Global Precipitation During the 1997-98 E1 Nifio and Initiation of the 1998-99 La Nifia By Scott Curtis ""- Joint Center for Earth Systems Technology University of Maryland Baltimore County Laboratory for Atmospheres NASA/Goddard Flight Center and Robert Adler, Laboratory for Atmospheres NASA/Goddard Flight Center George Huffman, Eric Nelkin, David Bolvin Science Systems Applications Inc. Laboratory for Atmospheres NASA/Goddard Flight _enter Abstract: The 1997-99ENSOcyclewasvery powerful,but alsowell observed.Thebest satelliterainfall estimatescombinedwith gaugeobservationsallow for aglobalanalysis ofprecipitationanomaliesaccompanyingthe 1997-98E1Nifio andinitiation ofthe1998- 99La Nifia. For theperiodApril 1997to March 1998the centralto easternPacific, southeasternandwesternU.S.,Argentina,easternAfrica, SouthChina,easternRussia, andNorth Atlantic wereall morethan two standarddeviationswetter than normal. During thesameyeartheMaritime Continent,easternIndianOcean,subtropicalNorth Pacific,northeasternSouthAmerica,andmuchofthemid-latitudesouthernoceanswere morethantwo standarddeviationsdrierthannormal.An analysisoftheevolutionofthe E1Nifio and accompanyingprecipitation anomaliesrevealedthat a dry Maritime ContinentledtheformationoftheE1Nifio SST,whilein thecentralPacific,precipitation anomalieslaggedtheE1Nifio SSTby a season.A rapidtransitionfrom E1Nifio toLa Nifiaoccurredin May 1998,butasearlyasOctober-Novembe1r997precipitationindices capturedprecursorchangesin Pacific rainfall afiomalies. Differences were found betweenobservedandmodeled(NCEPfNCARreanalysis)precipitation anomaliesfor 1997and98. Inparticular,themodelhadabiastowardspositiveprecipitationanomalies andthemagnitudesoftheanomaliesintheequatorialPacificweresmallcomparedtothe observations.Also, theevolutionof theprecipitationfield, includingthedrying of the Maritime Continentandeastwardprogressionof rainfall in the equatorialPacific,was lesspronouncedforthemodelcomparedtotheobservations.Onedegreedaily estimates of rainfall showclearly theMaddenJulianOscillation andrelatedwesterlywind burst eventsovertheMaritimeContinent,whicharekeyindicatorsfortheonsetofE1Nifio. 1.Introduction It canbearguedthatthe 1997-98E1Nifio wasthestrongestENSOeventever recorded. The unusuallywarmwatersof theequatorialPacificwereaccompaniedby changesinthe largescalecirculation,followedbyprecipitationanomaliesrangingfrom severedroughtto floods. The Maritime Continent,AmazonandCongobasins,and Central America experienceddrought during somepart of 1997and 1998,while Argentina,Peru,andEastAfrica werehardesthit by flooding:.,__fIanct,therehavebeen manystudiesdocumentingthevariability ofregionalprecipitationduringthe97-98event (Montroyetal. 1998,Bell andHalpert 1998,Bell etal. 1999,Jaksic1998,Mullen 1998, PaviaandBadan 1998,Harrison andLarkin 1998,Kogan 1998,Jensenet al. 1998, McPhadden1999). The E1Nifio wasfollowed by a persistentLa Nifia. The regional precipitation anomalies for this phase of the ENSO cycle have been less well documented.In this studysatelliteobservationsandgaugesareusedto examine global precipitation anomalies associated with the 1997-98 El Nifio and 1998-99 La Nifia. Another focus is the timing of the onset and decay of precipitation anomalies over the Maritime Continent and East Pacific associated with the evolution of ENSO. Finally, the observations will be compared to model generated precipitation during this period. An experimental version of the Global Precipitation Climatology Project's (GPCP) community data set (reference section 2) was used to construct precipitation anomalies for the year average April 1997 to March 1998 (Fig. 1). Overlain is Ropelewski and Halpert's (1987) (hereafter RH) schematic figure showing areas typically dry and wet during an E1 Nifio. Ropelewski and Halpert's pioneering work was a compilation of regional studies of precipitation during ENSO. The paucity of surface 2 observationsrestrictedaglobalanalysis.Recentdataproductsbasedonremotelysensed precipitationestimatesinvited arevisitationofglobalprecipitationanomaliesassociated with ENSOevents,in particularthepowerful97-98event. 