) had been taken out of context. Media representatives responded to this point by noting that their stories must fit a space or a time slot constraint and while they do not change a quote, they may shorten it. To get better marks from the public for the forecasts of El Niño or La Niña and its impacts, it is necessary for the press to report these in probablistic terms. However, the media, among others, believe that the public has great difficulty in understanding, let alone using, probabilities. One participant noted that he had been made to look foolish by the media because he spoke in terms of probabilities of occurrence of precipitation in southern California and, when the rains did not occur when expected, the media referred to El Niño as "El No-Show." Yet, the rains did come and were as heavy as predicted. It was proposed that scientists should work more closely with the media to identify ways that they could educate one another. Participants suggested that workshops organized to center on the interactions between these groups be organized as a fruitful way to improve the understanding of the strengths and weaknesses in reporting ENSO and its impacts to the general public. A media person noted that it was necessary to scrutinize their sources more carefully, given the proliferation of web sites on the Internet and no one to oversee the quality or reliability of their content. Which sites were providers of reliable information was a concern to some of these media representatives. The same care must be taken to sort out biases that accompany the scientific information; were the forecasts of the onset of El Niño (or La Niña) made as early or as successfully as some researchers have suggested? Were they as reliable as some have suggested? And so forth. A recent article in Science (Kerry, 1998) suggested that "Models win big in forecasting El Niño" with respect to an advanced warning of the onset of the 1997-98 El Niño. Such a conclusion has not been supported by retrospective assessments of the forecasts (e.g., Barnston and Glantz, BAMS, February 1999). Suggestions about devising a way to rate the reliability of various web sites generated heated debate including comments about a fear of censorship of web sites that presented views in opposition to mainstream views. There are real, legitimate differences among scientists over scientific views and opinions about the ENSO cycle. This makes it difficult for the media which would prefer to report such stories in black and white terms. There was confusion between forecasting the El Niño phenomenon itself and forecasting its impacts several months following the onset. Claims of El Niño forecast success sometimes relate to the forecast of its impacts, even though the forecast of the El Niño phenomenon was issued about the same time that SST observations showed the onset of an El Niño. These are not the same thing. Successes for the latter (forecasting impacts) do not necessarily translate into successes for the former (forecasting El Niño). The session ended with comments from several foreign participants about coverage in their national media about El Niño (e.g., Australia, Ecuador, Ethiopia). The Australian situation was confused because, in addition to its national by developed El Niño forecasts, the media in Australia were bombarded with forecasts (often conflicting) from the U.S. agencies. It became very clear that the 1997-98 El Niño was a global story that was carried in the media just about everywhere. It seems, however, that few places matched the amount of coverage that was produced by the media in the United States. Media interest in La Niña is following on the coattails of interest in this major 1997-98 El Niño. It becomes clear that there is less known about it, as suggested by the following newspaper headlines (Figure 10) recently devoted to La Niña. Key Points (Media):
How Well Did We Forecast the 1997-98 El Niño? Forecasting El Niño is still fraught with problems. And, to date a reliable long range El Niño forecast system has not been developed. The situation is similar for forecasting the onset of La Niña. In fact some researchers have suggested that La Niña events are more difficult to forecast than El Niño events. Sometimes the sea surface temperatures linger around "normal" and then return to warmer temperatures. Thus, not every El Niño is followed by a La Niña. As of now (early September 1998) the recent El Niño is over and there are different views on whether a strong La Niña is emerging. These conflicts, often played out in the media, are confusing to the public and reduce the potential value of La Niña forecasts. Tony Barnston (NOAA's National Center for Environmental Prediction, climate Prediction Center) presented a review of forecasts of ENSO conditions, based on a review of 15 dynamical and statistical models for the 1997-98 El Niño event and the initial stages of the 1998-99 La Niña. He noted that, while most of the models forecasted some degree of warming one to two seasons prior to the onset of the El Niño in the boreal (Northern Hemisphere) spring of 1997, none had detected its strength until the event was already becoming strong in early summer. Neither the dynamical nor the statistical models, as groups, performed significantly better than the other during this episode. The 2-4 best-performing statistical models and 1-2 of the best-performing dynamical models forecast SST anomalies of about +1°C (versus 2.5 to 3° observed) prior to any observed positive anomalies. He noted that the most comprehensive dynamical models performed better than the simple dynamical ones. Once the El Niño had developed in mid-1997, a