Meteorologist, oceanographer, Python solvem probler, experimenter, Raspberry Pi tinkerer, geek dad, manager, space enthusiast, ham radio lover, science columnist, ADHDer, synthetic intellect curious, linux admin cosplayer.

The 1997-98 El Niño: Validation of Operational Seasonal Forecasting

Research notes for the next NWP-history blog post on the TOGA-TAO buoy array. Focus: the 1997-98 El Niño as the validation moment for the entire TOGA + Cane-Zebiak + NCEP CFS pipeline. Companion to Post 27 The Forecast on a VAX and to research/computers/ENSO_discovery_history.md.

Tagline

The 1986 Cane-Zebiak forecast called the spirits from the vasty deep. The 1997-98 event was when the spirits came in front of a watching world – and TOGA-TAO buoys, NCEP coupled models, and the IRI were already in place to record it, predict it, and explain it.

The 1997-98 El Niño was, by some measures, the strongest on record. It was also the first event of its scale in the satellite-and-buoy era, the first to be observed continuously from the western Pacific warm pool to the South American coast, the first to be predicted operationally by multiple coupled-model systems running at multiple national weather centres, and the first whose impacts were communicated to the global public months in advance with sufficient confidence that governments and individuals could prepare. By some metrics, the early-onset and rapid-growth phases were stronger than the legendary 1982-83 event. By the end of December 1997 the equatorial cold tongue had warmed to something approaching the temperature of a warm bathwater everywhere across the eastern Pacific. By the end of June 1998 it had collapsed. The cumulative damages have been estimated, with very wide error bars, between US$32 billion and US$96 billion in 1998 dollars; the confirmed death toll was at least 22,000, with the highest individual-country contributions from Indonesia, Peru, Kenya, and southern China.1

This document covers the science of the event, the operational forecast, the cultural moment, and the institutional consequences.

1. Atmospheric and oceanic anomalies

1.1 First detection: late 1996 / January 1997

The 1997-98 El Niño was preconditioned by an accumulation of warm water in the western equatorial Pacific in 1996 – a deeper-than-normal thermocline, greater upper layer thickness (ULT), and SSTs about 0.5 °C above normal in the western basin. This is the standard precursor of major El Niño events: the basin must be charged with warm water in the west before the discharge phase that becomes El Niño.2

The first specific observational signal that would become the 1997-98 event was a series of westerly wind bursts (WWBs) over the western Pacific warm pool in late 1996 and early 1997. These were associated with the Madden-Julian Oscillation (MJO). On the TAO/TRITON buoy array, the WWBs registered as:

  1. Eastward displacement of the convective rising branch of the Walker circulation from the Indonesian region toward the date line.
  2. Westerly anomalies on the equator in the western Pacific lasting several weeks at a time.
  3. Subsurface warm anomalies propagating eastward at Kelvin-wave speed (~2.7 m/s), reaching the eastern Pacific in approximately two months.

McPhaden’s Science paper (1999) identifies the genesis of the event with the second WWB, in March-April 1997, although the precursor warming was visible in TAO subsurface data months earlier:3

“The 1997-98 El Niño-Southern Oscillation was, by some measures, the strongest on record, with major climatic impacts felt around the world. A newly completed tropical Pacific atmosphere-ocean observing system documented this El Niño from its rapid onset to its sudden demise in greater detail than was ever before possible.”3

By January 1997, deep-water temperature probes in the central Pacific detected an area of unusually warm subsurface water; by April, surface anomalies were “fully established”; by May, they exceeded 5 °C off South America; by September, surface temperatures between South America and the international dateline averaged 2-4 °C above normal across nearly a quarter of Earth’s circumference.4

1.2 Build-up: summer 1997

Through the boreal summer of 1997 the event grew with extraordinary speed. Each WWB excited a new downwelling Kelvin wave; each Kelvin wave, on reaching the eastern Pacific, pushed the thermocline deeper; the deepening thermocline reduced upwelling-driven cooling; the warming SST further weakened the trade winds; the weakened trades produced more eastward-shifted convection and more WWBs. The Bjerknes feedback was running with particular vigour.

McPhaden, summarising the development period: “The El Niño developed so rapidly that each month from June to December 1997 a new monthly record high was set for SST.”3

By August 1997 the equatorial Niño 3.4 SST anomaly (the standard index, defined over the region 5°N-5°S, 170°W-120°W) exceeded +2 °C. The thermocline depth in the eastern Pacific had been depressed by more than 90 metres relative to normal. The 28 °C isotherm that normally marks the western Pacific warm pool now extended across the entire equatorial basin – “28° to 29°C water filled the equatorial basin,” in McPhaden’s phrase.3

1.3 Peak: November-December 1997 / January 1998

The mature phase of the event was attained in November-December 1997.2 Niño 3.4 SST anomaly peaked at approximately +2.5 °C in November 1997 (some analyses put the peak in the Oceanic Niño Index three-month average at +2.4 °C). The Niño 1+2 region (off Peru and Ecuador, 0-10°S, 90°W-80°W) was even more anomalous: SST anomalies exceeding 5 °C persisted from July 1997 through May 1998, with peak coastal anomalies of 11 °C above average reported off Peru in January 1998.4

Wallace, Rasmusson, Mitchell, Kousky, Sarachik, and von Storch (1998), in their landmark J. Geophys. Res. synthesis of TOGA-era observations, documented that the 1997-98 event reset essentially every metric of ENSO climatology that had been compiled during the TOGA decade.5

For the quantitative comparison with 1982-83, the standard reference is Enfield (1999) in the Bulletin of Marine Science:2

“Both the measured temperatures and the anomalies for NINO-3 and NINO-1+2 rose faster in mid-1997 than in mid-1982. Throughout the boreal summer and fall the indices were larger in 1997 than 1982. In the boreal winter-spring season the 1983 temperatures and anomalies were equal to or greater than those of 1998.”

Enfield’s summary judgement: “The 1982-83 and 1997-98 events, coming only 15 years apart, are generally comparable in magnitude and impacts, and are by far the strongest warmings observed since 1950.”2 In the eastern Pacific (NINO-1+2 region), 1997-98 was somewhat stronger and lasted longer; in the central Pacific (NINO-3, NINO-4 regions) and on most atmospheric indices, 1982-83 was at least equally strong.

Wolter and Timlin (1998), using a more objective Multivariate ENSO Index, concluded slightly in favor of 1982-83 as the stronger event overall; Davey and Anderson (1998), looking at SST, wind stress, sea level pressure, and thermocline depth indices, called the comparison “essentially a tie.”67

1.4 Walker circulation collapse

The atmospheric signature of a strong El Niño is the collapse of the Walker circulation – the east-west overturning that under normal conditions has its rising branch over Indonesia, eastward upper-tropospheric flow, sinking branch over the cold tongue off Peru, and westward surface return flow as the trade winds. During the 1997-98 event:

  1. The rising branch shifted eastward to the central Pacific (around 160°W).
  2. The trade winds collapsed across the central and eastern Pacific.
  3. Convective rainfall shifted from Indonesia to the central Pacific.
  4. The descending branch shifted westward, producing severe drought over Indonesia.

Wallace et al. (1998) characterised the atmospheric structure during the mature phase: “The paradigm of a quasiperiodic ‘ENSO cycle,’ phase locked with the annual march, does not capture the complexity of the evolution of the anomalies.”5 In other words, the 1997-98 event was atypical in some details of its evolution – the rapid onset in particular was faster than canonical ENSO theory predicted.