2. DataandIndices Theprimary datasourcefor this studyis anexperimentalversionof theGlobal PrecipitationClimatology Project's(GPCP;Huffmanetal. i997) communitydataset, a hereafter referred to as GPCPx. This data set is like Huffman et al. (1997) except there has been a slight modification in the merger technique; it uses TOVS data (Susskind et al. 1997) to fill in missing or uncertain data in the high latitudes, and the record extends from January 1979 to July 1999. GPCPx is compared with NCEP/NCAR's reanalysis product (Kalnay et al. 1996). Precipitation is computed from a 62 wave triangular, 28 layer spectral model "Medium Range Forecast" run. For the examination of smaller time and sl_ace scales, an experimental one degree daily (1DD; Huffman et al. 1999) precipitation analysis_was used. !DD includes precipitation estimates from geosynchonous IR, and TOVS. Each day of the daily combination is scaled such that each month of data sums to the corresponding GPCPx month of data. This is done to ensure consistency between GPCP products and introduce gauge data as a constraint on 1DD. The product currently extends from January 1997 to December 1998. NCEP/NCAR also produces a daily reanalysis precipitation data set, with zonal grid spacing of 1.875 °, and this was used to compare with 1DD. Finally,theENSOwastrackedusinganobservedSSTanomalydatasetat0.5° resolution(ReynoldsandSmith1995). Severalindices wereusedto monitor monthly temperatureandprecipitation anomaliesinthePacificbasinandMaritime Continent.Following theClimatePrediction Center'sanalysis,SSTandprecipitationwerecomputedwithin thecoreof theMaritime Continent(INDO; Fig. 2). Nino 3.4 (Fig. 2) waschosento quantify SSTin theEast Pacific. A precipitationindexwasalsoconstructedfor theNino_3.4block. In addition, areaaveragesof precipitation,thesizeof INDO andNino3.4, weremovedthroughout largerdomainsencompassingtheMaritime Continent(mc) andEastPacific(p) (Fig.2) asdescribedby Curtis andAdler (1999). Moving blocks areespecially useful in describingprecipitationbecauseofthespatiallyvaryingnatureofrainfall. Themc-value denotestheminimumvaluein themovingblocksfoundwithin mc andp+ themaximum valuefoundwithin p. Amc- andAp+refertotheminimum andmaximumprecipitation anomaliesrespectively. Thus,theseindices capturethe centersof action over the Y Maritime Continent and central Pacific and can often lead fixed indices in detecting climate change. The normalized difference of Ap+ minus Amc- equals the E! Nifio Index (EI), a measure of the westward gradient of rainfall anomalies. The normalized difference of Amc+ minus Ap- equals the La Nifia Index (LI), a measure of the eastward gradient of rainfall anomalies. The El Nifio Index minus the La Nifia Index yields the ENSO Precipitation Index (ESPI). The normalized ESPI has been shown to be well correlated with Nino 3.4 and Southern Oscillation Index (SOI) and a good measure of the strength of the Walker circulation (Curtis and Adler 1999). 4 3. MonthlySSTandPrecipitationfor 1996-99 This sectionusesmapsandindicesof SST andprecipitation to describethe evolutionof the1997-98E1Nifioandinitiationofthe1998-99LaNifia. Inparticular,the tropicsareexaminedtoidentify SSTandprecipitationsignalswhichmaybeprecursors totheonsetordecayof ENSOevents. 