el niño: catastrophe or opportunity - irigoddard/papers/goddard_dilley_2005.pdf · the reported...

15
El Niño: Catastrophe or Opportunity LISA GODDARD AND MAXX DILLEY International Research Institute for Climate Prediction, The Earth Institute of Columbia University, Palisades, New York (Manuscript received 2 December 2003, in final form 3 June 2004) ABSTRACT El Niño–Southern Oscillation (ENSO) phenomenon, a periodic warming of sea surface temperatures in the eastern and central equatorial Pacific, generates a significant proportion of short-term climate variations globally, second only to the seasonal cycle. Global economic losses of tens of billions of dollars are attributed to extremes of ENSO (i.e., El Niño and La Niña), suggesting that these events disproportionately trigger socioeconomic disasters on the global scale. Since global El Niño/La Niña–associated climate im- pacts were first documented in the 1980s, the prevailing assumption has been that more severe and wide- spread climate anomalies, and, therefore, greater climate-related socioeconomic losses, should be expected during ENSO extremes. Contrary to expectations, climate anomalies associated with such losses are not greater overall during ENSO extremes than during neutral periods. However, during El Niño and La Niña events climate forecasts are shown to be more accurate. Stronger ENSO events lead to greater predictability of the climate and, potentially, the socioeconomic outcomes. Thus, the prudent use of climate forecasts could mitigate adverse impacts and lead instead to increased beneficial impacts, which could transform years of ENSO extremes into the least costly to life and property. 1. Introduction The reported “cost” of El Niño events contributes greatly to misconceptions about the global climate ef- fects and socioeconomic impacts of El Niño and La Niña. The estimated costs of the two largest El Niño events of the twentieth century were 8–18 billion U.S. dollars (USD) for the 1982/83 event (UCAR 1994; Sponberg 1999), and approximately 35–45 billion USD for the 1997/98 event (Sponberg 1999). But those fig- ures neither refer to increases in economic losses, nor to losses from climate variability explicitly attributable to El Niño. Rather, they represent a gross estimate of all hydrometeorological impacts worldwide in those years. How these losses compare with those during ENSO- neutral periods has not been established. Furthermore, during El Niño events, El Niño is implicitly assumed be associated with all climate-related losses. Natural disasters are loss events with socioeconomic as well as natural causes. Hazard events, including those related to climate, such as drought, floods, strong winds, and temperature extremes, constitute one group of causal factors. A second group of causal factors is the vulnerability characteristics of the exposed elements (i.e., people, infrastructure, and economic activities) that make them susceptible to damage from the occur- rence of a specific type of hazard event. Impacts of climate anomalies are attenuated by the vulnerabilities of the exposed elements. Like climate, socioeconomic vulnerability varies over both space and time. This study seeks to establish the degree to which ENSO events and disasters are related on two levels. The first level is to look for the signal of ENSO ex- tremes in terms of disasters, because this often fuels the global anxiety surrounding ENSO events. In this line of inquiry, hazard exposure and vulnerability are con- founding factors. Thus, the second level is to examine the effect of ENSO extremes on climate anomalies di- rectly. Using disaster data and climate data from both ob- servations and predictions, three questions are consid- ered: “To what extent do climate-related disasters in- crease during years of ENSO extremes?”; “Do climate anomalies become more severe or widespread during ENSO extremes?”; and “What is our ability to predict climate variations during ENSO extremes relative to other times?”. This latter question addresses the poten- tial to reduce negative climate-related impacts by re- ducing vulnerability to climatic hazards. 2. Data and methods a. Disaster data Currently, the only publicly available global disaster event database is the Emergency Disasters Data Base Corresponding author address: Dr. Lisa Goddard, International Research Institute for Climate Prediction, The Earth Institute of Columbia University, 61 Route 9W, 228 Monell Bldg., Palisades, NY 10964-8000. E-mail: [email protected] 1MARCH 2005 GODDARD AND DILLEY 651 © 2005 American Meteorological Society JCLI3277

Upload: others

Post on 11-Jun-2020

4 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: El Niño: Catastrophe or Opportunity - IRIgoddard/papers/Goddard_Dilley_2005.pdf · The reported “cost” of El Niño events contributes greatly to misconceptions about the global

El Niño: Catastrophe or Opportunity

LISA GODDARD AND MAXX DILLEY

International Research Institute for Climate Prediction, The Earth Institute of Columbia University, Palisades, New York

(Manuscript received 2 December 2003, in final form 3 June 2004)

ABSTRACT

El Niño–Southern Oscillation (ENSO) phenomenon, a periodic warming of sea surface temperatures inthe eastern and central equatorial Pacific, generates a significant proportion of short-term climate variationsglobally, second only to the seasonal cycle. Global economic losses of tens of billions of dollars areattributed to extremes of ENSO (i.e., El Niño and La Niña), suggesting that these events disproportionatelytrigger socioeconomic disasters on the global scale. Since global El Niño/La Niña–associated climate im-pacts were first documented in the 1980s, the prevailing assumption has been that more severe and wide-spread climate anomalies, and, therefore, greater climate-related socioeconomic losses, should be expectedduring ENSO extremes. Contrary to expectations, climate anomalies associated with such losses are notgreater overall during ENSO extremes than during neutral periods. However, during El Niño and La Niñaevents climate forecasts are shown to be more accurate. Stronger ENSO events lead to greater predictabilityof the climate and, potentially, the socioeconomic outcomes. Thus, the prudent use of climate forecastscould mitigate adverse impacts and lead instead to increased beneficial impacts, which could transformyears of ENSO extremes into the least costly to life and property.

1. IntroductionThe reported “cost” of El Niño events contributes

greatly to misconceptions about the global climate ef-fects and socioeconomic impacts of El Niño and LaNiña. The estimated costs of the two largest El Niñoevents of the twentieth century were 8–18 billion U.S.dollars (USD) for the 1982/83 event (UCAR 1994;Sponberg 1999), and approximately 35–45 billion USDfor the 1997/98 event (Sponberg 1999). But those fig-ures neither refer to increases in economic losses, nor tolosses from climate variability explicitly attributable toEl Niño. Rather, they represent a gross estimate of allhydrometeorological impacts worldwide in those years.How these losses compare with those during ENSO-neutral periods has not been established. Furthermore,during El Niño events, El Niño is implicitly assumed beassociated with all climate-related losses.

Natural disasters are loss events with socioeconomicas well as natural causes. Hazard events, includingthose related to climate, such as drought, floods, strongwinds, and temperature extremes, constitute one groupof causal factors. A second group of causal factors is thevulnerability characteristics of the exposed elements(i.e., people, infrastructure, and economic activities)

that make them susceptible to damage from the occur-rence of a specific type of hazard event. Impacts ofclimate anomalies are attenuated by the vulnerabilitiesof the exposed elements. Like climate, socioeconomicvulnerability varies over both space and time.

This study seeks to establish the degree to whichENSO events and disasters are related on two levels.The first level is to look for the signal of ENSO ex-tremes in terms of disasters, because this often fuels theglobal anxiety surrounding ENSO events. In this line ofinquiry, hazard exposure and vulnerability are con-founding factors. Thus, the second level is to examinethe effect of ENSO extremes on climate anomalies di-rectly.