1.5 The sudden demise

The collapse of the event was as dramatic as the onset. Between May and June 1998, the central and eastern equatorial Pacific cooled by 5-8 °C in a matter of weeks. McPhaden reported the most extreme observation: “At one location (0°, 125°W), SST dropped 8°C in 30 days, more than 10 times the normal cooling rate.”3

The decay was driven by a combination of (a) wind-forced upwelling Kelvin waves excited by the return of easterly trade-wind anomalies in the central Pacific, and (b) reflection of upwelling Rossby waves off the western boundary of the basin. The recharge phase of the recharge oscillator (Jin 1997) was already under way in the western Pacific by late 1997, with a cool subsurface anomaly accumulating that would, by July 1998, become the strong La Niña event of 1998-2000.2

McPhaden’s summary on demise: “The demise of the El Niño was related to a complex interplay of wind-forced and western boundary-forced waves that preconditioned the ocean to sudden sea surface temperature cooling initially centered between 125°W and 170°W along the equator.”3

2. The forecast: NCEP, Cane-Zebiak, and ECMWF

2.1 Cane-Zebiak at Lamont

The Cane-Zebiak coupled model – the model whose 1986 Nature paper started the operational ENSO-forecasting era and which is the protagonist of Post 27 – was issuing experimental long-lead forecasts continuously through the 1990s under the auspices of Lamont-Doherty’s Climate Group, with Cane and Zebiak still personally overseeing operations. The model’s forecasts were published in NCEP/CPC’s Experimental Long-Lead Forecast Bulletin on a roughly quarterly basis from the mid-1990s onward.8

Performance during 1996-97 was mixed. The model’s January 1997 forecast (made on data through late 1996) called for “continuation of somewhat below normal Nino 3 conditions, with strongest negative anomalies for winter 1996-97” – in other words, weak La Niña conditions, exactly the opposite of what was about to happen.8 By March-April 1997 the model had reversed and was correctly predicting strong El Niño conditions for the second half of 1997, but only after the warming was already under way and being assimilated into the model’s ocean state. Across the suite of models surveyed by Barnston, Glantz, and He (1999) and by Landsea and Knaff (1999), Cane-Zebiak underpredicted the 1997-98 event at every long-lead horizon prior to mid-1997.910

Cane himself, in subsequent interviews, has been candid about the model’s limitations during 1997. In a 2018 National Science Review interview he framed the moment with characteristic dry retrospection: the Cane-Zebiak model “said no – and then it would happen a year later.”11

2.2 NCEP CFS

The NCEP Climate Forecast System (CFS) – NCEP’s operational coupled atmosphere-ocean model for ENSO and seasonal forecasting – was being developed under the leadership of Ants Leetmaa, Director of the Climate Prediction Center, and Ming Ji of the Environmental Modeling Center, throughout the 1990s. By 1997 the system was a forecast-quality coupled GCM running every month with initialisation from the TAO buoy array and the Reynolds optimal-interpolation SST analysis.9

The crucial NCEP forecast was issued on 26 June 1997. It was the official NOAA Climate Prediction Center “Climate Outlook” and “ENSO Diagnostic Discussion” for the upcoming North American winter. It predicted, with explicit reference to the El Niño then developing in the Pacific:9

  1. Above-normal precipitation across California, the southwestern US, and the southern Gulf states.
  2. Above-normal precipitation across the southeastern US.
  3. Below-normal precipitation across the Pacific Northwest.
  4. Above-normal temperatures across the northern half of the country.
  5. Wetter and stormier conditions across the southern third of the United States.

Of these, every single one was qualitatively correct. The Barnston, Leetmaa, Kousky, Livezey, O’Lenic, Van den Dool, Wagner, and Unger (1999) BAMS postmortem on the forecast called it: “Heavy winter precipitation across California and the southern plains-Gulf coast region was accurately forecast with at least six months of lead time, and dryness was also correctly forecast in Montana and in the southwestern Ohio Valley. The warmth across the northern half of the country was correctly forecast, but extended farther south and east than predicted.”9

This was the single most successful operational long-lead seasonal forecast in the history of the National Weather Service up to that time.

2.3 ECMWF

The European Centre for Medium-Range Weather Forecasts had been developing a coupled atmosphere-ocean seasonal forecasting system through the early 1990s under T. N. Stockdale and D. L. T. Anderson. Their model was the only system that successfully predicted, two seasons in advance, the rapid rise of SST anomalies in May 1997 – but, as Stockdale et al. (1998) noted, the prediction was made experimentally and was not made publicly available before it verified, so it did not influence the operational forecast community.12

Stockdale, Anderson, Alves, and Balmaseda (1998) in Nature: their seasonal-forecast system predicted, on data through January 1997, a +2 °C Niño 3.4 anomaly for the December-January-February 1997-98 season. Verification: +2.4 °C. The lead time was approximately 11 months.12

This was the ECMWF system’s coming-of-age moment. It would become, over the following two decades, the world’s best operational ENSO forecasting system, eventually evolving into SEAS5 (currently in operation in 2026).

2.4 The contrast with 1982-83

The contrast with the 1982-83 event is the central institutional point. In 1982-83, there was no operational ENSO forecast. There was, indeed, no operational coupled atmosphere-ocean model anywhere in the world – the Cane-Zebiak system was still being developed, ECMWF had no seasonal coupled model, NCEP had no Climate Forecast System. The 1982-83 event was, additionally, partially masked by stratospheric aerosol from the El Chichón eruption of March-April 1982, which biased early satellite SST retrievals; the actual scale of the warming was not appreciated by the operational community until the autumn of 1982, by which time the event was nearly half over. The post-mortem on 1982-83 is what drove the international community to build TOGA in the first place. (See research/computers/ENSO_discovery_history.md Section 9 for the 1982-83 detail.)

By 1997-98, fifteen years later:

  • The TAO buoy array was complete.
  • Multiple satellite SST analyses (NOAA OI SST, NCEP, ESA) cross-validated the in-situ data.
  • The Cane-Zebiak, NCEP CFS, ECMWF, COLA, and several other coupled models were running operationally.
  • The Climate Prediction Center had been formally established in 1995 as a successor to the Climate Analysis Center, with a specific operational mandate for seasonal forecasting.
  • The IRI had been founded in 1996 as a NOAA-Columbia partnership for operational seasonal climate forecasting.

The 1997-98 event was the first major test of the entire pipeline. The pipeline passed the test for short-range forecasts (1-3 seasons ahead, after the event was already developing) but failed it for the longest-range forecasts (more than three seasons ahead, before the event began). This is honest. It is also the standard pattern of operational forecasting maturation: the system worked well enough to be useful, well enough to drive the public-affairs response and the disaster-preparedness measures, but it did not deliver the perfect long-range forecast that the post-event hype sometimes claimed.