3.1 Global Anomalies Figs. 3 to 6 describe the evolution of the El Nifio and initiation of the La-Nifia in terms of global GPCPx precipitation anomalies. Bimonthly averages remove the 30-60 day oscillations and are thus useful in analyzing interannual variability. The monthly GPCPx fields for 1997-99 and climatologies were also used in the analysis, but not shown here. 1995-96 was characterized by a mild La Nifia. It was the fourth strongest according to 20 years of ESPI data, but relatively weak compared to long-term SST and SOI records. In January-February 1997 (Fig. 3a,b), the La Nifia pattern was weak, with negative SST and precipitation anomalies in the cenfral to east Pacific and wet conditions straddling the equator over the Maritime Continent. A dry anomaly extended from the eastern equatorial Indian Ocean to Indonesia. The relation between this feature and the onset of the E1 Nifio will be described further in section 3.2. In March-April 1997 a warm anomaly was located off the coast of Peru (Fig. 3c). In the East Pacific the typical double Intertropical Convergence Zone (ITCZ) was absent, while in the West Pacific the ITCZ and South Paciifc Convergence Zone (SPCZ) were further apart than normal. This pattern of precipitation was consistent with negative precipitation anomalies over the Maritime Continent, extending eastward in a band along theequatorandpositiverainfall anomalieslocatedimmediatelytothe northandsouth (Fig.3d). InMay-June1997(Fig.3e,f) warmE1Nifiowatershadextendedinto thecentral Pacificandpositiveanomalieswerefoundoff thewestcoastofNorthAmerica. Therain bandaccompanyingtheITCZ hadbeguntobroaden,intensify,andmigratesouthward, consistentwith a weakeningof the Walker circulation. Negative anomaliesover SoutheasAt siawereproducedbyadelayintheonsetofthe1997summermonsoon. .-,..v-- July-August 1997 (Fig. 3g, h) the positive precipitation anomaiies'in the In equatorial Pacific continued to strengthen and the Asian monsoonal rainfall was heavier than normal. By September-October 1997 (Fig. 4a, b) the Indian Ocean had begun to warm while rainfall increased over East Africa. At the same time there was a decrease in convective precipitation over the Congo and Amazon basins. The temperature anomalies in the East Pacific reached a maximum in November-December 1997 (Fig. 4c). Precipitationwithin the SPCZ and Pacific ITCZ merged, forming a region of positive precipitation anomaly east of the date line (Fig. 4d). The Atlantic ITCZ was anomalously dry and northern Argentina and the Southeast U.S. were anomalously wet during these months. The largest precipitation anomalies in the Pacific occurred during January-February 1998 (Fig. 4f) after the El Nifio (in terms of SST) had begun to weaken (Fig. 4e). Anomalously dry regions straddled the wet equatorial Pacific, providing evidence that during this stage of the ENSO an anomalous meridional, rather than zonal, circulation was dominant (Curtis and Hastenrath 1997). This will be 6 discussed in the context of the transition from El Nifio to La Nifia in section 3.2. The North Pacific storm track crossed the west coast of North America further south than normal bringing dry conditions to coastal Alaska and heavy rains to the Pacific Northwest. By March-April 1998 (Fig. 4g, h) positive precipitation anomalies in the eastern Pacific had weakened. Northeast Brazil was anomalously dry but the Maritime Continent drought had begun to end. o A weak E1 Nifio (in terms of SST) in May-June 1998 was accompanied by mostly negative precipitation anomalies in the Pacific (Fig. 5a, b). However, rainfall anomalies in excess of 10 mm day -_remained off the coast of Peru. Positive precipitation anomalies were replaced by negative anomalies in the Southeast U.S. In July-August 1998 the SST pattern in the Pacific was more characteristic of a La Nifia than an El Niho and the Walker circulation began to strengthen as positive precipitation anomalies covered the East Indian Ocean and western Maritime Continent and negative anomalies dominated the central equatorial Pacific (Fig. 5c, d). September-October 1998 (Fig. 5e, f) was