Using disaster data and climate data from both ob-servations and predictions, three questions are consid-ered: “To what extent do climate-related disasters in-crease during years of ENSO extremes?”; “Do climateanomalies become more severe or widespread duringENSO extremes?”; and “What is our ability to predictclimate variations during ENSO extremes relative toother times?”. This latter question addresses the poten-tial to reduce negative climate-related impacts by re-ducing vulnerability to climatic hazards.

2. Data and methods

a. Disaster data

Currently, the only publicly available global disasterevent database is the Emergency Disasters Data Base

Corresponding author address: Dr. Lisa Goddard, InternationalResearch Institute for Climate Prediction, The Earth Institute ofColumbia University, 61 Route 9W, 228 Monell Bldg., Palisades,NY 10964-8000.E-mail: [email protected]

1 MARCH 2005 G O D D A R D A N D D I L L E Y 651

© 2005 American Meteorological Society

JCLI3277

Page 2: El Niño: Catastrophe or Opportunity - IRIgoddard/papers/Goddard_Dilley_2005.pdf · The reported “cost” of El Niño events contributes greatly to misconceptions about the global

(EM-DAT 2003), which is maintained by the Center forResearch on the Epidemiology of Disasters in Brussels,Belgium. EM-DAT sources include United Nations(UN) agencies, nongovernmental organizations, insur-ance companies, research institutes, and press agencies.Criteria for inclusion in EM-DAT as a “reported disas-ter” includes exhibiting one or more of the followingcharacteristics: 10 people killed, 100 people affected,1

significant disaster (e.g., “second worst”), significantdamage, and declaration of a state of emergency or/andappeal for international assistance. EM-DAT containsapproximately 9000 natural disaster entries from 1900to the present. Reporting is more consistent in recentdecades, so that only approximately 7000 entries from1975 onward are considered.

Given the availability of data in EM-DAT and therange of losses associated with climatic hazards, fourvariables are candidates to represent disasters in ananalysis of El Niño/La Niña–related climate anomaliesand disasters. These include disaster frequency, mortal-ity, the number of affected people, and economiclosses. Disaster frequency is based on a binary deter-mination that a loss event occurred that exceeded EM-DAT’s inclusion criteria. The other three variables—mortality, the number of affected people, and economiclosses—require a more precise estimation of the mag-nitude of the losses.

Disaster frequency has several advantages, as well asone disadvantage, in the context of the current analysis.The disadvantage is that disaster frequency conflatesmajor disasters that involve high human or economiclosses with more numerous, much smaller, loss events.Using the frequency of disasters in the analysis, as op-posed to the magnitude of losses, therefore, does notallow us to completely rule out that disaster losses maybe higher globally during El Niño/La Niña events thanat other times. In other words, the frequency of disas-ters could be independent of ENSO events, but theaggregate losses are higher during ENSO events, eitherdue to higher average losses per disaster or, perhaps,because of exceptionally high losses in a few major di-sasters. As will be shown later, however, the globalcharacteristics of local and regional precipitationanomalies during ENSO extremes suggest that themagnitude of hydrometeorological hazard events is notgreater during El Niño/La Niña conditions than in neu-tral conditions.

The limitation that comes with employing disasterfrequency, however, must be considered in light of themultiple disadvantages that come with the use of losslevels as the outcome variables of interest. None of theindividual loss variables—mortality, the affected popu-lation, or economic losses—adequately characterizes

losses associated with the different types of hydrome-teorological hazard events, and these variables cannotbe easily combined into a multivariate measure of di-saster losses. Problems include missing data, a lack ofstandardized definitions and assessment methods, dif-ferences among loss variables in terms of how lossesassociated with different hydrometeorological hazardsare reflected, the confounding roles of exposure andvulnerability in creating losses, and problems with re-spect to the attribution of losses to particular types ofhydrometeorological hazard events.

A full inventory of losses in any given disaster mayinclude not only mortality, but also economic lossesacross the range of social, productive, infrastructure,and environmental sectors. Loss patterns vary spatially.Disasters in poor countries tend to have higher deathtolls than in rich countries. Gross economic losses perevent are much higher in rich countries than in poorones.

Data on aggregate economic losses in EM-DAT arelimited. Only a third of the entries include estimates ofeconomic losses, and the economic loss estimates thatdo exist are generally based on ad hoc reports ratherthan systematic, multisectoral assessments of combineddirect losses to assets and indirect losses to productionacross all of the affected sectors.

For these reasons, neither mortality nor economiclosses are adequate as a sole indicator of disaster mag-nitude. Another candidate variable, “affected people,”is not defined well enough to constitute an unambigu-ous measure of losses. Therefore, insufficient informa-tion is available to adequately characterize the magni-tude of disaster losses across all climatic hazards.

Finally, and in light of the role of vulnerability todisaster causality, loss attribution can be highly prob-lematic. For example, unlike flooding, where peoplecan be killed or property can be destroyed from thedirect impact of exposure to water, the impact ofdrought on people is entirely mediated by the economy(see Sen 1981; Benson and Clay 1998). Drought-relatedmortality, when it occurs, is always a result of complexnature–society relationships in which socioeconomicconditions play a particularly important role in the out-come. This complicates the use of disaster losses as avariable in an analysis of the contributions of climaticfactors to disaster causality.

Thus, the use of disaster frequency to measure thedegree to which ENSO events affect loss patterns mini-mizes the influence of confounding factors, incompleteaccounting of losses, and difficulties with the attribu-tion of losses to particular causes. At the same time, itcapitalizes on the fact that EM-DAT applies the sameinclusion criteria every year, so that the same biases, interms of the sources that are consulted, and inclusioncriteria are standardized over the entire study period.Variations in loss levels partly depend on varying de-grees of exposure and vulnerability. Frequency is lesssensitive to variations in exposure and vulnerability

1 “Affected” people are those requiring immediate assistanceduring a period of emergency; it can also be displaced or evacu-ated people.

652 J O U R N A L O F C L I M A T E VOLUME 18

Page 3: El Niño: Catastrophe or Opportunity - IRIgoddard/papers/Goddard_Dilley_2005.pdf · The reported “cost” of El Niño events contributes greatly to misconceptions about the global

than are losses for the purposes of assessing climaticcontributions to disaster causality. Finally, frequency isa more robust metric than that of losses, considering thedifferential impacts of the different hazards being con-sidered.

b. Rainfall data

Monthly to seasonally averaged precipitation anoma-lies are used to investigate the climate side of the hy-drometeorological disasters. Clearly, droughts occur onthese climate time scales, and longer. Similarly, though,positive precipitation anomalies, which may be experi-enced as weather anomalies, that are aggregated overtime lead to the regional flooding that then registers asa national disaster. Even extreme weather events suchas hurricanes demonstrate interannual variability instrength and frequency of occurrence, contributing toseasonally averaged precipitation anomalies.