3. The TOGA-TAO contribution

3.1 The buoy array

The Tropical Atmosphere Ocean (TAO) array of moored buoys was completed in December 1994 – 70 buoys spaced roughly every 2-3° of longitude along the equator from 137°E to 95°W, plus several off-equatorial moorings. Each buoy measured (typically): SST, sea surface salinity, surface wind speed and direction, air temperature and humidity, surface shortwave radiation, and – crucially – a thermistor chain providing subsurface temperature at 11 standard depths from the surface down to 500 m. Data were transmitted in near-real-time by Service Argos satellite link to PMEL in Seattle, where they were processed and made publicly available within hours. The array was operated by NOAA PMEL under Mike McPhaden’s directorship from 1992 onward.13

The 1997-98 El Niño was the first major ENSO event in which the entire array was fully operational throughout the event’s lifecycle. Before TOGA-TAO, ENSO observations had been a patchwork of expendable bathythermograph (XBT) lines from ships of opportunity, drifting buoys, island weather stations, and satellite SST retrievals – excellent in places, sparse and lagged in others. With TAO, the equatorial Pacific had, for the first time, a real-time three-dimensional view of the upper ocean.

Enfield (1999) summarised the TOGA-TAO contribution: “The 1997-98 El Niño was the best-measured event ever.”2

3.2 What TOGA-TAO saw in 1996-97

The early subsurface warm anomaly that became the 1997-98 El Niño was first detected in TAO subsurface data in mid-1996 – a slow eastward propagation of warm anomalies from the western Pacific along the thermocline.2 Through late 1996 and early 1997, TAO recorded a series of MJO-driven westerly wind bursts in the western and central Pacific and the associated downwelling Kelvin waves they excited. By March 1997, the leading Kelvin wave had reached the eastern Pacific and the surface warming off Ecuador had begun.

Crucially, this evolution was being observed in real time. Researchers and forecasters at PMEL, NCEP, IRI, ECMWF, and dozens of other institutions had access, via the PMEL website, to the TAO data within hours of collection. The McPhaden et al. (1998) J. Geophys. Res. TOGA observing-system paper – published just months before the 1997-98 event began – had documented the array’s capabilities and the scientific community’s readiness to use them.13

The TAO data fed directly into:

  1. The NCEP coupled-model initialisation (via Behringer-Ji-Leetmaa 1998 ocean assimilation).14
  2. The Cane-Zebiak monthly forecasts (which used a wind-stress-only initialisation but cross-validated against TAO subsurface temperatures).
  3. The ECMWF seasonal forecast system (via the ocean reanalysis OPA-OASIS).
  4. Real-time monitoring at the Climate Prediction Center, which issued its weekly ENSO Diagnostic Discussions on the basis of TAO SST and subsurface temperatures.

The June 1997 NCEP forecast that drove the public-affairs response was, fundamentally, based on TAO data showing that the El Niño was already developing strongly in mid-1997.9

4. Public reception

4.1 “El Niño” as household term

The 1997-98 El Niño was the moment “El Niño” entered American popular culture as a household phrase. The term had been used in scientific literature since the late nineteenth century, in oceanographic literature since the 1920s, and in the climate-science community since Bjerknes’s 1969 synthesis. It had got into mainstream press during the 1982-83 event but mostly as a foreign-news phenomenon affecting Peru and Indonesia. By autumn 1997, it was on the front page of every major American newspaper, the lead item on weeknight network news, and the punchline of Tonight Show monologues.

Hare (1998) in Eos counted: between June 1997 and June 1998, more than 1,000 stories on El Niño appeared in The New York Times alone; the Washington Post, Los Angeles Times, and USA Today together contributed several thousand more. Television weather forecasts began to mention El Niño explicitly during their nightly broadcasts; National Public Radio ran an entire series on it; Time and Newsweek both ran cover stories.15

4.2 The Clinton administration response

President Bill Clinton’s administration treated the 1997-98 El Niño as a serious operational challenge. Vice President Al Gore had been personally interested in climate science since his Senate years and had taken the public-communications portfolio for federal climate response. Federal Emergency Management Agency Director James Lee Witt – whom Clinton had elevated to cabinet rank in 1996, the first FEMA director with that status – was the lead operational figure for the El Niño response.16

Through the autumn of 1997 and the winter of 1997-98, the Clinton administration ran what amounted to a sustained public-affairs campaign on El Niño preparedness. NOAA Administrator D. James Baker Jr. was the lead public scientific spokesman; FEMA’s Witt was the lead public operational spokesman; Vice President Gore was the lead public political spokesman. The campaign included:

  1. A series of NOAA press briefings on the developing El Niño from June 1997 onward.
  2. A Vice-Presidential event in October 1997 showcasing a NOAA study suggesting that global warming might lead to more frequent and intense El Niños.17
  3. A White House Climate Conference on October 6, 1997, attended by 100 broadcast meteorologists and the cabinet (Energy Secretary Federico Peña, Health and Human Services Secretary Donna Shalala, Labor Secretary Alexis Herman, Commerce Secretary William Daley, EPA Administrator Carol Browner, FEMA Director Witt) – the briefing covered both climate change and El Niño preparedness.17
  4. A presidential visit to California in February 1998 to declare a major disaster after El Niño-driven flooding and to thank FEMA Director Witt and Senator Barbara Boxer for preparedness leadership.18
  5. The “Project Impact” disaster-preparedness initiative launched in late 1997, in which Clinton explicitly cited El Niño as a motivating example.19

Vice President Gore’s framing in February 1998 (verified at the Project Impact press conference): “Federal officials had the foresight to plan for the worst and prepare for the terrible impact of El Niño.” Gore explicitly thanked “FEMA Director James Lee Witt and Senator Barbara Boxer for their leadership in this area.”18

4.3 Cultural moment

By October 1997 “El Niño” had become a punchline. The most famous instance is Chris Farley’s appearance on Saturday Night Live on October 25, 1997, in his last episode as host before his death two months later. Farley appeared in a skit titled “Weather Scope” as a wrestling-style character (“El Niño”) issuing pseudo-Spanish challenges to other weather phenomena; the bit ran for several minutes and is widely remembered as one of the best of his SNL career. The transcript exists at SNL Transcripts Tonight (Bryan, “97delnino”); the YouTube clip remains one of the most-viewed comedy moments of the 1997-98 season.20

The pop-culture moment was broader than a single SNL bit. By the autumn of 1997 there were T-shirts, novelty mugs, stickers, and weather-merchandise products. Surf shops in California did brisk business in El Niño-themed surfwear. The Tonight Show with Jay Leno made El Niño a recurring punchline. Late Show with David Letterman ran weather-related Top 10 lists. The late-night comedy treatment was a marker of the term’s having graduated from technical jargon to public idiom.

5. Specific impacts and forecast usefulness

5.1 California: wettest winter on record

The Climate Prediction Center’s June 1997 forecast called for above-normal precipitation across California for the upcoming winter. Verification: California recorded its wettest February on record (since 1895). The state experienced six major Pacific storms between November 1997 and February 1998, with cumulative damages estimated at over US$550 million in San Francisco Bay region alone. Wind-driven waves of 15-20 m struck the central California coast; the National Park Service closed 95% of trails at Point Reyes.21

5.2 Pacific Northwest: predicted dry; was dry

Counter-intuitively, the Pacific Northwest experiences below-normal precipitation during strong El Niño events because the storm track shifts south. The CPC forecast got this right – Washington and Oregon recorded a relatively dry winter – though the post-event 1998-99 La Niña winter delivered the largest storms.21

5.3 Southeastern US: record rainfall and Florida tornadoes

The CPC June 1997 forecast called for above-normal precipitation across the southeastern US driven by the southward-shifted storm track. Verification: the Southeast had the wettest December-February on record (Florida 19.28 inches, South Carolina 19.54 inches; for comparison the 1963-64 record was 15.14 inches and 1981-82 was 17.38). Florida experienced the worst tornado outbreak in state history during a single 24-hour period in February 1998; 42 deaths.22

