FEBRUARY 1986 PHILIP E. ARDANUY AND H. LEE KYLE 415El Nifio and Outgoing Longwave Radiation: Observations from Nimbus-7 ERB PHILIP E. ARDANUYResearch & Data Systems Corp., Lanham, MD 20706 H. LEE KYLESpace Data and Computing Division, Code 636, NASA/Goddard Space Flight Center, Greenbelt, MD 20771(Manuscript received 13 April 1985, in final form 3 September 1985) ABSTRACT Five years of broad-band earth radiation budget measurements taken by the Nimbus-7 ERB experiment havebeen archived. This period encompasses the 1982/83 E1 Nifio/Southern Oscillation event, which reached a peaknear the beginning of the fifth data year (January 1983). A 41-month outgoing longwave radiation subset ofthis data set, extending from Jane 1980 through October 1983, has been further processed to enhance the spatialresolution. Analysis of the resultant fields and the anomalies from the pre-El Nifio climatology provides the first broadband glimpse of the terrestrial outgoing longwave radiative response to the E! Nifio event throughout its lifecycle. Of particular interest are the quasi-stationary planetary-scale tropical and midlatitude patterns whichemerge as the El Nifio reaches its peak intensity. Important new implications to the vertical motion field areaddressed.1. Introductiona. Background Without question, the 1982/83 E1 Nifio event is associated with one of the largest amplitude global climate perturbations in recent history. The meteorological ramifications of this event include droughts insouthern Africa, Indonesia, and Australia; flooding inthe central equatorial Pacific, Ecuador, and Peru; intense tropical cyclones over French Polynesia and theHawaiian Islands; and unusual storm activity acrossthe southern United States from California to the Gulfof Mexico. These global phenomena are discussed indetail by Rasmusson and Wallace (1983) and Rasmusson and Hall (1983). Arkin et al. (1983) produceda "quick-look" atlas of the event including the sea surface temperature, 850 mb and 200 mb wind, and theNOAA-7 AVHRR 11.5 ~tm-band outgoing longwaveradiation fields. These illustrate the time evolution ofthe large-scale features preceding and during the event,from September 1982 through August 1983. In order to be able to predict, on a year-to-year basis,fluctuations of our global climate system, and with itthe periodic occurrences of the El Nifio/Southern Oscillation (ENSO) events, it is important to develop aphysical understanding of the ENSO phenomenon. The1982/83 ENSO occurrence was observed on manyscales, by many instruments. One of these was the setof broad-band Earth radiation budget radiometers onboard the Nimbus-7 polar orbiter. This spacecraft issun synchronous, with local-noon and local-midnight'equatorial crossing times. This experiment was operational and taking data prior to, during, and followingthe ENSO event.b. Overview of the Earth Radiation Budget Experiment Wide field-of-view radiometers capable of measuringthe shortwave and total outgoing components of theearth's radiation budget at satellite altitude have beenin operation for the past nine years in ERB experimentson board the Nimbus-6 (1975) and Nimbus-7 (1978)spacecraft. In the new follow-on program, both wideand medium field-of-view radiometers are a majorcomponent of the multi-spacecraft Earth RadiationBudget Experiment (ERBE) (Barkstrom, 1984). Whensent into sun-synchronous near-polar orbits, these radiometers record satellite-altitude irradiances whichcan be used to monitor and derive, globally and regionally, such earth radiation budget parameters as albedo, net radiation and long, wave emission on a daily,seasonal and annual basis. On board the Nimbus-7 observatory are a wide variety of remote sensing experiments including the EarthRadiation Budget (ERB) experiment. The ERB instrument package consists of three sections: a ten-channelsolar telescope, four Wide-Field-of-View (WFOV) fixedearth-flux channels, and eight Narrow-Field-of-View(NFOV) scanning earth-radiance channels. The experiment is described in detail by Jacobowitz et al.(1984). The scanner failed in June 1980 after twentyc 1986 American Meteorological Society416 MONTHLY WEATHER REVIEW VOLUM[. 114months, but the solar and the WFOV Earth-flux channels are in their seventh year of operation and both arestill recording high quality data. The solar data hasbeen discussed in numerous places--see for instanceHickey and Alton (1983). The scanner data has beenanalyzed by Jacobowitz and Tighe (1984). In this paperwe consider the long-term WFOV Earth radiationbudget data set as archived on the ERB MATRIX tapeswhich are described by Jacobowitz et al. (1984). On 24 October 1978, the Nimbus-7 (N-7) spacecraftwas launched into a sun-synchronous near-polar orbitwith local-noon ascending, and local-midnight descending nodes and a retrograde inclination to theequator of 99.3- (Jacobowitz et al., 1978). To minimizedegradation in the optical trains of the various channels(caused by the interaction of deposited out-gassingcontaminants and the solar UV), the experiment wasnot activated until 16 November 1978. At this time,full-time collection of earth radiation budget observations commenced, limited only by the constraint ofavailable spacecraft power. The major imposition wasthat of a three day on-one day off instrument dutycycle. In the Nimbus ERB instruments, a total channel(12) is used to measure both reflected-solar and emittedterrestrial radiation in a spectral band of 0.2 to 50 ~tm.At the same time, shortwave channels (13 and 14)monitor reflected solar radiation in the spectral bandsof 0.2 to 3.8 #m and 0.7 to 2.8 t~m. The calibratedoutput of the channels and their differences then yieldtotal reflected shortwave, emitted longwave, nearinfrared reflected, ultraviolet visible reflected, and totalexitant irradiances. Channels 12, 13, and 14 view the entire visible earthdisc at all times. Satellite-altitude (955 km) irradianceobservations are taken at 4-second intervals throughoutthe 104-minute orbits. It should be emphasized at thistime that this satellite-altitude data set may not be simply reduced to the "Top of the Earth's Atmosphere"(TOA) due to the nonuniform weighting of radiancesincident within the fidds-of-view (FOVs) as noted inKing and Curran (1980). Thus, the WFOV observations are actually integrals across the radius of the 29.6-Earth central angle (ECA) visible earth disk; the halfpower point radius is approximately 7- ECA. Although calibrated to high precision in the laboratory at three times prior to the launch, with no evidence of any calibration drift (Hickey and Karoli,1974), exposure to the in-flight orbital environmentproduced both trends and periodicities in the filteredchannels' (13 and 14) data set, requiring the applicationof a set of calibration adjustments (Kyle et al., 1984;Ardanuy and Rea, 1984; Maschhoff et al., 1984). Observations of the total outgoing radiation as measured by the total channel 12 appear to be more thanadequately accurate and stable. The sensitivity of thischannel varies, after the first year in orbit, at a rate ofone-tenth of a percent per year. What this implies isthat the observed instantaneous net radiation (itself afunction of the orbital plane of the satellite) and relatedproducts are sufficiently stable to render interannualclimate observations possible (Ardanuy, 1983). Whenthe two ERB experiments