larger set of models was able to forecast its peak in late 1997 and dissipation and reversal to cold conditions in late spring/early summer 1998. Because ENSO extremes usually begin to develop in the boreal (Northern Hemisphere) spring or early summer and persist through the following winter, forecasting impact tendencies in extratropical North America for the winter (when impacts are most pronounced) with at 5 months of lead time is not difficult. This requires only good observations of the summer ENSO state and knowledge of the winter teleconnections. Because of the strength of the 1997-98 El Niño and the consequent skill of the 5-month lead forecasts of U.S. impacts in the winter 1997-98, the success of these forecasts was noticed to an unprecedented extent by the media and, hence, the general public. However, forecasting impacts in the austral (Southern Hemisphere) winter that occur simultaneously with the initial appearance of an ENSO extreme (e.g., in Chile, Uruguay, Kiribati, Ecuador, and Peru) requires forecasting the boreal spring/summer onset of ENSO events with several months of lead time. This latter task is difficult, as our performance remains mediocre as evidenced by the fact that formal announcements of a major El Niño did not occur until May or June 1997, leaving little time for users in some of the above regions to prepare for its potential impacts. Barnston summed up by noting that continued effort and computer resources were needed to advance our understanding of tropical and global ocean-atmosphere interactions and their simulation in coupled models. He then drew attention to a review by Glantz of the verbal summaries of ENSO forecasts that had been submitted for inclusion in the Experimental Long-Lead Forecast Bulletin produced by NCEP/CPC. These bulletins were issued to users worldwide before and during the 1997-98 El Niño event. (After the December 1997 issue, the ELLF bulletins are issued by COLA.) They were found to contain verbal ambiguities, when examined from the users' points of view. Given the need for forecasts to be expressed verbally and also to be precise enough for meaningful use and verification, a simple numerically based verbal classification system for describing ENSO-related forecasts was proposed. Key Points About Forecasting
Forecasting the Onset of a La Niña in 1998-99 This discussion session was opened by Nick Graham (International Research Institute/Scripps Institute of Oceanography), who noted that as of mid-July 1998 the trend in changes in SSTs in Niño3.4 was from a warm event toward a cold one (La Niña). He provided overheads comparing SST trends for several La Niña events. He noted that, as of 13 July 1998, Niño3 was average (not showing a warm or cold anomaly, although the trend had been from warm to normal). It was warm in the far western part of the tropical Pacific, and cold SSTs appeared in the central tropical Pacific. He concluded that observations strongly suggested that the SSTs were moving into La Niña conditions. Graham noted that the decay of the 1997-98 El Niño was more like that of 1972-73 than for the 1982-83 event. The 1982-83 event had a slow cooling, whereas the 1972-73 El Niño exhibited a rapid decline in SSTs in the central Pacific. Graham also noted that the different models used different data to initialize their forecast model runs. In response to a question, Graham remarked that some models, such as those of ECMWF and LDEO, initialized with the observed state. He then discussed how various models were faring with regard to the forecasting of the end of El Niño and the onset of a cold event. Using the COLA projections as an example, he showed that COLA forecast a La Niña event in June 1997, but the model's projection had not gotten as cold as observations, as of the time of the July Summit. He then discussed the progress of NCEP, noting it had forecast earlier in the year (1998) that a strong La Niña was likely. In January 1998, ECMWF produced a La Niña forecast for later in the year. Scripps' hybrid coupled model suggested as early as August 1997 that there would be cool SSTs in June or July 1998 and quite cold for 1999. The different models have different consistent biases, which Graham said should be accounted for in the projections. A participant asserted that it was very important to know the biases of the various models, so that their forecasts can be judged against that bias. Another participant suggested that it was considerably harder to forecast the onset of a La Niña than the onset of an El Niño and that a model's success can be assessed in part by its ability to reliably forecast a La Niña event. Discussion in this session then centered on lead time. These are on the 3-, 6- and 9-month leads. The best projections are for the 3-month leads as the forecast's reliability deteriorates as the lead time gets longer. The wide range of users have different needs, when it comes to lead time. While some users could benefit greatly from a 3-month forecast, others need much longer warnings lead times in order to better prepare for ENSO's impacts. Obviously, the goal is to forecast reliably as far into the future as possible and as reliable as possible. This issue is one for the users of ENSO forecasts to discuss. Any improvement would be beneficial with regard to extending lead times for ENSO forecasts and for improving reliability. A question arose again about which type of model had performed best: dynamical or statistical. Barnston suggested that his assessment showed that they performed about the same, but that in the future he expected that the dynamical models would outperform the statistical models. While the ECMWF makes forecasts based on the output of a coupled model, other models based their projections on SSTs in a set of forecast runs (i.e., an ensemble of forecasts). The dynamical models "know nothing about history," whereas the statistical models involve pattern matching. In response to an earlier comment by one participant, another participant challenged the comment that bigger models will produce improved forecasts, reminding participants that there was an ongoing debate in the forecast community about this point of view. Key Points:
