Organic Carbon: Oxidation and Transport in the Amazon River Jefirey E. Richey; James T. Brock; Robert J. Naiman; Robert C. Wissmar, Robert F, Stallard Science, New Series, Vol. 207, No. 4437 (Mar. 21, 1980), 1348-1351. Stable URL hitp:/ links jstor-org/siisici=0036-8075% 28 19800321%293%3A207%3A4437%3C 1348%3A OCOATIG3E2.0.CO%3B2-%23 Science is currently published by American Association for the Advancement of Science. ‘Your use of the ISTOR archive indicates your acceptance of JSTOR’s Terms and Conditions of Use, available at hup:/www,jstororglabout/terms.hml. ISTOR’s Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use ofthis work. Publisher contact information may be obtained at hutp:/www jstor.org/joumals/aaas html. Each copy of any part of @ JSTOR transmission must contain the same copyright notice that appears on the sereen or printed page of such transmission. STOR is an independent not-for-profit organization dedicated to creating and preserving a digital archive of scholarly journals. For more information regarding JSTOR, please contact jstor-info@ jstor.org, upswww jstor.org/ Wed Apr 21 14:18:43 2004ried out at NBS G). The recalculation Aiffers from the previously extant equa- tions (4) at a level far too small to explai the anomalous results, In addition, a re- ‘cent experiment which measured in vac~ ‘uum the mass difference between objects of differing D (5) has made it possible to determine p directly through Archi- rmedes’ principle. No anomalous behav- jor was observed. In an effort to reconcile the various re- sults the Center for Absolute Quantities (CAPQ) undertook a reexamination of the matter. (The CAPQ is responsible for the maintenance of the kilogram mass unit and of the high-accuracy mass mea surement capabilities at NBS.) The origi- nal data that were used to develop the Science report (I) were reviewed, and a new experiment was undertaken. In an analysis of his results after the publication of (/), Pontius became aware ‘of a number of defects in the original ‘work not initially obvious, the main one being a break in the chain of certification of the pressure-measuring devices at the high-clevation stations and another being. a questionable pressure sensor in the al- titude and decompression chambers. Forewarned of these difficulties, we en- deavored to avoid whatever systematic errors might have been present in the original study by measuring the experi- mental parameters as accurately as pos- sible and by maintaining calibrations rig- corously traceable to NBS. Since the effect seemed to occur en- tirely between normal atmospheric pres- sure and 80,000 Pa, the findings should bbe confirmed if the experiments are re- peated in a well-equipped laboratory at fan elevation of about 1600 m, Such a lab- oratory exists at Sandia Corporation, Al- >buquerque, New Mexico. With the coop- ‘eration of Sandia staff, CAPQ personnel repeated the experiment with all of the objects listed above and several other stainless steel kilograms with four times the surface-to-volume ratio of the stan- dards. State-of-the-art measurement sys- tems were used wherever possible (6). ‘The comparisons conducted at NBS ‘and Sandia show mass discrepancies be- tween weights with large volume dif. ferences which, if attributable to errors in the applied buoyancy corrections, ‘would place an uncertainty on the p al- ‘gorithm of less than 0.04 percent of p. This figure, if dependent on barometric pressure, would imply an errorin the cal- culation of p of less than 0.2 percent per atmosphere of pressure change and thus within the maximum uncertainty og nally expected by Pontius (). We have not been able to establish in detail the exact point at which the large, ne om sorsnOn21 unexpected systematic error erept into the initial experiments, The difficulties with pressure measurement mentioned above, the less than ideal thermal condi tions at the high-elevation laboratories, and the apparent effect of vibration in the deep-sea decompression chamber all seem possible sources for the troduction of systematic error. Even though the initial alarm con- cceming difficulties in comparing masses of different D now seems overstated, sig- nificant problems still exist. There re- ‘mains, for instance, an uncertainty in the ‘mass unit in materials other than plati- ‘num (for example, stainless steel) which exceeds the precision of the best Kilo- ‘gram comparators and which is a direct, result of presently accepted uncer- tainties in state-of-the-art buoyancy cor- rections, At 293.15 K, 101,325 Pa (I at- mosphere), and 50 percent relative hus midity, estimated uncertainties inp based on our most recent measurements ‘correspond to uncertainties in mass in the transfer between platinum-iridium and stainless steel artifacts (volume di ference, ~ 80 cm®) of approximately 40, ‘ug. This uncertainty may be compared to the precision of the best kilogram comparators, about 1 yg. It is also true that, in order to achieve even this accu- acy, a substantial effort is necessary both’ to measure accurately the proper- ties of air in the balance enclosure and to create stable conditions within the bal- ance case during weighings. R. M, ScHooNover RS. Davis, V. E. BOWER Center for Absolute Physical Quantities, National Bureau of Standards, Washington, D.C. 20034 Reterences and Note 1. PE Poti, Science 19,379 (1979, ‘Tae sever hgh ett ae ine of he crv or eae awe, vy gen eh Pee ih 08 percent ot fino ers formulation in one Case othe She tri ih ont te oe EOE! foes, J Rev. Nat, Bur, Stand. 