characterized by a strengthening La Niha and cooling of the Indian Ocean. Heavy rains accompanied the North American summer monsoon. From November-December 1998 to March-April 1999 (Figs. 5g, h and 6) a steep gradient of anomalous precipitation was formed in the western Pacific, as the convection center of the Maritime Continent and SPCZ continued to intensify and the largest negative precipitation anomalies shifted westward. In early 1999 (Fig. 6) enhanced rainfall was observed over the Amazon basin and southeastern Africa, while the Indian 7 Ocean ,,,,'as covered by dry anomalies. In March-April 1999 (Fig 6c, d) the La Nifia pattern remained strong in terms of SST and precipitation, but with positive precipitation anomalies located just south of the equator in the eastern Pacific. In fact, the La Nifia has persisted through boreal summer 1999 (not shown here). In summary, the dry anomalies over the Maritime Continent led the formation of the E1 Nifio. However, the maximum precipitation anomalies over the central Pacific lagged the peak in the E1 Nifio SSTs by a season. The largest tprecipitation anomalies o? , were confined to the equatorial Pacific basin and Maritime Continent from April to September 1997; were found throughout the global tropics and subtropics at the end of 1997; and extended as far poleward as Alaska by the beginning of 1998. Whether or not these precipitation anomalies were forced solely by the E1 Nifio remains unclear. 3.2 Indices The monthly evolution of observed temperature and precipitation anomalies in tl:e equatorial Pacific is captured by indices shown in Figs. 7-8. Bimonthly averages of precipitation remove the 30-60 day oscillation. Early indications of the 1997-98 E1 Nifio can be found in the decrease in rainfall over the Maritime Continent as described by the downward turn in INDO precipitation (Fig. 7b) and Amc- (Fig. 7c) in September-October 1996. The indices actually crossed the zero line in November-December. In the next bi- month positive precipitation anomalies were found in the Pacific search area (Fig. 7c). The Nino 3.4 SST index bottomed out in December 1996 and climbed to positive values in March 1997. Thus, negative precipitation anomalies were found over the Maritime Continent (ref. Fig. 3b) before the increase in rainfall and warming of the equatorial Pacific(ref.Fig.3c,d). Interestingly,thepositiveprecipitationandSSTsignalswerenbt co-locatedatthistime. Theincreaseinrainfalloccurredin thecentralPacific,just off the equator,andthewarmanomaliescoveredtheequatorialcoldtonguein theeastPacific (ref.Fig.3c,d). Infact,theanomalouslyclearconditionsandsubsequenint creasein solar radiationduringthisseasonmayhaveservedtostrengthenthedevelopingEl Nifio. By May-June positive SST anomalieshadextendedin to the centralPacific (Fig. 3e), providing the energyfor convectionin theNino 3.4region(F_Fig3.f, Fig.7b). The E1 ,?, Nifio evolved with the SST leading the precipitation signal in the central Pacific by about a month. However, the decrease in convection over the Maritime Continent occurred well before a small decrease in the water temperature surrounding the islands (Fig. 7a,b). A direct relation between these two signals is not evident. Dry conditions over the Maritime Continent (Amc-) were most extreme in October-November 1997, nearly simultaneous with the peak in E1 Nifio SSTs, but leading Ap+ (Fig. 7c). During the maturation of the E1 Nifio, in September-October 1997, the Amc+ and Ap- began to rise and fall respectively (Fig. 7d), giving the first indication of the coming La Nifia. The negative precipitation anomalies in the Pacific were found to the north of the anomalous enhanced rainfall (Fig. 4b) and positive anomalies surfaced over New Guinea in November-December, spreading throughout the Maritime Continent by March- April 1998.