Gridded precipitation data over global land points istaken from two sources, due to uncertainties in suchanalyses. The first is the gridded analysis of the raingauge data from the Global Historical ClimatologyNetwork (GHCN), which is available at 5° horizontalresolution from the National Climate Data Center (in-formation available online at http://www.ncdc.noaa.gov/oa/climate/research/ghcn/ghcngrid.html).This dataset is based on 2064 homogeneity-adjustedprecipitation stations (from the United States, Canada,and the former Soviet Union) that were combined witha dataset containing 20 590 raw precipitation stationsthroughout the world to create the gridded fields. Ingrid boxes with no homogeneity-adjusted data, GHCNraw data were used to provide the greatest possibleglobal coverage. The second is the global land precipi-tation dataset of the Climate Research Unit (CRU) ofthe University of East Anglia (UEA; Hulme 1992,1994), which is supplied at 2.5° x 3.75° horizontal reso-lution (information online at http://www.cru.uea.ac.uk/�mikeh/datasets/global). The station dataset on whichthis gridded dataset is based is an extension of the origi-nal CRU/U.S. Department of Energy (DoE) data de-scribed in Eischeid et al. (1991). The nearly 12 000 sta-tions used were subjected to extensive quality controlbefore incorporation into the gridded analysis.

Monthly mean precipitation totals are used only forthe period of 1950–95. Rainfall anomalies for each setare calculated relative to the respective climatologycovering the entire 46-yr period. The years 1950–95constitute the analysis period also for the climate modelresults (section 2c) and the SSTs (i.e., Niño-3.4 ENSOindex, section 2e). Although data for all of these vari-ables are available for more recent years, the quality ofthe precipitation data limits the analysis to 1995. Afterthat time, the aggregated rainfall values begin to di-verge, showing unexplainable trends (in opposite direc-tions, not shown), making both datasets suspect.

c. Rainfall forecasts

Three different atmospheric general circulation mod-els (AGCMs) were forced with monthly mean observedSSTs for the period of 1950–95. Information on modelresolution and ensemble size, as well as references toinformation on model physics, is provided in Table 1.The ensemble members from a particular AGCM differfrom one another only in their initial atmospheric statesat the beginning of the runs. Observed atmospheric ini-tial conditions are not inserted at any time.

The seasonal (i.e., 3-month average) precipitationfrom the three AGCMs is transmuted into a singleprobabilistic forecast. The ensemble of integrationsfrom each AGCM is converted into three-categoryprobabilities for above-, near-, and below-normal sea-sonal precipitation, based on each model’s historicalterciles for precipitation over the 1950–95 period. TheAGCM probabilistic rainfall predictions are then com-bined using a Bayesian approach (Rajagopalan et al.2002), based on the regional and seasonal skill of eachmodel, and a climatological forecast of equal probabili-ties for each category. The International Research In-stitute for Climate Prediction (IRI) currently employsthis approach to multimodel ensembling for theirprobabilistic seasonal forecasts for temperature andprecipitation. In terms of probabilistic measures, suchas the ranked probability skill score (RPSS) and reli-ability, the IRI’s Bayesian approach outperforms theindividual models and also outperforms simpler modelcombination strategies, such as the straight pooling ofall ensemble members (Barnston et al. 2003; Robertsonet al. 2004).

These are technically multimodel AGCM simula-tions, however, they will be referred to as forecasts.Thus, the skill of these forecasts is “potential skill” be-cause it assumes that the boundary conditions, such asSSTs, which force the predictable part of seasonal cli-mate, are perfectly predictable.

d. Skill metric

The RPSS (Epstein 1969; Wilks 1995) quantifies theskill of the multimodel rainfall forecasts. The RPSS

TABLE 1. The three atmospheric GCMs used for the rainfallforecasts: the National Center for Atmospheric Research(NCAR) Community Climate Model, version 3.2 (CCM3.2; Hacket al. 1998; available online at http://www.mpimet.mpg.de/en/extra/models/echam/index.php); the ECHAM model, version 4.5(ECHAM4.5; Roeckner et al. 1996); and the National Aeronau-tics and Space Administration’s (NASA’s) Seasonal-to-Interannual Prediction Project (NSIPP; online at http://nsipp.gsfc.nasa.gov/research/atmos/atmos_descr.html) at the Goddard SpaceFlight Center.

Model CCM3.2 ECHAM4.5 NSIPP1

Ensemble size 10 24 9Horizontal resolution T42 T42 2.5° � 2.0°Vertical resolution 18 19 34

1 MARCH 2005 G O D D A R D A N D D I L L E Y 653

Page 4: El Niño: Catastrophe or Opportunity - IRIgoddard/papers/Goddard_Dilley_2005.pdf · The reported “cost” of El Niño events contributes greatly to misconceptions about the global

measures the squared error between the cumulativeforecast and observed probabilities (CPF and CPO, re-spectively), then weighs that score against a referenceforecast of climatology, which is the historically ob-served frequency of occurrence in each category,

RPSS � 100 � �1 �RPSfcst

RPSref�, �1�

where

RPS � �m�1

Ncategories

(CPFm� CPOm

�2. �2�

A score of 100% implies that the observed categoryis always forecast with 100% confidence. A score of 0implies always, or effectively, forecasting climatologicalprobablities, and a score of �0 means the forecast sys-tem performs worse than climatology. There is no ana-lytical formula or lookup table for the statistical signifi-cance levels of the RPSS. One can estimate the level ofsignificance for RPSS via a Monte Carlo resampling ofthe forecasts, but the RPSS is a stringent metric; fore-casts drawn at random relative to the observed se-quence of years typically yield a RPSS worse than thatof climatological forecasts (Mason 2004). As a result,the Monte Carlo estimates of critical RPSS values, evenat the 99% confidence level, are usually negative. Still,it is desirable to designate a level of skill deemed highand useable even if that designation is subjective. Thus,a RPSS skill threshold of 5% is chosen to represent“good skill” of the forecasts, because this value exceedsthat of climatological forecasts and is substantiallygreater than that obtainable from random forecasts. Ina discussion of the results, presented in section 5, “skill-ful forecasts” refers to those that meet or exceed thesubjective RPSS threshold of 5%.

e. Definition of El Niño/La Niña conditions

The Niño-3.4 index, which is the average SSTanomaly within the region 5°S–5°N, 170°–120°W, de-termines the categorical state of ENSO for the pur-poses of the analyses presented here. The Niño-3.4 in-dex is one of the most widely used ENSO indices(Barnston et al. 1997) because of the prevailing effect ofSST variability within this region on the global climate.The SST data are taken from version 2 of the extendedreconstructed sea surface temperature (ERSST) data-set (Smith and Reynolds 2004).

The upper and lower 25% of the distribution of themonthly mean Niño-3.4 index delimit ENSO extremes;this categorization follows from the quasi-periodic re-currence of El Niño at about 4 yr. The middle 50% ofthe distribution refers to neutral conditions. Figure 1illustrates this classification, which leads to the identi-fication of El Niño and La Niña events that are largelyconsistent with existing lists (e.g., Trenberth 1997; Ma-son and Goddard 2001).

The language regarding ENSO is specific and delib-erate in this manuscript. The term ENSO refers to theentire range of SST variability in the equatorial central/eastern Pacific Ocean, even though in a broader con-text it refers to the associated tropical Pacific atmo-spheric variability, and even to the global teleconnec-tions. The terms El Niño/La Niña conditions andENSO extremes indicate the upper and lower 25% ofthe distribution of Niño-3.4 values, which are evaluatedat the monthly to seasonal time scale. The term ENSOevent refers to the entire life span of an ENSO extreme,from its initiation, typically in the boreal spring/summer, to its demise, typically in the following spring.For the purposes of our analyses, El Niño or La Niñaconditions must exist for at least five consecutivemonths to qualify as an event.