5.4 Northeast: warm winter, January ice storm

The CPC forecast called for above-normal temperatures across the northern US. The Northeast set warmth records in Connecticut, Iowa, Michigan, Minnesota, New Hampshire, Wisconsin. The January 5-9, 1998 ice storm in upstate New York, northern New England, and southeastern Canada killed 56 people and caused over US$2 billion in damages in Canada and over $300 million in the US. The link between the ice storm and El Niño remains debated; the El Niño-induced Pacific storm track that produced the warm and wet pattern was a contributing factor to the moisture that fell as freezing rain on the cold air mass over Quebec and northern New England.22

5.5 Peru: catastrophic flooding

Predicted, in classical El Niño style, by every forecast system. Verification: northern Peru flooding produced over 200 deaths, US$1.4 billion in damages, displacement of hundreds of thousands of people. The anchoveta fishery off Peru collapsed; boreal-winter production of fishmeal and oil fell by more than 80% relative to the previous year. Coral bleaching at the Galápagos islands; reduced seabird populations; widespread coastal mortality.23

5.6 Indonesia: drought and forest fires

Predicted to suffer drought, in classical El Niño style. Verification: Indonesia suffered the worst drought in 50 years. Approximately 9.7 million hectares of forest burned across Sumatra and Kalimantan during 1997-98 – an area larger than Hungary or South Korea. Smoke spread over eight Southeast Asian countries and affected 75 million people. Health impacts: respiratory illness, transportation disruption, loss of agricultural production. The Indonesian forest fires of 1997-98 were also a significant carbon-cycle event, adding to atmospheric CO₂ at a level comparable to a year’s worth of fossil-fuel emissions from Western Europe.24

5.7 Australia: predicted drought, less severe than 1982-83

Predicted to suffer drought; in fact suffered drought, but not as severely as 1982-83. Enfield (1999) notes “northern Australia and the southern cone of Africa suffered far less drought than in 1982-83.” Wheat production was below average, but national grain output was less affected than the 1982 case.2

5.8 East Asia: warm winter

Predicted; verified. The Asian monsoon in summer 1997 was disrupted, but less severely than in 1982. China experienced unusually warm winter conditions across the south.2

5.9 Africa: mixed results

The CPC and IRI forecasts for sub-Saharan Africa called for drought in southern Africa, flooding in eastern Africa, and warm conditions in northern Africa. Verification: Kenya and Somalia had widespread flooding in 1997 and a Rift Valley fever outbreak that killed 200-250 people and produced approximately 89,000 human cases of the disease. Mozambique had widespread flooding. Ethiopia and Djibouti had drought. Southern Africa drought was milder than predicted – a significant enough miss that Cane himself, in a SciDev.Net interview, has cited Africa as the case where over-confident forecasts produced unintended economic harm: banks reportedly withdrew credit from farmers anticipating reduced incomes, and the actual drought was less severe than the forecasts had implied.25

5.10 Total economic damage

The single most-cited damage figure is approximately US$36 billion globally, from Sponberg (1999) – the most-cited published estimate. The next-most-cited is approximately US$33 billion (Kerr 1999, Science). The lowest credible estimate is US$32 billion (Sponberg, low-end estimate in NOAA reports). The highest is US$96 billion (Swiss Re 1999). The widest-cited “global economic loss within five years” figure is US$5.7 trillion, but this includes secondary impacts and is not directly comparable to the primary direct-loss estimates.26

Flag for the blog: the damage estimates vary by an enormous range (low-end $32B vs high-end $96B is a 3x range). The most defensible single figure for the post is “between $32 billion and $96 billion in 1998 dollars, with a central estimate around $36 billion.”

5.11 Death toll

The most-cited figure is approximately 23,000 deaths globally, from Sponberg (1999) and Kerr (1999) in Science (specifically, Kerr quotes 23,000 deaths and US$33 billion damages).27 The UNEP report (Glantz et al. 2000) puts the figure at “more than 22,000 lives lost.”16

This figure is approximate. Major contributors to the death toll:

  • Indonesia forest fires and respiratory disease: several thousand attributable deaths.
  • Peru flooding: ~200-300 deaths confirmed; much higher figure likely if cholera and post-disaster mortality counted.
  • East Africa flooding and Rift Valley fever: ~200-250 RVF deaths plus much larger flooding death toll.
  • China flooding: hundreds.
  • US severe weather (tornadoes, ice storm, flooding): ~100-150 directly attributable.

Flag for the blog: the 23,000 figure is the standard, but it has very wide uncertainty bounds (probably ±50%). The Kerr 1999 Science news article is the canonical citation; secondary citations widely repeat the figure without verifying it.

6. Institutional and scientific consequences

6.1 The Climate Prediction Center

The Climate Prediction Center (CPC) had been formally established in 1995 as one of the seven service centers of the newly-named National Centers for Environmental Prediction (NCEP), succeeding the older Climate Analysis Center (CAC) which had operated under the National Meteorological Center.28 CPC’s founding mandate explicitly included operational seasonal forecasting – the kind of forecast that the 1997-98 event would test. Ants Leetmaa was the founding director (a position he held until 1998). The 1997-98 El Niño was CPC’s first major operational test, less than two years after its formal establishment. The Barnston et al. (1999) postmortem in BAMS is, by the dry standards of operational meteorology, almost a celebration. Vernon Kousky, a research meteorologist at CPC, was the public face of the CPC ENSO operations through 1997-98 and is the figure most-cited in contemporary news coverage.9

6.2 The IRI

The International Research Institute for Climate and Society (IRI), at Lamont-Doherty Earth Observatory, was formally established in 1996 – one year before the 1997-98 event began. The institutional initiative went back to a 1992 NOAA-organised Climate Forecasts Forum at which a coalition of climate scientists (Mark Cane, Stephen Zebiak, J. Shukla, V. Ramaswamy, others) and NOAA leadership (D. James Baker, Edward Frieman) had argued that the operational seasonal forecasting capability emerging from TOGA needed an institutional home outside the national weather services. The IRI’s formal founding in 1996 made it that home. Walter Munk of Scripps was a senior advisor in the founding period, though his role was institutional rather than scientific. Stephen Zebiak became IRI Director-General in 2003.29

The 1997-98 El Niño was the first major ENSO event of the IRI’s existence, and also the first major operational seasonal-forecast test for the institution. The IRI’s monthly ENSO forecast plume (which continues, in 2026, as the canonical operational multi-model ENSO forecast) was issued for the first time during the 1997-98 lifecycle.

6.3 WMO formal adoption of ENSO forecasting

Following the 1997-98 event, the World Meteorological Organization (WMO) formally adopted ENSO forecasting as a routine component of its operational mandate. The WMO 1999 Retrospective on the 1997-98 El Niño (cited as WMO 1999 in the UNEP “Once Burned” report) and the subsequent WMO Climate Information and Prediction Services (CLIPS) programme institutionalised ENSO forecasting as a service that national meteorological services around the world were expected to provide.16

The WMO’s adoption was, in turn, the trigger for the formal incorporation of ENSO and seasonal forecasting into the WMO Global Producing Centres (GPC) framework that operates today, with NCEP CFS, ECMWF SEAS5, UKMO GloSea, Météo-France, Environment Canada, and others as designated WMO Global Producing Centres.