are taken together, an observational record spanning a decade exists. Studieshave been performed transferring the N-7 calibrationsto yield compatibility between the two data sets (Ardanuy and Jacobowitz, 1984). This decade-long dataset is, however, not yet available as the final rcprocessing of the Nimbus-6 data has not yet been completed. In addition, the WFOV calibration was cha:agedstarting with the twentieth data month (June 1980),causing a noticeable discontinuity in the outgoinglongwave radiation maps. (See Kyle et al., 1985.) Whilework is being done to remove this discontinuity, thepresent study is restricted to the last 41 months of archived data: June 1980 through October 1983.c. Objective The principal objective of this study is the analysisof the Nimbus-7 broad-band ascending-node outgoinglongwave radiation (OLR) data set with respect to thetemporal and spatial evolution of the El Nifio-inducedanomalies. The present monthly resolution of the dataset permits a clear picture of the global developmentof this interesting climate phenomena. The descendingnode data were not included in this data set due to thepotential for latitude-dependent biases in the so-called"sun blip" regions, where data are rejected due to direct(in field) solar contamination. An additional objectiveis a comparison of the Nimbus-7 broadband andNOAA-7 AVHRR 11.5 ttm-band OLR fields to identify the differences in information content betwee:a thetwo methods. Solution of a similar problem for the albedo, shortwave reflected radiation fields, and the net radiation,is nearing completion at this time and will be reportedon in the near future. It is worth keeping in mind thatchanges in longwave radiation have a direct impact onthe atmosphere on monthly time scales, while changesin shortwave radiation have the greatest impact on theocean. Due to the need for the accurate modeling ofthe directional and bidirectional reflection characteristics of the shortwave radiation within the kernel ofthe integral, the resultant deconvolution problem isless easily solved than that for the longwave radiation.2. Technical approacha. Background The roots of this data set lie in the data record takenby the Earth Radiation Budget (ERB) experiment onboard the Nimbus-7 spacecraft. The experiment hastaken over six years of high-quality observations andI:'EBRUARY 1986 PHILIP E. ARDANUY AND H. LEE KYLE 417continues to do so. At present, five years of these observations have been archived. Because of a calibrationchange, only the latter four years (extending from June1980 through October 1983) are of the continuity requisite for the objectives stated earlier. The field mostamenable to higher-level analyses at this time is thatof the broad-band longwave flux at the so-called "topof the atmosphere." This parameter, during production, is binned in to 2070 target areas each approximately 500 by 500 km, and then time-averaged in adaily and in a monthly sense. The strengths of this dataset are that 1) these are broadband (4-50 um) measurements, making the resultant fields an energy parameter directly related to atmospheric cooling rates,rather than simply an index (as an energy parameter,it is useful either as an index or in energy calculations,but not for describing surface conditions); 2) they areglobal in extent; 3) the calibration is well known; 4)there are no discontinuities in the time coverage, and;5) due to the excellent "health" of the instrument andthe spacecraft, and due to reprocessing plans for thefirst 18 months, a consistently-calibrated extension ofthis data set spanning a decade may become availablein the future. What sets the present data set apart from the archivedNimbus-7 data set is the extensive postprocessing whichit has received. Through the use of additional information, already present but "hidden" in the data, it ispossible to substantially improve both the time andspace sampling of the data set. This is possible becausethe measurements are taken by WFOV radiometers,which effectively integrate the incident terrestrialemitted radiances over a hemisphere to get an irradiance at the satellite's altitude. To obtain an estimateof longwave flux at the top of the atmosphere, the inverse-square law is applied to the observation and it isthen binned in the target area directly below the subsatellite point. This occurs every 4 seconds during thecourse of a day. Because the orbits are 26- of longitudeapart and, at the equator for example, the target areasare 41/2- wide; only 14 of the 80 target areas are sampledeach day. This can result in the aliasing of time variability into spatial perturbations in the flux estimates.Further, because of the integration process, a smoothedfield is assigned to the top of the atmosphere when, infact, more spatial variability is present. The WFOVradiometers have a half-power radius of approximately7-; thus, a low-pass filter is effectively applied to theactual fluxes. Fortunately, it is possible to correct boththe time-sampling and space-smoothing deficienciesand improve further on an already good data set. Thetechnique has already been developed (Smith andGreen, 1981) and applied (Bess et at., 1981) to Nimbus6 ERB data. It is accomplished here for the Nimbus7 ERB data set by fitting a spectrally-truncated12th-degree spherical harmonic field to the binnedobservations. By incorporating the known angular instrument response, the effective spatial resolution isdoubled along the track and observations are effectivelyobtained at each target area for each day. This processhas been applied to the 4 l-month data set.b. Simulation Though a complete treatment of the developmentprocess that resulted in the deconvolution algorithmused here is outside the scope of this paper, some reviewis pertinent. As discussed in the previous section, aresampling of the WFOV data set is desirable both interms of instrument coverage and in terms of spatialresolution. There are two questions that may be asked:What is the optimal spectral resolution, and What isthe accuracy of the results? If one is to attempt to answer these questions, then a set of "truth" fields is required. For these, the set of 23 daily target-area NFOVascending-node longwave fields for June 1979 wereused (due to the three day on-one day off duty cycleimposed by power constraints, ERB data was not takenon every day of the month). For simulation purposes,the NFOV-derived "truth" fields are assumed perfect;the goal, therefore, is to replicate as closely as possiblethe "perfect" solution. Using the NFOV "truth fields"and the orbit parameters for the Nimbus-7 spacecraft,we simulate, on an observation-by-observation, dayby-day, basis, a set of synthetic WFOV longwave measurements. After deconvolution, the resultant spectralamplitudes can then be evaluated at each target areaand a global root mean square (rms) comparison madewith the truth field. The global minimum in the surfaceformed by a contour map of rms deviation betweenthe two fields as a function of spectral resolution (number of waves versus number of nodes) then yields thedesired spectral truncation. This simulation yielded,for a rhomboidal truncation, the combination of 7waves and 12 nodes, or 195 amplitudes. Although solutions to a higher wavenumber are possible, no globalgain in accuracy is obtained. Figure 1 a illustrates the rms error between fields