3. Frequency of disasters

Because data within EM-DAT have been consis-tently reported over the past 25–30 yr, it can serve as abaseline against which to assess the influence of ENSOevents on certain types of socioeconomic losses. Theupward trend in disaster frequency, the most obviousfeature of the time series (Fig. 2a), reflects some com-bination of the following: the increased reporting ofdisasters, greater concentrations of people and wealthin high-risk areas, increasing vulnerability, and long-term changes in the frequency or severity of climatichazard events. This trend is removed by fitting a sec-ond-order polynomial to the yearly data on a disasterfrequency; the detrended time series is used for theanalysis of disaster frequencies relative to ENSOevents.

If one assumes for the moment that all hydrometeo-

FIG. 1. Time series of the monthly mean Niño-3.4 index (SSTanomaly averaged 5°S–5°N, 170°–120°W) (°C). El Niño and LaNiña events are highlighted with bold red and blue lines, respec-tively. El Niño and La Niña conditions of insufficient duration toqualify as events are indicated by red and blue x’s, respectively.

654 J O U R N A L O F C L I M A T E VOLUME 18

Fig 1 live 4/C

Page 5: El Niño: Catastrophe or Opportunity - IRIgoddard/papers/Goddard_Dilley_2005.pdf · The reported “cost” of El Niño events contributes greatly to misconceptions about the global

rological hazard events in years of ENSO extremesarise from El Niño or La Niña, the question then be-comes, “How is the disaster frequency in years of ElNiño/La Niña different from that in neutral years?”.One would like to examine disaster frequencies overthe precise duration of El Niño or La Niña conditions;

however, particularly for drought data, information re-garding the specific month(s) is often not recorded inEM-DAT. ENSO events are rarely contained in a cal-endar year; they typically begin in the boreal spring/summer of one year, peak in the winter, and subse-quently decay the following year. Consequently, the

FIG. 2. (a) Reported hydrometeorological disasters. Relative incidences of floods (dotted), droughts (hatched),and “other” (brown), which include windstorms and temperature extremes, are demarcated. El Niño (thick redlines) and La Niña (thick blue lines) events, consistent with those shown in Fig. 1, are indicated along x axis. Solidgreen line represents time series of total disaster occurrence, detrended using a second-order polynomial (blackdashed line). (b) Relative disaster occurrences using detrended time series shown in (a) for calendar years thatexperienced six or more months of El Niño or La Niña conditions (red and blue circles, respectively) as determinedby the thick red and blue lines from (a), and listed in Table 2. The years that contained more than 6 months ofneutral conditions serve as the baseline (horizontal black lines). (c) Similar to (b), except for the onset and demiseyears of El Niño (red circles) and La Niña (blue circles). The years that contain no months within the intervals ofEl Niño or La Niña events serve as the baseline (horizontal black lines).

1 MARCH 2005 G O D D A R D A N D D I L L E Y 655

Fig 2 live 4/C

Page 6: El Niño: Catastrophe or Opportunity - IRIgoddard/papers/Goddard_Dilley_2005.pdf · The reported “cost” of El Niño events contributes greatly to misconceptions about the global

relative disaster frequency during El Niño/La Niña isanalyzed using two definitions of ENSO event years.First, each year is categorized as El Niño, La Niña, orneutral, depending on which condition was predomi-nant (i.e., existed for more than 6 months of the year)(Fig. 2b, Table 2). Second, relative disaster frequenciesin the year of onset and year(s) of demise for actualENSO events are considered relative to years with noonset or demise (Fig. 2c, Table 2).

Interannually, in either case, the number of hydrom-eteorological-related disasters appears to be largely in-sensitive to the occurrence of ENSO extremes (Figs.2b,c). The Kolmogorov–Smirnov test, which was em-ployed to compare the distributions of disaster frequen-cies in neutral years against those in El Niño or La Niñayears, or against those in the onset/demise years ofENSO events, does not reject the null hypothesis thatthese data are drawn from the same population at the95% confidence level. A similar result is obtained fortesting differences in the means by using a nonparamet-ric version of the t test. Globally, the number of flooddisasters within EM-DAT exhibits no consistent rela-tionship to ENSO events (Fig. 2a). Drought disasters,on the other hand, do tend to occur more frequently inselected regions during the demise year(s) of El Niñoevents (Fig. 2a; Dilley and Heyman 1995; Bouma et al.1997; Thomson et al. 2003). But, overall, hydrometeo-rological disasters experienced globally during El Niñoand La Niña events are statistically not significantlydifferent from what is experienced in neutral years. Di-sasters and associated losses occur in all years.

4. Precipitation anomalies and ENSO events

Moreover, not all disasters, or the climate anomaliescontributing to them, during ENSO events are due tothe ENSO event. The body of research documentingand describing El Niño/La Niña and their effects on

global and regional climate has grown considerably inthe last several decades (McPhaden et al. 1998; Tren-berth et al. 1998; Wallace et al. 1998). The claim that ElNiño may be predictable up to a year in advance2 (Caneet al. 1986), and the recognition that it yields somewhatrepeatable patterns of anomalous climate, has fueledthis wealth of research. These factors have also pro-vided an impetus for seasonal climate forecasting ef-forts (Goddard et al. 2001). One thing the climate com-munity has learned, however, is that the effects of ElNiño/La Niña are region specific. Observational studiesof the last 50 yr indicate that only approximately20%–30% of land areas experience statistically signifi-cant repeatability in the occurrence of categorical pre-cipitation anomalies during ENSO extremes in any par-ticular season (Mason and Goddard 2001). However,due to such factors as differences in the timing andpatterns of ENSO events, climate variability forced bySST anomalies in the tropical Atlantic and tropical In-dian Oceans that is independent of ENSO, and inher-ent uncertainty due to the internal chaotic dynamics ofthe atmosphere, the expected climate anomalies maynot occur during every El Niño or La Niña event forevery region.

To determine whether climate anomalies, in particu-lar, precipitation anomalies, become more severe orwidespread during ENSO events, a precipitation per-turbation index (PPI) is developed,

PPI�t� � �� �P��x, y, t� �

SP��x, y, t�dxdy, �3�

where SP(x,y) is the standard deviation of the precipi-tation anomalies for a given calendar month, P(x,y,t),

2 In practice the useable lead time for these forecasts is muchshorter than the early optimistic claims of 1 yr (e.g., Landsea andKnaff 2000).

TABLE 2. Years associated with analysis presented in Fig. 2, based on the Niño-3.4 index shown in Fig. 1. “Dominant conditions” areindicated if more than 6 months in the calendar year meet the categorical criterion. An “event” is identified as El Niño/La Niñaconditions occurring for at least four consecutive months, in which case the beginning year of the interval is the “onset” year, and theending (or subsequent) years are the “demise” years.