6.4 The TOGA-TAO array becomes operational

The TOGA observational programme had formally ended on 31 December 1994. Its successor in the Pacific was the operational maintenance of the TAO array under NOAA PMEL, with multinational participation. The 1997-98 El Niño was the proof-of-value moment that secured long-term US Congressional funding for the array. McPhaden’s PMEL group received significant institutional recognition; in 1997 the US Congress authorised long-term sustained support of the TAO array as part of an operational El Niño/Southern Oscillation observing system. The array remains in place in 2026, with TRITON (Japanese) buoys complementing the TAO western Pacific section. After 2000 the joint system was renamed TAO/TRITON.13

6.5 Climate Forecast System (CFS)

The NCEP coupled model that performed in 1997-98 was the precursor system to the formal NCEP Climate Forecast System (CFS), which became fully operational in August 2004. The 1997-98 performance was the institutional case for elevating the experimental coupled system to operational status. CFSv2, the current operational version, has been in service since 2011 and contributes to the IRI ENSO forecast plume monthly.30

7. Specific named scientists and their roles

The 1997-98 El Niño was the largest single climate-science institutional event of the late 1990s. The named scientists whose roles defined the moment:

Mark A. Cane (Lamont-Doherty Earth Observatory, Columbia University). G. Unger Vetlesen Professor of Earth and Climate Sciences. Co-founder, with Stephen Zebiak, of the Cane-Zebiak coupled model. Co-founder of the IRI in 1996. Vetlesen Prize 2017. The intellectual figure most associated with the operational ENSO-forecasting era. Born 1944 in Brooklyn, NY. (Detailed biography in Post 27.)

Stephen E. Zebiak (Lamont-Doherty / IRI). Cane’s first MIT PhD student (1985). Lead developer of the ZC87 model that became the operational Cane-Zebiak system. Director-General of the IRI 2003-2012. Currently Special Research Scientist at Lamont and CCAFS Climate Services Flagship Lead.

Michael J. McPhaden (NOAA Pacific Marine Environmental Laboratory, Seattle). Director of the TAO Array Project Office 1992-2007. Author of the canonical “Genesis and evolution of the 1997-98 El Niño” paper (Science, 1999) and the TOGA observing-system synthesis (J. Geophys. Res., 1998). The figure most associated with the observational backbone of the 1997-98 documentation. Currently a NOAA Senior Scientist at PMEL.

Vernon E. Kousky (NOAA NCEP Climate Prediction Center). Research meteorologist at CPC; the public face of CPC’s ENSO operations in 1997-98. Quoted in dozens of contemporaneous news articles as the senior NOAA spokesman on the developing El Niño. Co-author of the Wallace-Rasmusson-Mitchell-Kousky-Sarachik-von Storch (1998) TOGA synthesis.

Ants Leetmaa (NOAA NCEP CPC). Founding Director of the Climate Prediction Center 1995-1998. Lead scientist on the NCEP coupled-model development. After 1998, Leetmaa became Director of NOAA Geophysical Fluid Dynamics Laboratory (GFDL) at Princeton. He is the single figure most responsible for the operational success of the NCEP CFS during 1997-98.

Eugene M. Rasmusson (University of Maryland; previously NOAA). Born 1929 in Kansas, MIT PhD 1965 under Victor Starr. Chief of the Diagnostic Branch of NOAA’s Climate Analysis Center until 1986. Built the operational ENSO monitoring and diagnostic system that, in 1982-83, became the prototype for the modern CPC Diagnostic Discussion. Joined University of Maryland in 1986. Rasmusson is the single most important figure in the institutional transition of ENSO research from a curiosity-driven academic field into an operational service activity. Co-author of Wallace et al. (1998). Died 22 March 2015 at age 86.31

Tim P. Barnett (Scripps Institution of Oceanography, UCSD). Statistical climate dynamicist who, with the COLA group at Scripps, had been developing statistical and dynamical seasonal-forecast systems through the 1990s. Successfully predicted the 1997-98 El Niño and was widely-cited in contemporary press coverage. Born 1938; died 2022 at age 83.

Anthony G. Barnston (NOAA NCEP CPC, then IRI). Editor of the Experimental Long-Lead Forecast Bulletin. Lead author of the 1999 Barnston et al. BAMS postmortem on the 1997-98 forecast. After leaving CPC, Barnston joined the IRI and became one of the leads of the IRI’s monthly ENSO forecast plume.

T. N. Stockdale, D. L. T. Anderson, J. O. S. Alves, M. A. Balmaseda (ECMWF, Reading). The team responsible for the ECMWF coupled seasonal forecast system that successfully predicted the 1997-98 event two seasons in advance. The 1998 Nature paper (Stockdale et al. 1998) is the canonical reference for the ECMWF coming-of-age moment.12

John M. Wallace (University of Washington, Seattle). Joint Institute for the Study of the Atmosphere and Ocean. Lead author of the Wallace, Rasmusson, Mitchell, Kousky, Sarachik, and von Storch (1998) TOGA synthesis paper.5

Edward S. Sarachik (University of Washington, Seattle). Co-author of Wallace et al. 1998. Cane’s Lamont colleague and thesis advisor’s collaborator (the Sarachik with whom Cane wrote the early “Forced baroclinic ocean motions” papers).

Hans von Storch (Institute of Hydrophysics, GKSS Geesthacht, Germany). Co-author of Wallace et al. 1998. The European institutional voice on TOGA.

D. James Baker Jr. NOAA Administrator 1993-2001. The senior public scientific spokesman for the federal response to the 1997-98 event. Member of the National Academy of Engineering.

James Lee Witt FEMA Administrator 1993-2001. The senior public operational spokesman for the federal response. Cabinet-rank from 1996. Architect of the “Project Impact” disaster preparedness initiative.

8. Quotable contemporary press

The single most quotable contemporary press source is the McPhaden 1999 Science paper itself. Its abstract is the closest thing to a definitive technical statement of the event:3

“The 1997-98 El Niño-Southern Oscillation was, by some measures, the strongest on record, with major climatic impacts felt around the world. A newly completed tropical Pacific atmosphere-ocean observing system documented this El Niño from its rapid onset to its sudden demise in greater detail than was ever before possible. The unprecedented measurements challenge existing theories for El Niño-related climate swings and suggest why climate forecast models underpredicted the strength of the El Niño prior to its onset.”

The Wallace et al. (1998) abstract is the paired institutional statement:5

“Improved observations in the tropical Pacific during the Tropical Ocean-Global Atmosphere (TOGA) program have served to corroborate preexisting notions concerning the seasonally dependent relationships between sea surface temperature, sea level pressure, wind stress, rainfall, upper tropospheric circulation, and ocean thermal structure anomalies in the El Niño-Southern Oscillation (ENSO) phenomenon.”

Ants Leetmaa, on the patterns observed during the winter (quoted in NOAA NCEI Technical Report 98-02, The El Niño Winter of ‘97-‘98, Tom Ross et al., April 1998):22

“These are the patterns one would typically expect during a strong El Niño event. This year, the strong El Niño on top of the continuing gradual increase of temperature and precipitation set the stage for many all-time state records.”

Tom Karl, NOAA NCDC Director, in the same report:

“With the newest figures, the long-term trend of increasing temperatures and precipitation in the United States continues.”22

Mark Cane, in a 2018 interview, looking back at the operational forecasting era:11

“The model said no – and then it would happen a year later.”

Vice President Al Gore, at the February 26, 1998 Project Impact press event:18

“Federal officials had the foresight to plan for the worst and prepare for the terrible impact of El Niño.”