forthis truncation as a function of the number of daysbeing compared (from one day to one month). In thisfigure, "archived" refers to a simulation of the archiveddata set, where binning into a subset of the 2070 targetareas is applied. The term "deconvolved" applies tothe enhanced-resolution data set after the additionalprocessing, discussed in the previous section, is applied.For less than three days, the deconvolved data set isless accurate than the archived data set. This is becausethe former is evaluated over all 2070 target areas including regions between orbits which aren't well observed, whereas the latter is only evaluated for thosetarget areas that lie under the subsatellite tracks, Between two and six days of coverage (when all targetareas are filled except at the poles), the deconvolvedfield becomes relatively more accurate in comparisonto the archived results, gaining approximately 21/2 Wm-2. When the complete month is considered, a globalrms error of 8 W m-2 is obtained.418 MONTHLY WEATHER REVIEW VOLUME 1143O25~- 2O'"'""~',~. .... ARCHIVED~''' DECONVOLVED ~I I I I I I I I I I I I I I I I I 9 11 13 15 17 19 21 23NUMBER OF DAYS FiG. la. The global root-mean-square (rms) error as a function ofthe sampling interval (number of days) and at target-area resolution,when the WFOV fields are compared to the NFOV reference. Figure I b gives a comparison of the monthly sampiing for the 23 days observed (for June 1979) obtainedwith the two methods, defined here as "percent coverage'' (the number of samples per target area at eachlatitude relative to one sample per target area per day).Note that the archived fields, while w~ll-sampled in thehigh latitudes, are relatively poorly sampled in thetropics. Poleward of 81 o latitude, the extremes of theorbit track, no data are binned. In contrast, with thesampling improvement, data are obtained for all targetareas on all days. Fig. lc, for the same month, providesa comparison of the standard deviations of the WFOVmonthly-averaged archived and deconvolved data setswhen each is differenced with the NFOV truth tieldsas a function of latitude. The improvement in acc~~racyappears to be greatest in the tropical half of the Earth,as delineated by the 30- latitudes. Figure 1 d illustratesa similar monthly comparison, except for the zonalmean bias. Note that, once again, the major improvements are in the tropics. Note as well, for this month,that the largest error is associated with the ITCZ andis halved. The characteristic low-high-low bias pattern(with extrema at 20-S, 5-N, and 30-N in the archivedfield) is caused by the limited spatial resolution of theWFOV .sensors. After deconvolution, the errors aresharply reduced: the magnitudes and the wavelengthof the residual errors are cut in half. As a consequ~~nce,tropical features such as the ITCZ, the subtropical 'belts,and other nonzonally-symmetric features as we]il aremore sharply defined in the deconvolved fields. It should be recalled that, even after the resolutionimprovement implicit in the deconvolution process,some spatial smoothing remains in the OLR fields.Based on the simulation study, moderate-scale tropicalfeatures (meridional wavelength of less than 3000 km)and zonal wavenumber greater than 7 will not be fullyresolved for this spectral truncation. Residual -;rrorsresulting in the underestimation of extrema by anamount on the order of 5 W m-2 are indicated for thezonal averages (see Fig. 1 d). In addition to addressing the spatial resolution ofthe measurements, it is also necessary to have a:a understanding of the calibration stability of the instru100re'~ 20~r'Jl I I-80 -60 -40DECONVOLVEDI-20 o LATITUDE II ~ I20 40 60 80 F~G. lb. The ratio (percent) of the number of target areas actually sampled for eachday in the month relative to a reference of each target area sampled for each day, as afunction of latitude.FEBRUARY 1986 PHILIP E. ARDANUY AND H. LEE KYLE 41925]2~1510,., ,,, ,,,.. .^.,,, I I I I I I I I I-80 -60 -40 -20 0 20 40 60 80 ~TITUDE~o. lc. ~e s~n~d de~afion of ~e ~OV field ~lafive m ~e ~OV mferen~ m a fun~on of laftude.ment. This is addressed in part by Kyle et al. (1984).Due to the ability to effect solar calibrations by pitchingthe spacecraft, calibration stability of the measurementtime series can be achieved. While monthly interannuaichanges of the order of 2 W m-2 in some areas maynot be significant, there is no question that the magnitudes of the order observed during the ENSO eventare indeed real.3. Resultsa. Monthly averages Figs. 2a-d illustrate the local-noon OLR analysesfor June 1980, 1981, 1982, and 1983, respectively.Though the relatively short observing period does notdefine a climatology for the areas, it is interesting tonote the types of interannual variability present in different regions of the world. For example: 1) No interannual variability is noted near the Gulfof Guinea (0-N, 0-W). For the four years illustrated,the magnitude of the OLR minimum varies by only 2W m-2; e.g., 226, 226, 227, and 225. 2) In the three years preceding the bulk of the ElNifio of 1982/83, the South Pacific subtropical OLRmaximum (10-S, 130-W) departs by only 2 W m-2from a mean of 299. In the fourth year it departs by13 W m-2 from the pre-El Nifio mean. 20 15 10 ,,/-15~RCHIVED DECONVOLVED-80 -60 -40 -20 0 20 40 60 LATITUDEFIG. ld. The mean bias of the WFOV field relative to the NFOV reference as a function of latitude.8O420 MONTHLY WEATHER REVIEWa NIMBUS-7 ERB LONG, NAVE FLUX FOR JUNE 1980604O30200 20 40 60 80 100 120 140 160 -180 -160 -140 -120 -100 -80 -60 -40 -20 LONGITUDEVOLUME 114b NIMBUS-7 .ERB LONGWAVE FLUX FOR JUNE 19810 20 40 60 80 100 120 140 160 -180 -160 .140 -100 -80 -60 -40 -20 0 LONGITUDEFIGS. 2a-d. The local-noon analyses of outgoing longwave radiation for (a) June 1980, (b) 1981, (c) 1982, and (d) 1983, alter deconvolution. . 3) The Indian monsoon in June 1980 is "stronger"(lower OLR) than in the next two years. Here strongereither can be taken to mean either more intense (coldercloud tops or greater cloud cover) or simply more persistent cloud cover. In 1983, following the peak of E1Nifio, the monsoon is weaker (higher OLR) and later(implied by a position further to the south) than theprevious years. The pattern is typical of the May climatology (one month earlier), and July (not shown) issimilar to the June climatology, also suggesting a latemonsoon.Figs. 2e-g illustrate the local-noon OLR analyses forDecember 1980, 1981, and 1982. For these months,again, the following interesting interannual variabilitycharacteristics are present: 1) major departures in 1982 from the previous twoyears, due largely to the existence of the peak in E1Nifio for this month, notably over Australia, SouthAmerica, Indonesia, India, and the equatorial-e~tsternPacific; 2) lack of variability in certain regions, notablysouthern South America and central Africa; 3) changes in synoptic/planetary-scale wave 'phaseover the United States, with a pronounced OLR troughover the Gulf of Mexico in 1982.b. Anomalies during El Nigo During the 1982/83 E1 Nifio event, significant perturbations in a diverse set of geophysical fields occurred.Of special interest to this study are those planetaryscale fields that act to modify the outgoing longwaveradiation (OLR) field at the top of the atmosphere.The most important is the effective "cloudiness": specifically, perturbations from the climatological meanI~BRUARY 1986PHILIP E. ARDANUY AND H. LEE KYLE NIMBUS-7 ERB LONGWAVE FLUX FOR JUNE 19824216O504O302OlO-3040.6Od NIMBUS-7 ERB LONGWAVE FLUX FOR JUNE 1983 20 40 60 80 100 120 140 160 -180 -160 -140 -120 -100 '60 '60 -40 '20 LONGITUOEFIG. 