El Niño La Niña Neutral

Dominant conditions Event onset/demise Dominant conditions Event onset/demise Dominant conditions

1976/1977 1975 1975/1976 19771977/1978 1985 1984/1985 1978

1982 1982/1983 1988 1988/1989 19791983 1998 1998/1999, 2000 19801987 1986/1987, 1988 1999 19811991 1991/1992 2000 19841992 19891993 19901994 1994/1995 19951997 1997/1998 19962002 2002/2003 2001

* Even though it qualifies as an event, 1993 is not included in the El Niño onset/demise column because it began and ended in the samecalendar year.

656 J O U R N A L O F C L I M A T E VOLUME 18

Page 7: El Niño: Catastrophe or Opportunity - IRIgoddard/papers/Goddard_Dilley_2005.pdf · The reported “cost” of El Niño events contributes greatly to misconceptions about the global

at each grid point, (x,y). The PPI is an absolute valuemeasure. It represents an integral of normalized anoma-lous precipitation magnitude, or, equivalently, the num-ber of standard deviations that the local precipitationdeparts from the local normal. In this analysis the PPIis integrated over low-latitude (30°S–30°N) land points(Fig. 3), although the integral over global land points(not shown) yields identical results. The PPI indicatesthe overall perturbation to rainfall patterns; positive

and negative precipitation anomalies in different partsof the world do not cancel one another out. For pur-poses of presentation and correlation with ENSO indi-ces, the resulting time series for the PPI has been nor-malized to have zero mean and unit variance relative tothe monthly mean annual cycle for the 1950–95 period.

The PPI does not correlate with ENSO indices, suchas Niño-3.4 (Fig. 3a). Because the PPI is an absolutevalue measure of rainfall anomalies, the time series is

FIG. 3. PPI integrated over low latitudes (30°S–30°N) from two observed precipitation datasets (red: UEA CRUand blue: GHCN). (a) PPI using all land points, and (b) PPI using a subset of land points for which the magnitudeof precipitation anomalies exceeds 1�. (c) Number of grid points for which the magnitude of the precipitationanomaly exceeds 1�. Niño-3.4 index of ENSO is plotted for comparison in each graph (black). All time series havebeen normalized to have zero mean and unit variance relative to the monthly mean annual cycle over the 1950–95period.

1 MARCH 2005 G O D D A R D A N D D I L L E Y 657

Fig 3 live 4/C

Page 8: El Niño: Catastrophe or Opportunity - IRIgoddard/papers/Goddard_Dilley_2005.pdf · The reported “cost” of El Niño events contributes greatly to misconceptions about the global

compared against the absolute value of Niño-3.4 also,which can be interpreted as the magnitude of forcingfrom the tropical Pacific. The PPI does not correlatewith the absolute value of Niño-3.4 either.

Two similar analyses are carried out with variationson the PPI, comparing the perturbation time seriesagainst Niño-3.4 and its absolute value. Both cases fo-cus on more extreme rainfall impacts, with similar re-sults. In the first case, the PPI considers only landpoints for which the magnitude of the normalizedanomaly exceeds one standard deviation (Fig. 3b); thisvariant focuses on the overall severity of extremeanomalies. One could imagine the possibility that thePPI was relatively constant over time, resulting fromspatially concentrated extreme precipitation anomaliesduring El Niño/La Niña conditions and weaker, butmore widespread, anomalies during neutral years.When including in the PPI only the extreme precipita-tion anomalies, these potential differences between ElNiño/La Niña and neutral conditions would be high-lighted. The other case counts the number of landpoints for which the normalized anomaly exceeds onestandard deviation (Fig. 3c); this variant focuses on therelative spatial coverage of extreme anomalies but nottheir magnitude. In both cases, the resulting precipita-tion time series bears no correlation to the occurrenceor magnitude of ENSO extremes.

Although other research suggests that the spatial ex-tent of below-normal rainfall increases in response toEl Niño events (Lyon 2004, manuscript submitted to J.Climate), leading to an increase in reported droughtdisasters in some regions (Dilley and Heyman 1995;Bouma et al. 1997), in other years equivalently anoma-lous precipitation is observed in the same or differentparts of the world. No linear relationship exists be-tween ENSO extremes and the overall frequency orseverity of precipitation anomalies. That does not nec-essarily mean that ENSO extremes have no influenceon the overall frequency or severity of precipitationanomalies.

Distributions of the rainfall indices relative to theNiño-3.4 index of ENSO do show a discernible, thoughsubtle, change during El Niño and La Niña conditionscompared to neutral conditions (Fig. 4). The clearestshift in the frequency distribution exists for the PPI ofall anomalies (Fig. 4a2), but this shift represents anincrease of only about 3%–4% in the mean relative tothe neutral conditions (Table 3). For the PPI of largeanomalies, the distributions during El Niño and neutralconditions are statistically indistinguishable at the 95%confidence level, according to the Kolmorgorov–Smirnov test; and for the spatial coverage of large pre-cipitation anomalies, the two rainfall datasets give dif-fering results for the uniqueness of the frequency dis-tribution under La Niña conditions compared to that ofneutral conditions (Table 3).

A relevant finding illustrated in Fig. 4 is that thelargest values of the rainfall indices are not greater dur-

ing ENSO extremes than during neutral conditions.Changes in the expected frequency of “high” values ofthe indices (i.e., exceeding 1�) are also small (Table4), typically at only a few percent. The largest differ-ences between neutral and El Niño/La Niña conditionsexists in the “low” tail of the distribution, with overallless perturbation more likely under neutral conditionscompared to ENSO extremes. La Niña conditions leadto the most notable increase in the frequency of highvalues—those for the PPI of large anomalies (20%compared to the �14% observed under neutral condi-tions). The high values of the precipitation indices areof primary concern because these indicate the greatestpotential for climatic hazard events. Although ENSOextremes do modify the expected frequencies of thosehigh values relative to neutral conditions, the differenceis small. The risk of severe and/or widespread rainfallanomalies under neutral conditions remains compa-rable to that during ENSO extremes.

This is not intended to discount the devastating cli-matic hazard events during El Niño and La Niña eventsthat frequently occur in ENSO teleconnection regions(Ropelewski and Halpert 1987; Ropelewski and Hal-pert 1989; Mason and Goddard 2001), such as thedroughts over parts of southern Africa and southeast-ern Asia and the excessive rainfall and flooding oversoutheastern South America and parts of the GreatHorn of Africa during El Niño events. Many countriesof the world have come to expect adverse climate con-ditions from El Niño/La Niña events, which in manycases lead to loss of life and property. But because theaggregate rainfall anomalies do not change much dur-ing ENSO events, adverse climate conditions will occursomewhere in all years.

5. Climate predictability and ENSO events

The big difference during ENSO events is that cli-mate forecasts are potentially more skillful at thosetimes, as is demonstrated below for the 1950–95 period.Similar conclusions were reached by Brankovic andPalmer (2000) and Shukla et al. (2000) based on thepredictability of 500-mb heights in the Northern Hemi-sphere. However, the current analysis arrives at thatconclusion from a different perspective, and for a farmore socially relevant variable (i.e., precipitation).

The first point is that, relative to neutral conditions,skillful forecasts can be made for a greater proportionof land areas during El Niño/La Niña conditions (Fig.5). The percent of land area for which precipitationforecasts achieve good skill (i.e., where RPSS exceeds5%, see section 2d) increases markedly in all seasons.However, the greatest relative increases appear nearthe end of the year, when ENSO events typically reachtheir peak magnitude. At that time, the amount of landareas for which forecasts potentially have good skilldoubles, globally, during ENSO extremes relative toneutral conditions.