The Eos count of media coverage (Hare 1998):15 more than 1,000 stories on El Niño in The New York Times between June 1997 and June 1998. The William J. Broad and Andrew C. Revkin science-and-environment beats at the Times were the lead authors. (William K. Stevens, the Times science correspondent of that era, was also a major contributor.)

The Chris Farley SNL “Weather Scope” sketch transcript (October 25, 1997, SNL Transcripts Tonight):20 not directly quotable here for length and tonal reasons, but the bit ran for several minutes and is widely available on YouTube. It was Farley’s last SNL appearance; he died on December 18, 1997.

9. Comparison: 2015-16 El Niño

The 2015-16 El Niño was the next event of comparable magnitude to 1997-98. The events were similar in some respects, different in others.32

Magnitude. The 3-month average Oceanic Niño Index (ONI) peak in November 2015 - January 2016 was +2.3 °C, equal to the 1997-98 peak. The single-month Niño 3.4 SST anomaly peak in November 2015 was +3.0 °C, slightly higher than the November 1997 peak of +2.8 °C. In the eastern Pacific (Niño 1+2 region), 2015-16 was substantially weaker than 1997-98; the 1997-98 event remained the strongest event on record in the eastern Pacific.

Geographic pattern. The 2015-16 event was an “El Niño Modoki” or central Pacific event, with the SST maximum in the central Pacific rather than the eastern Pacific. The 1997-98 event was a canonical eastern-Pacific event. The atmospheric responses differed accordingly: 2015-16 produced a different teleconnection pattern over North America, with less intense California precipitation than 1997-98.

Forecast lead time. By 2015, ENSO forecasting was mature. The IRI/NMME (North American Multi-Model Ensemble) forecast plume included over 50 dynamical and statistical models from a dozen national centres. The 2015-16 event was forecast with high confidence approximately 12 months in advance – twice the lead time of the 1997-98 event. ECMWF SEAS5, the operational evolution of the Stockdale et al. (1998) system, was the world’s most skillful model and predicted the event onset, peak, and decay all within one standard deviation.32

Models. The 1997-98 event had been forecast skillfully by perhaps 5-6 dynamical and statistical models worldwide; the 2015-16 event was forecast by 50+. The IRI ENSO forecast plume in mid-2015 had over 30 contributing systems. The maturity of the operational seasonal forecasting community by 2015 was, in a real sense, the long-term consequence of the 1997-98 event having been the wake-up call.

Verification of operational skill. The 2015-16 success demonstrated that the 1997-98 success was reproducible. By 2015, the operational seasonal-forecasting capability was a fully institutional service, not the experimental capability of a small community of researchers as in 1997-98. The operational ENSO forecasting that the 2026 climate-services community takes for granted was, in 1997-98, still being established – and the 1997-98 event was the proof that establishment was warranted.

10. What was verified, and what was flagged

(a) Verified

  • The McPhaden (1999) Science paper text and abstract, verified at NOAA PMEL.
  • The Enfield (1999) Bulletin of Marine Science paper text, verified via NOAA AOML PDF.
  • The Wallace et al. (1998) J. Geophys. Res. paper text, verified via University of Washington PDF.
  • The NOAA NCDC Technical Report 98-02 (Ross, Lott, McCown, Quinn, 1998) on the El Niño Winter of ‘97-‘98, verified.
  • The NCEP June 26, 1997 forecast date, verified via Barnston et al. (1999) BAMS.
  • The Stockdale et al. (1998) Nature paper on the ECMWF lead-time-11-months success, verified by reference.
  • The Cane-Zebiak experimental forecasts in the CPC March 1996 Long-Lead Bulletin, verified.
  • The CPC founding 1995, the IRI founding 1996, the TAO array completion 1994, the WMO post-1998 institutional adoption: all verified.
  • Chris Farley SNL October 25, 1997 “El Niño” sketch verified at SNL Transcripts Tonight.
  • The “23,000 deaths globally and US$33 billion in damages” figure from Kerr (1999) Science news article, verified at NCAR retrospective.
  • The Barnston, Glantz, He (1999) postmortem in BAMS (statistical models 1997-98), verified.
  • The ECMWF central role on long-lead forecasting in 1997-98, verified via Stockdale et al. 1998.

(b) Flagged as widely-cited-but-uncertain

  • Total damages: estimates range from US$32 billion (Sponberg 1999) through US$36 billion (Kerr 1999) to US$96 billion (Swiss Re 1999). The “US$5.7 trillion within five years” figure includes secondary impacts and is not directly comparable. The blog post should cite the range “$32-96 billion in 1998 dollars, with central estimate near $36 billion.”

  • Death toll: standard figure is “approximately 23,000 globally” (Sponberg 1999, Kerr 1999, repeated in UNEP “Once Burned” report). Wide individual-source variation; some sources say 22,400, others up to 50,000 if cholera and post-disaster mortality are counted. Recommended: cite “more than 22,000” with acknowledgment of wide uncertainty.

  • Indonesia forest-fire hectares burned: estimates range from 5.2 million hectares (lower-end peer-reviewed) to 9.7 million hectares (upper-end including all fire-affected land). Recommended: cite “approximately 5-10 million hectares of Indonesian forest, with most-cited figure of 9.7 million.”

  • The Cane interview “we cannot forecast” exact quote: not verified in primary sources; the closest verified quote is “The model said no – and then it would happen a year later” from National Science Review 2018. The dramatic “we cannot forecast” or “scoffed” quotes from contemporaneous interviews need caveats.

  • Walter Munk’s specific founding role at IRI: institutional advisor in the 1992-1995 founding period, but exact role and dates are imprecise. He was not a formal scientific co-founder.

(c) Not directly verified

  • Specific dates for Bill Clinton press events on El Niño in October 1997. The October 6, 1997 White House Climate Conference attended by 100 broadcast meteorologists is well-documented. The October 22, 1997 Clinton speech at the National Geographic Society on climate policy (pre-Kyoto) is well-documented. A specific Clinton “El Niño press conference” in October 1997, separate from these climate-policy events, has not been independently verified.

  • The exact text of the NCEP June 26, 1997 advisory. The advisory date and content are summarised by Barnston et al. (1999) but the original text is hard to find online. Recommended: paraphrase rather than quote.

11. Three richest anecdotes for the blog post

  1. The 8 °C cooling in 30 days at 0°, 125°W. McPhaden’s Science paper records this as the most extreme observation of the entire 1997-98 event. From the date of peak warmth (late 1997) to the collapse of the equatorial cold tongue back into existence (May-June 1998), the ocean at the dateline warmed and cooled at rates unprecedented in the instrumental record. This is the single observation that, more than any other, demonstrates that the Bjerknes feedback can run in reverse as quickly as it runs forward.

  2. The Chris Farley SNL October 25, 1997 sketch as the inflection point of public awareness. Farley appeared in a wrestling-style “El Niño” character bit on his last SNL hosting appearance (he would die two months later on December 18, 1997). The sketch, parodying weather reporting, made “El Niño” a punchline overnight. The SNL bit is the moment when “El Niño” graduated from technical jargon to public idiom. It is also a reminder of how strange the late 1997 cultural moment was: a technical climate-science term, less than a year after the formal IRI founding, became the most-quoted weather phrase in the English language.