2. (Continued)of cloud cover, height, thickness, water content, drop/crystal size distribution and emissivity. Also importantare changes in surface temperature and atmosphericwater vapor content and, to a lesser extent, atmospherictemperature. Changes in one or more of these parameters, regionally or globally, will cause correspondinganomalies in the broadband OLR at the top of theatmosphere. It is recognized that the base periods, or mean fields,used in this study for the production of the anomalyfields are hardly stable means. As such, they do nottruly represent a long-term climatology, but containadditional anomalies themselves (Trenberth, 1984;Short and Cahalan, 1983). Indeed, one source of potential noise in the anomalies is the well-known 4050 day oscillation (Madden and Julian, 1971; Weickmann, 1983). Nevertheless, the strength of the 1982/1983 ENSO event signal in the broadband OLR fieldsis such that the examination of temporal and spatialevolution is possible. To facilitate the examination of the time evolutionof the El Nifio event from the perspective of the set oftop of the atmosphere OLR fluxes, monthly-averagedtime-anomaly fields have been generated. These aredefined in terms of departures (W m-2) from the baseperiod for that month. The term "base period" is usedhere to indicate the two years between June 1980 andMay 1982. Thus, a two-year mean pre-E1 Nifio seasonal cycle is removed in the creation of the anomalymaps. Selected examples for the period encompassingJuly 1982, through September 1983, are illustrated inFigs. 3a-g. Analysis of these fields indicates that theOLR anomaly response to the E1 Nifio event of 1982/83 can be characterized as having four modes of behaviour: onset, intensification, expansion, and decay. 1) ONSET OF THE EL NIRO The onset phase of the OLR response exists betweenJuly and September 1982. The phenomenological422MONTHLY WEATHER REVIEWVOLUME 1 14 NIMBUS-7 ERB LONGWAVE FLUX FOR DECEMBER 1980 3O f NIMBUS7' ERB LON~A~ FLUX FOR DECEMBER 1981 ) 20 40 ~ 80 1~ 120 1~ 160 -1~ -1~ -1~ -120 -1~ -~ -~ ~ -20 ' g NIMBUS7 ERB LON~A~ FLUX FOR DECEMBER 1982~ o - ~ ~GS. 2e-g The I~-nooh an~y~ of outgoing lon~ave m~afion for (e) ~m~r 1980, (~ 1981, and (g) 1982, a~er d~onvolu~on.FEBRU^P.Y 1986 PHILIP E. ARDANUY AND H. LEE KYLE 423characteristics for these months are those of weakanomalies with little spatial extent. Indeed, were it notfor the ensuing intensification of these anomalies intomajor perturbations, one would equate them with thoseother anomalies "typically" noted in non-El Nifioyears. They are, however, stationary and do intensify. In July (Fig. 3a), the incipient El Nifio is only apparent as a weak negative (-21 W m-2) OLR anomalyat 175-W on the equator and a corresponding weakpositive anomaly (27 W m-2) over Borneo/Celebes. By August, the anomalies have remained stationaryand intensified slightly, with the OLR deficiency overthe central Pacific Ocean increasing to -33 W m-2. By September (Fig. 3b), the positive anomaly overIndonesia has extended longitudinally to encompassSumatra and New Guinea, and intensified to 30 W m-2.Over the Central Pacific Ocean the OLR deficit hasreached -36 W m-2 and is already extending alongthe equator towards South America. It is now centeredslightly south of the equator at 5-S, 175-W. 2) INTENSIFICATION OF THE EL NIglO Intensification phase of the OLR response is definedbetween October 1982 and January 1983. During thisperiod the characteristics of the E1 Nifio response arerapid intensification and expansion. In October, the Indonesian OLR excess has temporarily split into two centers, each of which is moreintense than in the preceding month. A large negativeanomaly of -50 W m-2 is present for the first timeover the Arabian Sea. Over the equatorial central Pacific Ocean, the OLR deficit has now strengthened substantially to -60 W m-2, almost doubling in intensityin one month. An eastern extension to 135-W is evident. An important phenomenon now appearing is thedoublet of positive anomalies evident to the north andsouth of the Pacific equatorial rainfall center. This pairof anomalies have important dynamic implications;they are suggestive of enhanced subsidence caused bya stronger Hadley-type circulation and will be discussedin detail in a later section. In November, (Fig. 3c) the two Indonesian OLRanomalies have remerged over Sumatra/Borneo. A ringof negative centers surrounds the maximum, whichclearly dominates the region. The primary anomalyover the equatorial Pacific Ocean has intensified furtherto -78 W m-2 and has moved eastward 15- to 160-Wrelative to the previous month. It, like the Indonesiananomaly, is surrounded by OLR centers of oppositesign. To its north and south are positive anomaly centers; the northern-most center at 15-N, 135-W morethan doubled in magnitude in one month to matchthe intensity of its southern partner at 25-S, 165-W.Strong quasi-stationary features are now becoming evident. These will persist throughout this intensificationphase. Two midlatitude centers of special interest arethose anomalies now appearing over the southernUnited States (-29 W m-2) and over central Europe(23 W m-2). In December, an eastward movement of the two primary anomalies occurs. The Indonesian positiveanomaly becomes centered over Borneo, while thetropical Pacific OLR deficit continues to intensify,reaching -83 W m-2, and moves 20- further east to anew position at 140-W. The accompanying positiveregions to the north and south, however, remain stationary, but continue to intensify. The center notedpreviously over the southern United States, centralEurope and the Arabian Sea/Indian Ocean persist. Interestingly, the tropical high-low-high pattern at 140-Wis closely mirrored by a strikingly similar low-high-lowpattern at 120-E. In January (Fig. 3d), the peak amplitudes of the OLRanomalies are generally reached. The negative radiationcenter in the equatorial Pacific has reached -88 Wm-2. To its north and south, the accompanying positiveanomalies average now half its magnitude. An interesting large-amplitude pattern exists along the equator.The three areas that are normally quite active, convectively, at this time of the year are Indonesia, theAmazon river basin and the Congo river basin. Theynow show positive OLR anomalies indicative of reduced convection. These are replaced, instead, withthe negative anomalies over the Arabian Sea, the IndianOcean, and the central equatorial Pacific ocean. Thecenter over Europe has intensified, while the centerover the United States has now moved into the Gulfof Mexico. In this Figure (3d), the true global natureof the El Nifio event is evident.3) EXPANSION OF EL NIlqO Between the months of February and May 1983, thenegative OLR anomaly over the equatorial PacificOcean expands eastward to the South American coast.At the same time in the western Pacific Ocean, theanomaly patterns become less stationary. In February, the OLR minimum over the equatorialPacific Ocean continues to drift eastward and weakensslightly to -77 W m-2. Some oscillation in the magnitudes of the positive