658 J O U R N A L O F C L I M A T E VOLUME 18

Page 9: El Niño: Catastrophe or Opportunity - IRIgoddard/papers/Goddard_Dilley_2005.pdf · The reported “cost” of El Niño events contributes greatly to misconceptions about the global

The second point is that where general predictabilityof the climate exists, forecast skill increases as ENSOextremes become stronger (Fig. 6). Seasonal climateconsists of “signal,” due to the boundary forcing pri-

marily from changing SST patterns (e.g., Goddard et al.2001), and “noise,” due to the chaotic influences of theatmosphere’s internal dynamics. As El Niño or La Niñaconditions become stronger, the signal becomes more

FIG. 4. Distribution of rainfall indices relative to ENSO category, using the UEA CRU rainfall data. (a1)Scatterplot of monthly mean PPI index calculated over all low-latitude land points (30°S–30°N) vs monthly meanENSO category (La Niña conditions: blue; neutral conditions: black; El Niño conditions: red). (a2) Associatednormalized frequency distributions of PPI values shown in (a1) in each ENSO category. (b1), (b2) Similar to (a1)and (a2), except that the PPI integration includes only those rainfall anomalies exceeding 1� magnitude. (c1), (c2)Similar to (b1) and (b2), except the index quantifies relative spatial coverage of rainfall anomalies exceeding 1�magnitude rather than their strength. All data have been normalized to have zero mean and unit variance relativeto the monthly mean annual cycle over the 1950–95 period. A similar figure based on GHCN rainfall data, notshown, is nearly identical.

1 MARCH 2005 G O D D A R D A N D D I L L E Y 659

Fig 4 live 4/C

Page 10: El Niño: Catastrophe or Opportunity - IRIgoddard/papers/Goddard_Dilley_2005.pdf · The reported “cost” of El Niño events contributes greatly to misconceptions about the global

discernible relative to the noise, and the likelihood forexperiencing an “expected” climate anomaly due to theENSO extreme becomes greater.

The linear relationship between the strength of theNiño-3.4 index and the probabilistic skill of three cat-egories of rainfall forecasts calculated over global landareas exceeds the 99% confidence level3 for statisticalsignificance in all seasons for both positive and negativevalues of Niño-3.4 [e.g., the October–November–December (OND) season shown in Fig. 6]. The in-crease in average forecast skill during ENSO extremesrelative to ENSO neutral conditions is 99% statisticallysignificant4 in all seasons, except for July–August–September (JAS) during El Niño conditions where thedifference only exceeds the marginal significance of90%. In general, the difference in forecast skill betweenLa Niña and El Niño events is not significant. Further-more, within the El Niño category, the linear relation-ship between forecast skill and El Niño strength ex-ceeds 99% significance in all seasons. The linear rela-tionship between forecast skill and La Niña strengthexceeds the 97.5% level of significance during OND,when events tend to peak, and during January–February–March (JFM), when events are weakening,but atmospheric noise, particularly in the NorthernHemisphere extratropics, is relatively low (Kumar andHoerling 1998). Even over areas without robust ElNiño/La Niña teleconnections, climate predictions ben-efit from a stronger signal-to-noise ratio during ENSOextremes (Fig. 6b).

Much of the increase in overall forecast skill (Fig. 6a)during ENSO events results from increases in regionalskill over places that exhibit a statistically significantrepeatability in precipitation response to ENSO events(Fig. 6c, Fig. 7; Mason and Goddard 2001). For ex-ample, the forecasts exhibit substantially higher skill forENSO events in JFM over Hawaii, the southern tier ofthe United States, southern Africa, eastern China, andthe Philippines; in April–May–June (AMJ) over north-eastern Brazil and the southern and central United

States; in JAS over the Caribbean, western Sahel, In-donesia, and Australia; and in OND over Indonesia,eastern Africa, southeastern Brazil, the southernUnited States, and Mexico. Given the difficulty in de-termining the statistical significance of RPSS, however,these maps of skill improvement (Fig. 7) should be in-terpreted qualitatively. The weak and incoherent ap-pearance of the negative regions infers little change inthe skill owing to El Niño/La Niña. For the few semi-coherent regions of reduced skill, such as parts of In-donesia and the northern part of the United States forOND, it is likely that forecast skill actually degradesduring ENSO extremes for the forecast methodologyapplied here. Where coherent regions show improve-ment, particularly for differences in the RPSS that ex-ceed 10%, one can assume that the seasonal forecastswill perform better during ENSO extremes.

These results (Figs. 6 and 7) constitute an approxi-mate upper limit in forecast skill for seasonal precipi-tation because the dynamical models’ atmosphereswere forced with observed SSTs. In a real forecast set-ting skills decrease because imperfect SST predictionsintroduce additional uncertainty. At this time the mostskillful SST predictions are for the tropical Pacific (i.e.,for El Niño/La Niña). So, actual forecast skill will de-crease even more during neutral conditions than duringENSO extremes. Thus, the increase in skill duringENSO events, that is realizable in practice is underes-timated here. Increasing attention is being focused onpredictions for the other tropical ocean basins, as theimportance of accurate SSTs in the tropical Indian andAtlantic Oceans for dynamical climate prediction be-comes recognized (Goddard and Graham 1999; God-dard and Mason 2002). At this time, most of the fore-cast skill that resides in these basins is that due tochanges associated with ENSO (Penland and Ma-trosova 1998). Less observational data exist for thetropical Atlantic and Indian Oceans, and less is under-stood about the dynamics of interannual variability inthese basins that are independent of ENSO. Thus, con-siderable room for improvement remains for SST pre-dictions in these regions.

6. Summary and discussion

Conclusions from the above analysis for global, oreven tropical, scales include the following:

3 Statistical significance of linear relationships is based on thePearson product-moment correlation coefficient, testing the nullhypothesis that the correlation coefficient differs from 0.

4 Statistical significance of difference in means or medians isbased on the Mann–Whitney U test, a nonparametric analog ofthe t test.

TABLE 3. Results from the Kolmogorov–Smirnov (KS) test for uniqueness of sample distributions (1 � can reject null hypothesis thatsample populations are effectively the same, at the 95% confidence level), and percent difference in the population mean for ENSOextremes relative to neutral conditions. The first value is based on monthly mean gridded rainfall data from UEA CRU; the secondparenthetical value is based on GHCN gridded rainfall data.

All P (PPI) Extreme P (PPI) Extreme P (spatial coverage)

KS test Mean diff (%) KS test Mean diff (%) KS test Mean diff (%)

El Niño vs neutral 1 (1) 3.5 (4.5) 0 (0) 0.6 (�0.8) 1 (1) 3.8 (6.0)La Niña vs neutral 1 (1) 3.6 (2.3) 1 (1) 1.9 (1.9) 1 (0) 3.4 (1.5)

660 J O U R N A L O F C L I M A T E VOLUME 18

Page 11: El Niño: Catastrophe or Opportunity - IRIgoddard/papers/Goddard_Dilley_2005.pdf · The reported “cost” of El Niño events contributes greatly to misconceptions about the global

• Overall perturbation to precipitation over land areasis only weakly affected by ENSO extremes. The riskof widespread extreme precipitation anomalies dur-ing ENSO extremes is comparable to that duringneutral conditions. Furthermore, the highest valuesof integrated rainfall perturbation are not greaterduring ENSO extremes than during neutral condi-tions.