  3. The contrast with 1982-83 as institutional motivation. The 1982-83 El Niño was missed by the operational community for months because (a) there was no operational ENSO forecasting system, (b) the El Chichón volcanic eruption masked early satellite signals, and (c) there was no real-time observational backbone in the equatorial Pacific. By 1997-98, fifteen years later: TOGA-TAO was complete, the Cane-Zebiak model had been operational for a decade, NCEP CFS was running, ECMWF SEAS5’s predecessor was running, the IRI was founded, the CPC was established. The June 26, 1997 NCEP forecast got essentially every major regional impact qualitatively correct, six months in advance. This contrast – 1982 missed entirely; 1997 forecast successfully – is the single most important institutional fact about the 1997-98 event. It is what made operational seasonal forecasting a routine government service.

12. Primary-source bibliography

  • McPhaden, M. J. (1999). “Genesis and evolution of the 1997-98 El Niño.” Science 283, 950-954. DOI PMEL full text PMEL abstract PubMed.

  • Wallace, J. M., Rasmusson, E. M., Mitchell, T. P., Kousky, V. E., Sarachik, E. S., and von Storch, H. (1998). “On the structure and evolution of ENSO-related climate variability in the tropical Pacific: Lessons from TOGA.” Journal of Geophysical Research 103(C7), 14,241-14,259. PDF at U. Washington.

  • McPhaden, M. J., Busalacchi, A. J., Cheney, R., et al. (1998). “The Tropical Ocean-Global Atmosphere observing system: A decade of progress.” Journal of Geophysical Research 103(C7), 14,169-14,240. PMEL summary.

  • Enfield, D. B. (1999). “Evolution and historical perspective of the 1997-1998 El Niño-Southern Oscillation event.” Bulletin of Marine Science (in press, August 1998 resubmission). PDF at NOAA AOML.

  • Barnston, A. G., Leetmaa, A., Kousky, V. E., Livezey, R. E., O’Lenic, E. A., Van den Dool, H., Wagner, A. J., and Unger, D. A. (1999). “NCEP forecasts of the El Niño of 1997-98 and its U.S. impacts.” Bulletin of the American Meteorological Society 80(9), 1829-1852. AMS Journals CPC PDF.

  • Barnston, A. G., Glantz, M. H., and He, Y. (1999). “Predictive skill of statistical and dynamical forecasts during the 1997-98 El Niño episode and the 1998 La Niña onset.” Bulletin of the American Meteorological Society 80, 217-243.

  • Landsea, C. W., and Knaff, J. A. (2000). “How much ‘skill’ was there in forecasting the very strong 1997-98 El Niño?” Bulletin of the American Meteorological Society 81, 2107-2119.

  • Stockdale, T. N., Anderson, D. L. T., Alves, J. O. S., and Balmaseda, M. A. (1998). “Global seasonal rainfall forecasts using a coupled ocean-atmosphere model.” Nature 392, 370-373. DOI.

  • Trenberth, K. E. (1998). “Development and forecasts of the 1997/98 El Niño: CLIVAR scientific issues.” CLIVAR Exchanges 3, 4-14.

  • Davey, M. K., and Anderson, D. L. T. (1998). “A comparison of the 1997/98 El Niño with other such events.” Weather 53, 295-302.

  • Wolter, K., and Timlin, M. S. (1998). “Measuring the strength of ENSO events: How does 1997/98 rank?” Weather 53, 315-324.

  • Wang, C., and Weisberg, R. H. (2000). “The 1997-98 El Niño evolution relative to previous El Niño events.” Journal of Climate 13(2), 488-501. AMS Journals.

  • Behringer, D. W., Ji, M., and Leetmaa, A. (1998). “An improved coupled model for ENSO prediction and implications for ocean initialization. Part I: The ocean data assimilation system.” Monthly Weather Review 126, 1013-1021.

  • NCDC Technical Report 98-02 (1998). Ross, T., Lott, N., McCown, S., and Quinn, D. The El Niño Winter of ‘97-‘98. NOAA National Climatic Data Center, Asheville, NC, April 1998. PDF.

  • WMO (1999). The 1997-98 El Niño Event: A Scientific and Technical Retrospective. World Meteorological Organization, Geneva.

  • Glantz, M. H. (Principal Investigator) (2000). Lessons Learned from the 1997-98 El Niño: Once Burned, Twice Shy? UNEP/NCAR/UNU/WMO/ISDR Assessment, October 2000. PDF UN Press Release.

  • Sponberg, K. (1999). “Compendium of climatological impacts.” University Corporation for Atmospheric Research / NOAA Office of Global Programs.

  • Kerr, R. A. (1999). “Big El Niños ride the back of slower climate change.” Science 283, 1108-1109. (The key contemporary news article in Science.)

  • Hare, S. R. (1998). “Recent El Niño brought downpour of media coverage.” Eos 79(40), 481.

  • Harrison, D. E., and Larkin, N. K. (1998). “Seasonal U.S. temperature and precipitation anomalies associated with El Niño: Historical results and comparison with 1997-98.” Geophysical Research Letters 25, 3959-3962.

  • Oberhuber, J. M., Roeckner, E., Christoph, M., Esch, M., and Latif, M. (1998). “Predicting the ‘97 El Niño event with a global climate model.” Geophysical Research Letters 25, 2273-2276.

  • NOAA Climate Prediction Center (1997). Various ENSO Diagnostic Discussion bulletins, June-December 1997. Archive.

  • Wikipedia (2026). “1997-98 El Niño event.” link. (Useful as secondary reference; primary citations should be to McPhaden, Wallace, Enfield, Barnston, etc.)

  • Saha, S., Moorthi, S., Pan, H.-L., et al. (2006). “The NCEP Climate Forecast System.” Journal of Climate 19(15), 3483-3517. AMS Journals. (For the operational evolution of the NCEP coupled-model lineage from 1997-98 onward.)

  • Saha, S., Moorthi, S., Wu, X., et al. (2014). “The NCEP Climate Forecast System Version 2.” Journal of Climate 27(6), 2185-2208. (CFSv2, the current operational descendant.)

  • L’Heureux, M. L., Takahashi, K., Watkins, A. B., et al. (2017). “Observing and predicting the 2015/16 El Niño.” Bulletin of the American Meteorological Society 98(7), 1363-1382. AMS Journals.

Footnotes

Cross-references in the NWP-history series

  • The Forecast on a VAX — Cane and Zebiak’s 1986 Nature paper, the VAX 11/780, the founding of operational ENSO forecasting. The 1997-98 event is the validation moment for the entire pipeline that started on that VAX.
  • He Made Walker Right — Bjerknes 1969 and the feedback mechanism that the 1997-98 event ran with such vigour.
  • The Butterfly That Broke the Forecast — Lorenz 1963 and the predictability limit that ENSO does not respect.
  • The next post in the series is on TOGA-TAO as observational backbone; this research file is the 1997-98 event content for that post.

Document compiled May 2026. Sources verified to the extent possible from open primary sources; key uncertainty flags noted in Section 10. Recommended citation pattern: McPhaden 1999, Wallace et al. 1998, Enfield 1999, Barnston et al. 1999, Glantz 2000 are the canonical primary sources; secondary sources should be verified against these.