anomaly centers to the northand south is evident in this and the next several months.The northernmost intensifies to 47 W m-: from 32 Wm-2, while the southernmost weakens to 28 W m-2.The positive OLR anomaly over Indonesia weakensslightly as well to 62 W m-2. Two accompanying features are significant with respect to the Australiandrought in progress at this time: the weakening of thesouthern tongue of the OLR maximum over Australiaand the formation of a strong (-45 W m-2) negativeanomaly to the west (20-S, 90-E) of the subcontinent.The positive anomaly over Brazil has maintained itsintensity while the center over Africa has weakened.The negative anomaly that was centered over the Gulf424 MONTHLY WEATHER REVIEW VOLUME 114NIMBUS -7 ERB LONGWAVE ANOMALIES FOR JULY 1982 NIMBUS- 7 ERB LONGWAVE ANOMALIES FOR SEPTEMBER 1982 b30 -~ o .,,,o 620 40 60 80 100 120 140 160 180 -160 ~14~ -120 -100 -80 -60 -40 -20 LONGITUDEc NIMBUS-7 ERB LONGWAVE ANOMALIES FOR NOVEMBER 1982 o L ~o; '600 20 40 60 160 -180 -160 -1~ -120 ' -1~ -80 -60 -40 -20 O F~G. 3. Monthly-averaged anom~ies of the outgoing lon~ave radiation relative to ~e "preE1 Nifio" mean for (a) July 1982, (b) ~ptem~r 1982, (c) Novem~r 1982, (d) Januau 1983, (e)April 1983, (f) June 1983, and (g) Septem~r 1983.FEBRUARY 1986 PHILIP E. ARDANUY AND H. LEE KYLE 425NIMBUS-7 ERB LONGWAVE ANOMALIES FOR JANUARY 1983LONGITUOEr NIMBUS-7 ERB LONGWAVE ANOMALIES FOR APRIL 198340 0 20 40 60 80 1 oo 120 140 160 -180 - 160 - 140 -120 - 1 oo -80 -60 -40 -20 0 LONGITUDEf NIMBUS-7 ERB LONGWAVE ANOMALIES FOR JUNE 19830 20 40 60 80 1~ 120 140 160 -180 -160 -140 -120 -100 -80 -60 -40 -20 LONGITUDEI~G. 3. (Continued)426 MONTHLY WEATHER REVIEW VOLUME 114 NIMBUS- 7 ERB LONGWAVE ANOMALIES FOR SEPTEMBER 198360FIG. 3. (COntinued)of Mexico has now extended across the Atlantic Oceanto the African coast. In March, the primary Pacific anomaly weakenedslightly to -71 W m-e, and both moved further eastward and elongated along the equator toward the coastof South America. The accompanying OLR maximato the north and south also intensify. The large positiveanomaly that had remained stationary over Indonesiahas drifted to the east, while to the south an OLR deficitis noted to drift eastward over Australia, signifying theend of their drought. Positive anomalies are still evidentover South America and equatorial Africa. By April (Fig. 3e) the OLR deficit along the easternPacific equator now clearly extends from the datelineto the coast of South America. It is bounded to thenorth and south by elongated OLR maxima, to the eastby the maximum over South America, and to the westby the large maximum to the northeast of Indonesia/Borneo. A very strong (-55 W m-2) OLR minimumis present over Australia. By May, the primary OLR minimum in the easternPacific ocean continues its drift towards South America, reaching 115-W and intensifying slightly. The fourOLR maxima that surround it decrease in spatial extent. The positive anomaly of 38 W m-2 in the Arabiansea is iladicative of a late monsoon for which the onsetover India appears to lag one month compared to otheryears. 4) DECAY OF EL NIg~O The decay phase of the OLR response is definedbetween June and September 1983, at which time theradiative anomalies no longer show any of the coherentstructure present a year before. In June (Fig. 3f), the weakening OLR minimumapproaches the equatorial west coast of South America.The OLR maxima on each of its four sides have continued to weaken. The late monsoon is evident overthe Indian peninsula. It is interesting to note that theanomaly pair (the deficit of-54 W m-2 offthe SouthAmerican coast and the excess of 48 W m-2 now overthe Philippines) are still, after one year, the largest amplitude perturbations in the global OLR field. In the next months of July, August, and September(Fig. 3g), the Indonesian OLR maximum vanishes,while the eastern-equatorial Pacific minimum weakensand moves against the coast of South America. This,in the OLR field, signifies the end of the E1 Nifioinduced anomalies.4. Characteristics of the 1982/83 ENSO eventa. Eastward propagation The eastward propagation of the OLR anomaliesaccompanying the El Nifio along the equator are illustrated in Fig. 4. The onset of the negative ano~naliesis first evident in July 1982 at the dateline. The positiveOLR anomaly over Indonesia occurs .simultaneously,even though the winter monsoon has not yet begun.At this latitude, peak values are generally reached inJanuary 1983, after the primary anomaly has propagated some 30- eastward to 150-W. By June 1983,when the anomaly loses its definition, its center hastraveled one-quarter of the way around the wc,rld to90-W, in one year.b. Zonal averages Figure 5 is a latitude-time section of the zonallyaveraged OLR anomalies. Interestingly, up until thepeak of this ENSO event, no persistent perturbationto the zonal averages is evident. The implication isthat, at least in the OLR field, regional anomalies inthe radiation.patterns are compensated by ano]naliesof the opposite sign at other longitudes. In other words,FEBRUARY 1986 PHILIP E. ARDANUY AND H. LEE KYLE 427 EQUATORIAL OLR ANOMALY TIME SECTION (W/r~) o~,.,~o o \ o o~,....~o o o~,,..,~ o o~o, ~o , , , ,o o ~o ~o ,o,~o ,~o ,,o -,~o -,~o -~o -~o -~o o ~G. 4. A fime-lon~tude ~om ~tion, ~en ~ong the ~uator, of the outgoing lon~averadiation anom~ies. No smoo~ing other than that implicit in ~e le~t-~uares resolution enhan~ment is applied.the E1 Nifio-induced OLR patterns are, up to thispoint, produced solely by longitudinal shifts in opticallyactive fields, principally clouds. By January, however, biases from the pre-E1 Nifioclimatology become apparent. Interestingly, even thesetend to compensate when averaged over latitude, indicating once again that there is primarily a displacement, rather than an absolute change, in the OLR field.c. Zonally symmetric planetary scale It is possible to ascribe some physical meaning toseveral of the spherical harmonics. In particular, forwavenumber zero, the harmonics with zero, one, andtwo zero crossings carry information about the globalmean, the Northern Hemisphere to Southern Hemisphere gradient, and the tropical-to-polar gradient, re ZONALLY-AVERAGED LONGWAVE ANOMALIES (W/m2)60N i / .____ /-7/ ,,.- L --,, ~.~ ~-; _~ -. ' - ~, ,-5 --?---~ ./ \6os I -1 '~ I ~ I ~ I ~ I)~ I ~ I JUN AUG OCT DEC FEB APR JUN AUG OCT 1982 1983 FIG. 5. A latitude-time cross section of the z~nally averaged longwave radiation anomalies.428 MONTHLY WEATHER REVIEW VOLUME i!4spectively. It is instructive to examine the time seriesof these three parameters to determine whether thereis any perturbation which one might associate with theENSO event. Figures 6, 7, and 8 illustrate the amplitudes of thethree chosen harmonics. In each of these, part A contains the two-year pre-E1 Nifio climatology (year 1= June 1980 through May 1981; year 2 = June 1981through May '1982) replotted as background over theENSO period along with the 1982/83 signal. In addition, part B contains the anomaly component beforeand after the application of a weak (1/4, 1/2, 1/4) low-passfilter to remove any monthly "spurious" componentin the result. The vertical bars span the range of thedeviations between the two years composing themonthly climatology and give some estimate of a"normal" range of variation in the absence 0fan ENSOevent. It is difficult to make any quantitative statementon statistical significance here, solely on the basis ofarguments derived from this relatively short data record. At a minimum, however, one can state that, whenthe vertical bars do not enclose zero, the anomaliesduring the ENSO event are outside the "normal" climatological range. It is immediately obvious that each of the three timeseries is perturbed in the vicinity of the peak of the E1Nifio. Figure 6 illustrates the global mean. For thisparameter, two observations can be made: the first is GLOBAL MEAN i---,, /,,,,~ 228 ,' ',,,,, \ /,.,.