• The frequency of reported climate-related disastersdoes not increase during El Niño/La Niña years rela-tive to neutral years. The degree to which overalllosses may increase during ENSO events has notbeen estimated, but in the absence of increases in themagnitude of global aggregate climate anomaliesduring ENSO events any such increases would be dueto socioeconomic, rather than climatic, causal factors.

• Forecast skill of seasonal rainfall increases, in mag-nitude and coverage, during ENSO extremes.

It is hoped that these results will motivate furtheranalysis and improved representation of El Niño/LaNiña–related climatic hazards and associated disasters.Disasters occurring during ENSO extremes must bediscussed within the context of baselines, such as disas-ter occurrences in previous years and in neutral years.More effort should be given to understanding the attri-bution of disaster causality; would they have occurredin the absence of the ENSO event? More comprehen-sive and precise disaster databases and more compre-hensive estimates of losses will aid in such research.Also, positive benefits of ENSO extremes should bemore carefully documented, yielding a more complete

FIG. 5. Percentage of land area over which seasonal precipitation forecasts exhibit “good skill” (see section 2d)as measured by RPSS over the 1950–95 period. (top) RPSS that is averaged globally, and (bottom) RPSS that isaveraged over low latitudes (30°S–30°N), for ENSO extremes (solid black), and neutral conditions (dashed black).Gray line (right scale) indicates the relative increase in land area for which good skill exists in ENSO extremes vsneutral conditions.

TABLE 4. Observed frequencies of categorical monthly mean rainfall indices (Low: ��1�; �1� � Med �1�; High: �1�) under ElNiño, La Niña, and neutral conditions. The first value is based on gridded rainfall data from UEA CRU; the second parenthetical valueis based on GHCN gridded rainfall data.

All P (PPI) Extreme P (PPI) Extreme P (spatial coverage)

Low Medium High Low Medium High Low Medium High

El Niño 11% (10%) 72% (68%) 17% (22%) 17% (16%) 68% (77%) 15% (7%) 9% (8%) 75% (71%) 16% (21%)Neutral 23% (19%) 65% (69%) 12% (12%) 17% (15%) 70% (70%) 13% (15%) 20% (17%) 66% (71%) 14% (12%)La Niña 10% (9%) 70% (78%) 20% (13%) 10% (10%) 70% (72%) 20% (19%) 9% (11%) 72% (78%) 19% (11%)

1 MARCH 2005 G O D D A R D A N D D I L L E Y 661

Fig 5 live 4/C

Page 12: El Niño: Catastrophe or Opportunity - IRIgoddard/papers/Goddard_Dilley_2005.pdf · The reported “cost” of El Niño events contributes greatly to misconceptions about the global

appreciation of the socioeconomic impacts of El Niñoand La Niña events. Finally, increased efforts should bedirected toward improving seasonal climate predictioncapabilities for neutral ENSO conditions, because over-all rainfall perturbations can be just as great as duringENSO extremes.

At this time, given that society is faced with climate-related disasters in all years, seasonal climate forecastscan make a targeted contribution to reducing the asso-ciated losses. Because overall forecast skill is highestduring El Niño/La Niña, advanced warning that anENSO event is likely to develop and reliable climateforecasts for regions that are likely to be affected allows

international, regional, and national agencies and localcommunities to prepare. Vulnerability reduction canpotentially reduce the frequency and severity of hy-drometeorological disasters incurred in those years.The United States realized such a reduction in adverseimpacts during the 1997/98 El Niño event; Californiaestimated a 1 billion dollar savings in property damagesdue to better preparedness in response to seasonal cli-mate forecasts (Chagnon 1999; Weiher 1999). Further-more, beneficial effects of ENSO have also been noted.For example, tropical Atlantic hurricanes that threatenthe southeastern United States, the Caribbean, andeastern Central America occur less frequently during

FIG. 6. Forecast skill (RPSS) for three categories of forecasts of OND rainfall for the 1950–95 period. Skill scoresfor each year are ordered by the corresponding OND Niño-3.4 index. (a) RPSS is calculated globally over landpoints for which the temporal RPSS over all years exceeds 5%. (b) Similar to (a), but including only those pointswith no statistically significantly repeatability of climate response to El Niño/La Niña. (c) Similar to (a), butincluding only those points with statistically significantly repeatability of climate response to El Niño/La Niña. Barsare color-coded by their respective category: La Niña (blue), neutral (black), and El Niño (red).

662 J O U R N A L O F C L I M A T E VOLUME 18

Fig 6 live 4/C

Page 13: El Niño: Catastrophe or Opportunity - IRIgoddard/papers/Goddard_Dilley_2005.pdf · The reported “cost” of El Niño events contributes greatly to misconceptions about the global

FIG. 7. Differences in skill (RPSS) for three categories of seasonal rainfall forecasts between ENSO extremesand neutral conditions for the 1950–95 period. Positive values indicate higher skill during ENSO extremes.

1 MARCH 2005 G O D D A R D A N D D I L L E Y 663

Fig 7 live 4/C

Page 14: El Niño: Catastrophe or Opportunity - IRIgoddard/papers/Goddard_Dilley_2005.pdf · The reported “cost” of El Niño events contributes greatly to misconceptions about the global

El Niño years (Gray 1984). Also, warmer winter tem-peratures commonly are observed in the northernUnited States during El Niño, leading to less energy useand, therefore, lower energy prices (Chagnon 1999).Thus, between mitigating adverse climate effects andtaking advantage of beneficial ones through the pru-dent use of climate forecasts, El Niño and La Niñayears may eventually result in substantially lower so-cioeconomic losses, globally, than are realized in otheryears.

Acknowledgments. The authors thank Mark Cane,Dave DeWitt, Simon Mason, Tony Barnston, KevinTrenberth, and two anonymous reviewers for com-ments on this work. The authors are supported by acooperative agreement from the National Oceanic andAtmospheric Administration (NOAA) NA07GP0213.The views expressed herein are those of the authorsand do not necessarily reflect the views of NOAA orany of its subagencies.

REFERENCES

Barnston, A. G., M. Chelliah, and S. B. Goldenberg, 1997: Docu-mentation of a highly ENSO-related SST region in the equa-torial Pacific. Atmos.–Ocean, 35, 367–383.

——, S. J. Mason, L. Goddard, D. G. DeWitt, and S. E. Zebiak,2003: Increased automation and use of multimodel ensem-bling in seasonal climate forecasting at the IRI. Bull. Amer.Meteor. Soc., 84, 1783–1796.

Benson, C., and E. J. Clay, 1998: The impact of drought on sub-saharan African economies: A preliminary examination. TheWorld Bank Tech. Paper 401, 80 pp.

Bouma, M. J., R. S. Kovats, S. A. Goubet, J. S. H. Cox, and A.Haines, 1997: Global assessment of El Niño’s disaster bur-den. Lancet, 350, 1435–1438.