  1. Damage estimates from Sponberg (1999) at the low end (US$32 billion), Kerr (1999, Science) at the central estimate (US$33-36 billion), and Swiss Re (1999) at the high end (US$96 billion). The “US$5.7 trillion” five-year secondary-impact figure cited in the Wikipedia article is from Hsiang et al. (2023, Science), and is not directly comparable to the contemporaneous direct-loss estimates. Death toll of 22,000-23,000 from Sponberg (1999), Kerr (1999), repeated in UNEP/NCAR Glantz et al. (2000) “Once Burned” report. ↩

  2. Enfield, D. B. (1999). “Evolution and historical perspective of the 1997-1998 El Niño-Southern Oscillation event.” Bulletin of Marine Science (in press, August 1998 resubmission). Verified text at NOAA AOML PDF. ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9

  3. McPhaden, M. J. (1999). “Genesis and evolution of the 1997-98 El Niño.” Science 283, 950-954. Direct quotes verified at PMEL full text and PMEL abstract. ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7

  4. SST anomaly chronology synthesised from McPhaden (1999) and Wikipedia “1997-98 El Niño event.” The “11 °C above average off Peru in January 1998” figure comes from Wikipedia citing contemporary Peruvian meteorological reports; it refers to SST in the Niño 1+2 region during peak coastal warming. ↩ ↩2

  5. Wallace, J. M., Rasmusson, E. M., Mitchell, T. P., Kousky, V. E., Sarachik, E. S., and von Storch, H. (1998). “On the structure and evolution of ENSO-related climate variability in the tropical Pacific: Lessons from TOGA.” Journal of Geophysical Research 103(C7), 14,241-14,259. Verified text at University of Washington PDF. ↩ ↩2 ↩3 ↩4

  6. Wolter, K., and Timlin, M. S. (1998). “Measuring the strength of ENSO events: How does 1997/98 rank?” Weather 53, 315-324. ↩

  7. Davey, M. K., and Anderson, D. L. T. (1998). “A comparison of the 1997/98 El Niño with other such events.” Weather 53, 295-302. ↩

  8. Cane-Zebiak experimental forecast at NCEP CPC, March 1996 Experimental Long-Lead Forecast Bulletin, link. Performance of the Cane-Zebiak model during 1996-97 documented in Barnston, Glantz, and He (1999) and Landsea and Knaff (2000) BAMS. ↩ ↩2

  9. Barnston, A. G., Leetmaa, A., Kousky, V. E., Livezey, R. E., O’Lenic, E. A., Van den Dool, H., Wagner, A. J., and Unger, D. A. (1999). “NCEP forecasts of the El Niño of 1997-98 and its U.S. impacts.” Bulletin of the American Meteorological Society 80(9), 1829-1852. PDF. The June 26, 1997 advisory date is verified in this paper. ↩ ↩2 ↩3 ↩4 ↩5 ↩6

  10. Landsea, C. W., and Knaff, J. A. (2000). “How much ‘skill’ was there in forecasting the very strong 1997-98 El Niño?” Bulletin of the American Meteorological Society 81, 2107-2119. ↩

  11. Cane interview in National Science Review (2018), link. ↩ ↩2

  12. Stockdale, T. N., Anderson, D. L. T., Alves, J. O. S., and Balmaseda, M. A. (1998). “Global seasonal rainfall forecasts using a coupled ocean-atmosphere model.” Nature 392, 370-373. ↩ ↩2 ↩3

  13. McPhaden, M. J., Busalacchi, A. J., Cheney, R., et al. (1998). “The Tropical Ocean-Global Atmosphere observing system: A decade of progress.” Journal of Geophysical Research 103(C7), 14,169-14,240. The TAO array completion December 1994 from PMEL chronology, link. McPhaden’s PMEL TAO Project Office directorship 1992-2007 from PMEL bio. ↩ ↩2 ↩3

  14. Behringer, D. W., Ji, M., and Leetmaa, A. (1998). “An improved coupled model for ENSO prediction and implications for ocean initialization.” Monthly Weather Review 126, 1013-1021. ↩

  15. Hare, S. R. (1998). “Recent El Niño brought downpour of media coverage.” Eos 79(40), 481. ↩ ↩2

  16. Glantz, M. H. (Principal Investigator) (2000). Lessons Learned from the 1997-98 El Niño: Once Burned, Twice Shy? UNEP/NCAR/UNU/WMO/ISDR. Verified at PDF. ↩ ↩2 ↩3

  17. Clinton White House Climate Change Conference, October 6, 1997, Clinton White House archive. The cabinet attendees were Energy Secretary Federico Peña, Health and Human Services Secretary Donna Shalala, Labor Secretary Alexis Herman, Commerce Secretary William Daley, EPA Administrator Carol Browner, FEMA Director James Lee Witt, and others. ↩ ↩2

  18. Vice President Gore at the Project Impact press event, February 26, 1998, Clinton White House archive. ↩ ↩2 ↩3

  19. Project Impact disaster preparedness initiative. James Lee Witt biographical material from Wikipedia. ↩

  20. Chris Farley SNL “Weather Scope” sketch, October 25, 1997. Transcript at SNL Transcripts Tonight. Episode info at SNLArchive. ↩ ↩2

  21. California impacts and Pacific Northwest impacts from NOAA NCEI Technical Report 98-02 (1998), and NPS retrospective at NPS. ↩ ↩2

  22. NOAA NCEI Technical Report 98-02. Ross, T., Lott, N., McCown, S., Quinn, D. The El Niño Winter of ‘97-‘98. April 1998. PDF. This is the canonical contemporary documentation of US winter impacts. ↩ ↩2 ↩3 ↩4

  23. Peru flooding figures from Enfield (1999) and Weather Underground 1998 reports; the “US$1.4 billion in damages” figure is contemporary and is from the Peru-specific damage assessment cited in Wikipedia. ↩

  24. Indonesian forest fire figures: 9,755,000 hectares (1997) from various sources cited at Wikipedia “1997 Indonesian forest fires”; 5.2 million hectares from peer-reviewed estimates at lower end; “smoke spread over eight countries and 75 million people” from Asian Development Bank 1999 report. Carbon-cycle implications from Page et al. (2002) Nature. ↩

  25. African impacts mixed-results discussion from Cane interview in SciDev.Net. ↩

  26. Damage estimates: low-end Sponberg (1999); central Kerr (1999, Science) “$33 billion”; high-end Swiss Re (1999) US$96 billion; cited in Glantz (2000) “Once Burned” UNEP report; range of $32 to $96 billion from same source. ↩

  27. Death toll 23,000 globally from Kerr (1999) Science news article; UNEP report of “more than 22,000.” ↩

  28. Climate Prediction Center founded 1995 from Wikipedia “Climate Prediction Center”. Predecessor was the Climate Analysis Center (CAC). The 1995 reorganisation of NMC into NCEP made CPC one of seven service centers. ↩

  29. IRI founding 1996 from iri.columbia.edu and Lamont LDEO. The 1992 Climate Forecasts Forum is referenced in the IRI 25th anniversary materials. Walter Munk’s role described as senior advisor at Scripps. ↩

  30. NCEP Climate Forecast System (CFS) operational from August 2004; Saha et al. (2006). Predecessor coupled system in operation during 1997-98 was a development version. CFSv2 from 2011, Saha et al. (2014). ↩

  31. Eugene Rasmusson biographical material from National Academies of Engineering memorial and University of Maryland obituary, link. ↩

  32. 2015-16 El Niño comparison from L’Heureux et al. (2017) BAMS and the ECMWF Newsletter 151 (2016/2017), link. The IRI ENSO forecast plume in 2015 had over 30 contributing dynamical and statistical models. ↩ ↩2