--' ~ 224 -4 JUN AUG OCT DEC FEB APR JUN AUG OCT 1982 1983 FIO. 6a. The time series of the globally averaged outgoing longwaveradiation during the period of June 1982 through October 1983(dashed line) relative to that of the preceding two years (June 1980May 1982), which are used here as the reference period (solid line).This is the spherical-harmonic amplitude for the zero wave/zero nodecomponent. FIG. 6b. The anomaly of the global average relative to the pre-ElNifio reference period before (solid line) and after (dashed line) filtering. The "error" bars show, on a month-by-month basis, the difference between the pre-E! Nifio years and thus give a measure ofthe variability inherent in the reference period. NORTHERN-TO-SOUTHERN HEMISPHERE GPAOIi".NT 10 0 . b JUN AUG OCT DEC FEB APR .JUN AUG OCT 1982 1983FIG. 7a. AS in Fig, 6a for the Northern Hemisphere to Southern, Hemisphere gradient; the zero wave/one node component.FIG. 7b. As in Fig. 6b for the Northern Hemisphere to Soulhern Hemisphere gradient; the zero wave/one node component.that the signal for almost the entire anomaly period isnegative, indicating that the OLR global mean is, suppressed by approximately 1% in this period; the secondis that, at the peak of the El Nifio, an enhancement ofthe OLR global mean occurs, indicating that the ]Earthis losing more thermal radiation to space. In Fig. '7, theanomaly of the Southern Hemisphere to NorthernHemisphere gradient changes sign at the peak of theENSO event. The anomaly magnitudes of 1 W m-~are comparable to those of the equator-to-pole: gradients. In Fig. 8, the equator-to-pole gradient is slightly-26-28-30 2"1EQUATOR-TO-POLE GRADIENT,-x.~-- ,, \ ~ ,,' ,,:,,,...~-2JUN AUG OCT DEC FEB APR JUN AUG 1982 1983 FIG. 8a. As in Fig. 7a for the equator-to-pole gradient:, the zero wave/two node component. FIG. 8b. As in Fig. 6b for the equator-to-pole gradient; the zero wave/two node component.... :./,,-i %~'x --~ ~' i t OCTFEBRUARY 1986 PHILIP E. ARDANUY AND H. LEE KYLE 429weaker than climatology (approximately 1-2%) in thefive months preceding the peak of the event. It thenbecomes enhanced (by approximately 4-6%) for an extended period following the peak of the event. The suggestion is that either the poles emit less thermal radiation to space or that the tropics become more efficientemitters of OLR. Physically, due to the presence of thestrong ENSO event, one expects the latter to be thecase. It is difficult to isolate effects which might presumably be due to the ENSO event from those possiblycaused by the E1 Ch!chtn eruption series of 29 March4 April 1982. The enhanced stratospheric aerosolamounts are believed to reduce the terrestrial OLR field(Chou et at., 1984; Harshvardhan, 1979). For an aerosoloptical depth increase of 0.1, the OLR could be expected to decrease by 0.5 W m-: at the poles and I to1.5 W m-: in the tropics. This could account for thenegative anomalies in Fig. 6 (but not the positive). Itis interesting to note that the last six months of thesecond year of climatology also have a slightly lowerglobal mean than those of the first year. Similarly, thesudden poleward transport of the aerosol cloud in December 1982 may, at least partially, account for thechange in sign noted in Fig. 8 (Ardanuy and Kyle,1985). It is interesting that, locally, the rise in the sea surfacetemperature (SST) normally observed during an ENSOevent (Rasmusson and Wallace, 1983) causesadecreasein the outgoing longwave radiation at the top of theatmosphere (TOA); in other words, the derivative ofthe TOA outgoing longwave radiation with respect toSST is negative. Globally, however, the brief positiveand near-zero OLR anomaly, occurring at the ENSOpeak superimposed on a background of persistentnegative anomalies possibly related to El Chichtn,may indicate a direct manifestation of the large SSTanomaly.d. Regional changes Figure 9a, b illustrate the annually averaged OLRfields for the two "climatology" years. The similaritiesbetween the two are obvious and striking (note in particular the monsoon regions and the Pacific Ocean, the240 W m-2 contour2 Differences between the two yearsare generally less than 5 W m-2. In sharp contrast,Figure 9c shows the annually averaged OLR field during the 1982/83 ENSO event. The absence of the monsoon "low" over Indonesia, and the intensification ofthe ITCZ and the two subtropical high belts in thePacific Ocean are the more obvious changes. These arehighlighted in Fig. 9d, which provides the annualanomaly of the ENSO event from climatology. Thefollowing can be noted: 1) The convectively induced OLR deficit over theeastern equatorial Pacific Ocean extends from thedateline to the coast of Equador and Northern Peru.It is completely surrounded by a ring of positive OLRanomalies. 2) The positive OLR anomaly over Indonesia andthe Philippines, caused largely by an absence of monsoon-induced cloudiness, extends from the northerncoast of Australia to Southeast Asia. This anomaly issurrounded by a ring of negative OLR anomalies whichextend through central Australia, the Indian Ocean,the Arabian Sea, the Himalayas, China, and the PacificOcean. 3) A train of anomalies extends from New Zealand,across the equator, to the southwest United States. Theteleconnection pattern is centered about the equatorialconvective anomaly, which appears to provide theforcing mechanism. The amplitudes of the extremadecrease away from the equator. A possible mechanismfor the two positive anomalies immediately to the northand south of the OLR deficit is increased subsidencedue to a thermally direct circulation caused by the enhanced convection along the equator. The couplingbetween the equatorial anomaly and that over theUnited States is particularly evident in Fig. 9d.e. Time evolution Figure 10 illustrates the monthly averaged peakmaxima associated with the primary ENSO low OLR(convective) anomaly (C), as well as the centers of thepositive anomalies immediately to its north (N) andsouth (S). Whereas the former can be traced from July1982 through the end of the data set in October 1983,the latter two anomalies are only evident for a tenmonth period centered about the ENSO peak in January 1983. They typically average half the magnitudeof the OLR deficit and are highly negatively correlated.This is consistent with the interpretation that these indicate centers of anomalous subsidence, radiatively evident through the suppression of cloudiness, a dryingof the atmospheric column, and a thinning of the planetary boundary layer. The anomaly tracks of the anomaly centers are plotted in Fig. 11 along with their amplitudes. The eastwardpropagation of the OLR deficit is evident during itsintensification (between October 1982 at 175-W andDecember 1982 at 140-W) and prior to its decay (between February 1983 at 140-W and June 1983 at95-W). The two OLR maxima are quasi-stationary(centered at 20-S, 160-W and 20-N, 130-W) untilApril and May 1983, when they propagate rapidlyeastward, apparently in response to the motion of theconvective center.5. Comparison to AVHRR It is instructive to compare the deconvolved broadband ERB data set to the three-month average OLRanomaly map, centered about the peak of the E1 Nifioevent in January 1983, as derived from the 11.5band AVHRR measurements on board the NOAA-7430 MONTHLY WEATHER REVIEW VOLUME 1146O50403020~o 0-~0-10-30-40-50-6O NIMBUS-7 ERB LONGWAVE AVERAGESa FOR JUNE 1981 THROUGH MAY 1982~ ~ C~~~~ ~~-: ~~ :~ ,~~::~ ~~- ~'~/ A ~ ~- 0 ~._ .