Brankovic, C., and T. N. Palmer, 2000: Seasonal skill andpredictability of ECMWF PROVOST ensembles. Quart. J.Roy. Meteor. Soc., 126, 2035–2067.

Cane, M. A., S. C. Dolan, and S. E. Zebiak, 1986: Experimentalforecasts of El Niño. Nature, 321, 827–832.

Chagnon, S. A., 1999: Impacts of 1997–98 El Niño generatedweather in the United States. Bull. Amer. Meteor. Soc., 80,1819–1828.

Dilley, M., and B. N. Heyman, 1995: ENSO and disaster:Droughts, floods and El Niño/Southern Oscillation warmevents. Disasters, 19, 181–193.

Eischeid, J. K., H. G. Diaz, R. S. Bradley, and P. D. Jones, 1991:A comprehensive precipitation data set for global land areas.U.S. Department of Energy Carbon Dioxide Research Divi-sion Tech. Rep. TR051, 81 pp.

EM-DAT, cited 2003: The OFDA/CRED international disasterdatabase. Université Catholique de Louvain. [Available on-line at http://www.cred.be.]

Epstein, E. S., 1969: A scoring system for probability forecasts ofranked categories. J. Appl. Meteor., 8, 985–987.

Goddard, L., and N. E. Graham, 1999: The importance of theIndian Ocean for simulating rainfall anomalies over easternand southern Africa. J. Geophys. Res., 104, 19 099–19 116.

——, and S. J. Mason, 2002: Sensitivity of seasonal climate fore-casts to persisted SST anomalies. Climate Dyn., 19, 619–631.

——, ——, S. E. Zebiak, C. F. Ropelewski, R. Basher, and M. A.Cane, 2001: Current approaches to seasonal-to-interannualclimate predictions. Int. J. Climatol., 21, 1111–1152.

Gray, W. M., 1984: Atlantic seasonal hurricane frequency. Part I:El Niño and 30 mb quasi-biennial oscillation influences. Mon.Wea. Rev., 112, 1649–1668.

Hack, J. J., J. T. Kiehl, and J. W. Hurrell, 1998: The hydrologicand thermodynamic characteristics of the NCAR CCM3. J.Climate, 11, 1179–1206.

Hulme, M., 1992: A 1951–80 global land precipitation climatologyfor the evaluation of General Circulation Models. ClimateDyn., 7, 57–72.

——, 1994: Validation of large-scale precipitation fields in generalcirculation models. Global Precipitations and ClimateChange, M. Desbois and F. Desalmand, Eds., NATO ASISeries, Springer-Verlag, 387–406.

Kumar, A., and M. P. Hoerling, 1998: Annual cycle of Pacific–North American seasonal predictability associated with dif-ferent phases of ENSO. J. Climate, 11, 3295–3308.

Landsea, C. W., and J. A. Knaff, 2000: How much skill was therein forecasting the very strong 1997–98 El Niño? Bull. Amer.Meteor. Soc., 81, 2107–2120.

Mason, S. J., 2004: On using “climatology” as a reference strategyin the Brier and ranked probability skill scores. Mon. Wea.Rev., 132, 1891–1895.

——, and L. Goddard, 2001: Probabilistic precipitation anomaliesassociated with ENSO. Bull. Amer. Meteor. Soc., 82, 619–638.

McPhaden, M. J., and Coauthors, 1998: The Tropical Ocean-Global Atmosphere observing system: A decade of progress.J. Geophys. Res., 103, 14 169–14 240.

Penland, C., and L. Matrosova, 1998: Prediction of tropical At-lantic sea surface temperatures using linear inverse modeling.J. Climate, 11, 483–496.

Rajagopalan, B., U. Lall, and S. E. Zebiak, 2002: Categoricalclimate forecasts through regularization and optimal combi-nation of multiple GCM ensembles. Mon. Wea. Rev., 130,1792–1811.

Robertson, A. W., U. Lall, S. E. Zebiak, and L. Goddard, 2004:Improved combination of multiple atmospheric GCM en-sembles for seasonal prediction. Mon. Wea. Rev., 132, 2732–2744.

Roeckner, E., and Coauthors, 1996: The atmospheric general cir-culation model ECHAM4: Model description and simulationof present-day climate. Max-Planck-Institüt für MeteorologieRep. 218, 90 pp.

Ropelewski, C. F., and M. S. Halpert, 1987: Global and regionalscale precipitation patterns associated with the El Niño/Southern Oscillation. Mon. Wea. Rev., 115, 1606–1626.

——, and ——, 1989: Precipitation patterns associated with thehigh index phase of the Southern Oscillation. J. Climate, 2,268–284.

Sen, A. K., 1981: Poverty and Famines: An Essay on Entitlementand Deprivation. Clarendon Press, 257 pp.

Shukla, J., and Coauthors, 2000: Dynamical seasonal prediction.Bull. Amer. Meteor. Soc., 81, 2593–2606.

Smith, T. M., and R. W. Reynolds, 2004: Improved extended re-construction of SST (1854–1997). J. Climate, 17, 2466–2477.

Sponberg, K., 1999: Navigating the numbers of climatological im-pact. Compendium of Climatological Impacts, UniversityCorporation for Atmospheric Research, Vol. 1, NationalOceanic and Atmospheric Administration, Office of GlobalPrograms, 13 pp. [Available online at http://www.cip.noaa.gov/docs/navimpact.pdf.]

Thomson, M. C., K. Abayomi, A. G. Barnston, M. Levy, and M.Dilley, 2003: El Niño and drought in southern Africa. Lancet,361, 437–438.

Trenberth, K. E., 1997: The definition of El Niño. Bull. Amer.Meteor. Soc., 78, 2771–2777.

——, G. W. Branstator, D. Karoly, A. Kumar, N.-C. Lau, and

664 J O U R N A L O F C L I M A T E VOLUME 18

Page 15: El Niño: Catastrophe or Opportunity - IRIgoddard/papers/Goddard_Dilley_2005.pdf · The reported “cost” of El Niño events contributes greatly to misconceptions about the global

C. F. Ropelewski, 1998: Progress during TOGA in under-standing and modelling global teleconnections associatedwith tropical sea surface temperatures. J. Geophys. Res., 103,14 291–14 324.

UCAR, 1994: El Niño and climate prediction. Reports to theNation on our Changing Planet, UCAR, No. 3, 25 pp.

Wallace, J. M., E. M. Rasmusson, T. P. Mitchell, V. E. Kousky,E. S. Sarachik, and H. von Storch, 1998: On the structure andevolution of ENSO-related climate variability in the tropical

Pacific: Lessons from TOGA. J. Geophys. Res., 103, 14 241–14 259.

Weiher, R. F., Ed., 1999: Improving El Niño forecasting: Thepotential benefits. U.S. Dept. of Commerce, National Atmo-spheric and Oceanic Administration, 57 pp. [Available onlineat http://www.publicaffairs.noaa.gov/worldsummit/pdfs/improving.pdf.]

Willks, D. S., 1995: Statistical Methods in the Atmospheric Sci-ences. Academic Press, 467 pp.

1 MARCH 2005 G O D D A R D A N D D I L L E Y 665