The Identification of Differences in Forecasting El Niño and La Niña Stephen Zebiak (Director of Modeling, International Research Institute/Lamont-Doherty Earth Observatory) opened this discussion session by stating that, from the perspective of simplified dynamical and statistical models, the procedures for capturing the physics of both extreme cold and warm events are the same. Thus, after 15 or so years of research, we have come to think of ENSO warm and cold events as a process, an oscillation, and not as a sequencing of random events. In the context of an oscillation, the same physics describes the whole state, including both extremes. We now see that the succession of warm and cold states have systematic differences. For example, it is believed that the cold states do not depart from average to the same extent that warm states do. Also, some decades such as the 1930s were relatively inactive periods with regard to extreme ENSO events while others have been dominated either by cold events (earlier in the century) or by warm events (in recent decades). Multivariate ENSO Index (Figure 11) Zebiak noted that there are physical differences between relatively warm and cold events relating to thermocline depth changes, convection expansion, and movement of the zone of convection. In a cold state (not necessarily a La Niña), the thermocline in the eastern equatorial Pacific is shallow to start with. It does not take much to bring it relatively closer to the surface. He also noted that for the models, the "balance of terms" of the mixed layer will be somewhat different. This taxes the ocean models in different ways. During strong cold events, the parameterizations are taxed. These are taxed somewhat differently in the atmosphere and ocean and in both extremes of ENSO. The "balance of terms" in the extratropics may vary according to longitude. He noted that there were processes in El Niño and La Niña conditions that were nonlinear in different ways. Zebiak stated that, although some of the physics at work in the two extremes are different, the predictability of El Niño and La Niña is not much different. There is a perception that forecasting La Niña has been more troublesome than forecasting El Niño. Questions were raised about the Lamont-Doherty Earth Observatory (LDEO) model which missed the forecast of the 1997-98 El Niño. Zebiak acknowledged that the newer model (LDEO2) performed much better with their use of real ocean temperatures. He noted that this model change underscored the value and importance of ocean observation data to ENSO modeling activities. He also noted that the ECMWF model projections were encouraging, but cautioned that its performance had not yet been validated over time. Zebiak identified four factors that tend to limit forecast skill: model flaws, flaws in the way that data are used (e.g., assimilation and initialization), gaps in the observing system, and the inherent limits of predictability. A participant suggested that a publicly "busted" (incorrect) forecast can help to stimulate support for research, and that the El Niño puzzle has not yet been solved (i.e., understanding El Niño is not yet a "done deal"). It was also noted that the ENSO forecast community should not shy away from making public their projections and predictions, as there will be forecast successes and misses. A question was raised about the terminology used to describe El Niño events -- weak, moderate, strong, very strong, extraordinary -- and whether those terms could be applied to La Niña events as well. It was suggested that the range between minimum and maximum temperature anomalies for cold events is narrower than for warm events, and that it would be possible but more difficult to apply these terms to a cold event. Some participants voiced their opposition to defining La Niña as an exaggerated condition of normal, while recognizing that perhaps there is a need to reconsider how these warm and cold states in the tropical Pacific are defined. A participant suggested that SSTs may not the best way to characterize changes in the tropical Pacific, noting that other indicators such as sea level pressure or easterly wind anomalies might provide a better way to characterize the whole system in the tropical Pacific. Another participant revisited the definition of the La Niña issue, suggesting that there is no "normal" state as such but that there are only two states: cold and warm. What we now consider normal may in fact be only a weak state of La Niña. Key Points:
Monitoring La Niña Historically, El Niño researchers have used regions in the tropical Pacific referred to as Niño3 and Niño3.4 (Figure 12) to generate indices of SSTs that are convenient for describing the evolution of an El Niño event. The session focused on whether these historical indices were appropriate for describing and understanding the character of La Niña as well. Antonio Busalacchi (NASA's Goddard Space Flight Center [GSFC]) opened the session on monitoring by raising the question, "Are there gaps in the monitoring system when viewed from a La Niña perspective?" Stated in another w |