~ . ~.(o. ~ . , .. ... ..-~ ~:.~~ ~~ I I I I I I I I I I I I I I I I I I I I I I I I I I I I I ~ ~ t ~ ~ ~ 20 40 ~ 80 1~ 120 140 160 -180 -160 -140 -120 -1~ -80 '60 '40 -20 LONGI~DE NIMBUS-7 ERB LONGWAVE AVERAGESb FOR JUNE 1980 THROUGH MAY 1981 NIMBUS-7 ER8 LONGWAVE AVERAGES FOR c JUNE 1982 THROUGH MAY 19836040 I~GS. 9a-c. The annually-averaged outgoing longwave radiation field for the (a) first pre-El Nifio year, (b) the second pre-E1 Nifio year, and (c) the E1 Nifio year.FEBRUARY 1986 PHILIP E. ARDANUY AND H. LEE KYLE 431 NIMBUS-7 ERB LONGWAVE ANOMALIES FOR d JUNE 1982 THROUGH MAY 1983FIG. 9d. The annu~ anom~y of the annu~ly avemg~ outgoing longwave m~ation field dudng the El Ni~o ye~ relative to ~e two pre~ous 7era.(Arkin et al., 1983). While both data sets are in excellentagreement on the magnitude of the equatorial-PacificOLR minimum (between 80 and 90 W m-2), they dis100QUASI-STATIONARY PACIFICOCEAN ANOMALY AMPLITUDES 6(~ S~ 20 S N<z -20 -60 -80 -~oo , I , I ~ I ~ I ~ I ~ I ~ I JULY SEP NOV JAN MAR MAY JUL SEP OCT 1982 MONTH 1983 FIG. 10. The time series of the monthly averaged peak maxima ofthe ENSO central convective anomaly (labeled C), as well as thoseof the centers~of the positive anomalies to the north (labeled N) andsouth (labeled S).agree strongly in terms of the magnitudes of the positiveOLR anomalies immediately to the north and south.The three-month average for ERB (not shown) for thecenters near 20-N and 20-S, 150-W is 35 and 41 Wm-2, respectively; for the AVHRR, approximately 22W m-2 each. The resolution of the discrepancy maylie in the base oeriods, which are different in the twodata sets. This is addressed in section 3b. Alternatively,it may lie in the specific humidity field, for which awindow-channel radiometer (AVHRR) will be less responsive to changes in OLR than a broadband radiometer. Thompson and Warren (1982) find that for surfacetemperatures greater than about 0-C, the dear-skyOLR at TOA is sensitive to the vertically integratedtropospheric water vapor content. For a reference surface temperature of 30-, for example, a change froma tropospheric mean relative humidity of 10% to amean of 90% results in a corresponding OLR decreaseof 70 W m-2 at TOA. This relationship allows an examination of the anomaly regions to determine if, infact, atmospheric moisture changes have the potentialto cause an observable difference between the two experiments. Anomalies in the total atmospheric water vaporcontent for the three months were observed by thescanning multichannel microwave radiometer(SMMR) aboard the Nimbus-7 spacecraft (Prabhakaraet al., 1985). Large positive anomalies of 1-1.5 gcm-2(more water vapor than the pre-El Nifio climate) areevident along the equatorial Pacific ocean. However,negative anomalies (drier than pre-E1 Nifio climate)are evident to the north and south, at about the samemagnitude. In round numbers, the climatological winter mean vertically integrated atmospheric moisturecontent was found to be 3 gcm-2 at 20-N, 150-W;4 gcm-2 at 150-W along the equator; and 4 gcm-2 at432 MONTHLY WEATHER REVIEW VOLUME 114 QUASI-STATIONARY PACIFIC OCEAN ANOMALY TRACKS 50N ~ 12 12 28 30'~ [ S,O,N.D, ............lOS :~0 ~~~__~ _71~A~ -65 ~ ] 20S S,O,N,~, i ~1% 40S 50S 60S I i i i i -~80 -160 -140 -120 -~00 -80 -e0 -40 FI6. 11. The anomaly trac~ of ~ ~rce Pa~fic ~ean cente~throu-out ~eir life ~ycle. The positions ~ noted by the tint leRerof the month (e.g., J = July, A = August). A~o~ denote tmnslatioain mean ~sition Etwen months. The ~k am-itudcs ~ $soprovidcd ncxt to the monthly position of each anomey ~nter. The~h~d clli~s denote the re-ons Mthi~ which the positive OLRanom~i~ remen~ qu~i-SationaW for ~ven4iet mo~ths. Notethat, althou- the ~tive anom~y w~ fi~t cMdent in ~uly 0) 1982,the ~n ~sitiv~ anoB~ies we~n't pre~nt until ~ptem~r 5.20-S, 150-W in the South Pacific Convergence Zone(SPCZ) region. The relative changes in the water vaporcontent for the three regions are then -35 to -50%25 to 40%, and -25 to -40%, respectively. This wouldsuggest, based on the work of Thompson and Warren(1982) for clear-sky fluxes, TOA flux changes of 30 to45 W m-e, -20 to -35 W m-e, and 20 to 35 W m-e,respectively. On top of this must be added the effects of clouds.Over the SPCZ, a reduction in high cloud during theENSO event will act to enhance the clear-sky OLRchange estimates. Over the equator, the developmentof the convective region with substantial high cloudcover will act to mask any effects of humidity changes,as these are concentrated in the lower troposphere(Warren and Thompson, 1983). Thus, the fact that theERB and AVHRR measurements are in agreementalong the equator is not surprising as relative humiditychanges will be relatively unimportant. Over the areasto the north and south, however, relative humiditychanges alone seem capable of accounting for OLR increases of 30-40 W m-2, with cloud changes con tributing additionally in the same direction. The dis agreement between the two experiments of some 13 19 W m-2 is thus within the amount of OLR flux change apparently explainable by observed atmo spheric water vapor changes. Thus, the change in the moisture field is a reasonable explanation for the dif ference between the two data sets, but we do not con tend that it is the only possible explanation, as other factors may also be important. Dynamically, the equatorial Pacific convection willrelease large amounts of latent heat. A thermally directHadley-type overturning can result, with descendingbranches to the north and south. This subsidence hasthe potential of drying out the atmospheric columns,e.g., at 20-N and 20-S, thus increasing the amount ofOLR radiation released by the Earth's surface to space.The implications regarding the vertical motion and irrotational circulation fields are important. Determination of these parameters from the narrow-band OLRfields (e.g., Julian, 1984) can lead to an underestimationin the strength of the descending branches of the local"Hadley cell type" circulation and, perhaps, an overestimation in the east-west "Walker-type" circulation.Vertical motion and irrotational circulation fields de' rived with narrow-band measurements may thereforesignificantly underestimate the subsidence, and therefore the velocity potential anomalies. The result maybe an incorrect partitioning of the divergent part of thecirculation into Walker (east-west) rather than Hadley(north-south) modes.6. Conclusions The Nimbus-7 Earth Radiation Budget WFOV dataset is of the needed accuracy and stability to observethe OLR anomalies induced by the 1982/83 F;NSOevent. Enhancement of the instrument's resolutionthrough a deconvolution method to a spectrally truncated spherical harmonic series at the top of the atmosphere substantially improved the tropical observations. At the event's height, in January 1983, planetaryscale teleconnection patterns are evident (see Table 1).These involve the entire tropics and are noted to extendinto the midlatitudes, particularly into the westernUnited States. It is worth noting that the annual-meanOLR anomaly pattern illustrated in Fig. 9d is strikingly similar to the schematic showing the key regionsin the global teleconnection pattern arising from OLRFluctuations in the equatorial central Pacific Ocean(Fig. 10 of Lau and Chan, 1983). The former is based,on Nimbus-7 ERB OLR anomalies during the 1982/83 E1 Nifio event; the later is based on seven years ofNOAA polar orbiter OLR data (1974-81) an6 doesnot include the most recent El Nifio event. Based on the existence and strength of the positiveOLR anomalies which accompany the principal conFEBRUARY I986 PHILIP E. ARDANUY AND H. LEE KYLE 433 TABLE 1. A comparison of the peak amplitudes noted in themonthly averaged anomaly results during the 1982~83 ENSO eventas a function of scale. Note that there exists almost one order ofmagnitude decrease in amplitude from regional to zonal scales, dueto longitudinal compensations, and another order of magnitude decrease in amplitude from zonal to planetary scale, due to latitudinalcompensations.Peak anomaly Positive NegativeScale (Wm-2) (Wm-2)Regional 65 -88Zonal average 17 - 10Equator-to-pole gradient l/: 1 l/:Northern to Southern Hemisphere gradient 1/2 -2Global average 1 -2vective region to its north and south, and combinedwith supportive observations of atmospheric verticallyintegrated water vapor content, the intensification ofthe thermally direct Hadley-type overturning in thevicinity of the convective anomaly may be muchstronger than previously suspected. Acknowledgments. One of us, Philip Ardanuy, wassupported by NASA/Goddard Space Flight Center aspart of the ERB Experiment Data Set Developmentand Quality Control Effort under the Continued Nimbus Operations Project. The authors also wish to acknowledge the help of Richard Hucek and Gloria Hoy,also of Research and Data Systems Corp., in the development of the deconvolution algorithm and for theword processing. Arnold Gruber of NOAA/NESDIS,the editor, and the anonymous reviewers all madevaluable suggestions incorporated into this study. REFERENCESArdanuy, P. E., 1983: Determinability of inter-annual global and regional climate changes of the Earth radiation budget. Preprints Fifth Conf on Atmospheric Radiation, Baltimore, Amer. Meteor. Soc., 410-413. , and H. Jacobowitz, 1984: A calibration adjustment technique combining ERB parameters from different remote sensing plat forms into a long-term data set. J. Geophys. Res., 84, 5011 5019. , and H. L. Kyle, 1986: Observed perturbations of the Earth's radiation budget: A response to the El Chichon stratospheric aerosol layer? J. Climate Appl. Meteor., 25 (3, in press). , and J. Rea, 1984: Degradation asymmetries and recovery of the Nimbus-7 Earth radiation budget shortwave radiometer. J. Geophys. Res., 84, 5039-5048.Arkin, P. A., J. D. Kopman and R. W. Reynolds, 1983:1982-1983 E1 Nino/Southern Oscillation event quick look atlas. Climate Analysis Center NOAA/NMC. 80 pp.Barkstrom, B., 1984: The Earth radiation budget experiment (ERBE). Bull. Amer. Meteor. Soc., 65, 1170-1185.Bess, T. D., R. N. Green and G. L. Smith, 1981: Deconvolution of wide field-of-view radiometer measurements of Earth-emitted radiation. Part II: Analysis of first year of Nimbus-6 ERB data. J. Atmos. Sci., 38, 474--488.Chou, M. D., L. Peng and A. Arking 1984: Climate studies with a multilayer energy balance model. Part III: Climate impact of stratospheric volcanic aerosols. J. Atmos. Sci., 41, 759-767.Harshvardhan, 1979: Perturbation of the zonal radiation balance by a stratospheric aerosol layer. J. Atmos. Sci., 36, 1274-1285.Hickey, J. R., and A. R. Karoli, 1974: Radiometer calibrations for the earth radiation budget experiment. Appl. Opt., 13, 523-533.--, and B. M. Alton, 1983: Extraterrestrial irradiance results from the ERB experiment of Nimbus-7. Preprints 5th Conf on At mospheric Radiation, Baltimore, Amer. Meteor. Soc., 444--447.Jacobowitz, H., L. L. Stowe and J. R. Hickey, 1978: The earth ra diation budget (ERB) experiment. Nimbus-7 User~ Guide. NASA Goddard Space Flight Center. ' , H. V. Soule, H. L. Kyle, F. B., House and the Nimbus-7 ERB experiment team, 1984: The Earth radiation budget experiment: an overview. J. Geophys. Res., 89, 5021-5038. ., R. J. Tighe and the Nimbus-7 ERB Experiment Team, 1984: The Earth radiation budget derived from the Nimbus-7 ERB Experiment. J. Geophys. Res., Atmos., 89, 4997-5010.Julian, P. R., 1984: Objective analysis in the tropics: A proposed scheme. Mon. Wea. Rev., 112, 1752-1767.King, M. D., and R. J. Curran, 1980: The effect of a nonuniform planetary albedo on the interpretation of earth radiation budget observations. J. Atmos. Sci., 37, 1262-1278.Kyle, H. L., F. B. House, P. E. Ardanuy, H. Jacobowitz, R. H. Maschhoff and J. R. Hickey, 1984: New in-flight calibration adjustment ofthe Nimbus-6 and -7 Earth radiation budget wide field-of-view radiometers. J. Geophys. Res., 89, 5057-5076. , P. E. Ardanuy and E. J. Hurley, 1985: The status of the Nimbus 7 Earth Radiation Budget Data Set. Bull. Amer. Meteor. Soc., 1378-1388.Lau, Ka-ming, and P. H. Chan, 1983: Short-term climate variability and atmospheric teleconnections from satel/~te-observed out going longwave radiation. Part I: Simultaneous relationships. J. Atmos. Sci., 40, 2735-2750.Madden, R. A., and P. Julian, 1971: Detection of a 40-50 day oscillation in the zonal wind in the tropical Pacific. J. Atmos. Sci.,28, 702-708.Maschhoff, R., A. Jalink, J. Hickey and J. Swedberg, 1984: Nimbus Earth radiation budget sensor characterization for improved data radiation fidelity. J. Geophys. Res., 89, 5049-5056.Pmbhakara, C., D. A. Short and B. E. Volner, 1985: El Nino and atmospheric water vapor. Observations from Nimbus-7 SMMR. Submitted to J. Climate Appl. Meteor., 24, 1311-1324. Rasmusson, E. M., and S. M. Hall, 1983: The great equatorial Pacific Ocean warming event of 1982-1983. Weatherwise, 36, 166 175.--, and J. M. Wallace, 1983: Meteorological aspects of the E1Nifio/ Southern Oscillation. Science, 222, 1195-2222. Short, D. A., and R. F. Cahalan, 1983: Interannual variability and climatic noise in satellite-observed outgoing !ongwave radiation. Mon. Wea. Rev., 111, 572-577. Smith, G. L., and R. N. Green, 1981: Deconvolution of wide field of-view radiometer measurements of Earth-emitted radiation. Part I: Theory. J. Atmos. Sci., 38, 461--473. Thompson, S. L., and S. G. Warren, 1982: Parameterization of out going infrared radiation derived from detailed radiative calcu lations. J. Atmos. Sci., 39, 2667-2680. Trenberth, K. E., 1984: Some effects of finite sample size and per sistance on meteorological statistics. Part II: Potential predict ability. J. Atmos. Sci., 112, 2364-2379. Warren, S. G., and S. L. Thompson, 1983: The climatological min imum in tropical outgoing infrared radiation: Contributions of humidity and clouds. Quart. J. Roy. Meteor. Soc., 109, 169 185. Weickmann, K. A., 1983: Intraseasonal circulation and outgoing longwave radiation modes during northern winter. Mon. Wea. Rev., 111, 1838-1858.