83, 419 z mn Mas and Many Vales (XS Waahtgtn, D.Co T0t0y Smaaonaay Mere? moet fale, Ras Es aon Be ea, Mashing WSF, Roch, B'8. Davis, VE, Bower, J. Res Nod, er ind 8,20) (3 RMU Schoonover BS. Davis RG. Dever, VBE Bower, id 88, 27 (30), 26 December 197) Organic Carbon: Oxidation and Transport in the Amazon River Abstract. Spatial and temporal patterns in the organic carbon load (<1 millime= ter) of the Amazon River indicate that oxidation was constant throughout the river at ‘any one time but was much greater at rising water than at high water, whereas transport was constant. The total effective efflux, as the sum of oxidation plus trans- port in the river, was about 10" grams of carbon per year. Estimates for other river, ‘systems suggest that global riverine carbon fluxes exceed 10" grams per year. ‘The transport and oxidation of organic ccarbon in the Amazon River reflect, over Jong distances, upstream flooding events, and the geologic and vegetative structure of the drainage basin (I). The sum of the transport and processing fluxes yields the amount of organic carbon effectively exported from Amazonia. The sum for all large rivers constitutes the riverine role in the global carbon cycle. This role is not yet defined. As far as we know, there are no published data for the trans> port of organic carbon in any major river that encompass the hydrologic year (2). Existing data on organic carbon from the ‘Amazon are either from restricted reach- es or were taken at only one stage of the hydrologic cycle. We report results from two cruises of the RV. Alpha Helix which assess for the first time spatial and temporal pat- tems in the organic carbon load of the ‘Amazon River. Using these data as a model, we then estimate global riverine ‘carbon fluxes. We test three hypotheses, 134850500 Copyright © 1980 AAAS ‘based on the work of earlier investiga- tors (3), concerning the sources, utiliza tion rate, and downstream export of or- ganic matter. They hypothesized that cchanges in the input and utilization of dissolved and particulate organic matter should be predictable on the basis of riv- er size. For very large rivers (order 9 through 12), the hypotheses are as fol- ows: () Swamp hardwood forests and flood plains decompose fine particulate matter ‘during periods of low water; this material is subsequently returned to the rivers by flooding water and surface runoff. (i) The relative rates at which organic matter is utilized tend to be constant from the headwaters to the sea in an un- perturbed river system. (ii) The seasonally pulsed nature of ‘organic inputs is damped by biological processes and retention, such that the to- tal exports of organic matter do not in- ‘crease downstream. ‘Transect 1 extended 2000 km, from SCIENCE, VOL. 207, 21 MARCH 1980,‘Manaus, Brazil, to Iquitos, Peru, in Feb- ruary to March 1977 (RJ.N. and J.T-B.) atthe beginning of the rainy season when. water levels were rising about 1m per week. Transect 2 extended 3400 km, from Iquitos, Peru, to Belem, Brazil, in May to June 1977 (J.B.R., R.C.W., and R.FS.) during peak flooding when the river inundates terrestrial habitats for several kilometers on each side of the ‘normal channel in upstream reaches to 20 to 100 km in downstream reaches. Chemical measurements included partic tlate organic carbon < I mm (POC), dis solved organic carbon (DOC), the organ- jc matter content of seston, and the res- piratory oxidation activity (4). ‘The amount of organic matter present in the fine suspended load averaged about 10 percent during both transects (Table 1) G). The concentration of POC during the rising-water period was 15 t0 20 g mr upriver and decreased to 8.2 2 m downriver at Manaus, whereas at high water the POC concentration reached an upriver maximum of 3.7 g ‘my and exhibited downstream values of Ito 2 gm. During both rising and high water, the DOC concentration was rela- tively uniform throughout the river, av= raging 4.2 and 6.5 g m*, respectively. ‘The POC and DOC values appear com- parable to those of other investigators at single stations with regard to both wet- and dry-season values (6), Respiration ‘riputaries Tributaries 127,66 NT, 8 4 4 Uoriver Between reach 6 tutes) each 1 | tetove Manaus) | pose 2 Lipase Singu) transport] (pio Somes) [~ tanspert ) (mi Amazonas) iss 84.49 nena [ wr. 18 oxidation NT. 01 Fig. 1. Evaluation of total organic carbon (< 1 mm POC + DOC) transport and oxidation in 180 reaches ofthe Amazon River, from Iquitos, Peru, to just above the confluence ofthe Solinges ‘with the Rio Negro at Manaus (2200 kn) and from Manaus to above the delta region below the confluence with the Rio Xingu (1300 km), at sing water (frst number) and peak high water (econd number), in grams per second % 10° (NT, not taken). Inputs to each reach include upriver flows and tributaries (reach 1: Rios lea, Juta, Jura, Japura, Purus, and Jandiatuba; reach 2: Rios Negro, Madeira, Trombetas, Tapajos, Tocantins, and Xingu), Outputs include downriver transport ‘and oxidation (seston respiration). Data are from Table 1. Transport is jen by the concentration multiplied by the discharge: oxidation i given bythe respiration rate ‘multiplied by the reach volume, where the volume isthe cross-sectional area (dischargelmean Yelocty) multiplied by the reach length fof uncertainty in measurements and possible during rising water was higher by about two orders of magnitude than at high wa ter, averaging 26 and 0.2 mg m-? hour" throughout the river, respectively. Rela tive utilization rates (7) showed that dur- ing rising water 3 percent of the POC pool turned over per day versus about 0.3 percent at high water and that the uti> lization rate was constant throughout the river. Since an undetermined fraction of the organic carbon polis refractory, the ‘oxidation of the labile fraction would be more rapid. ‘What are the sources and fates of the puts and outputs do not balance in al eases because ‘exclusion of a fax ‘organic carbon in the river which might ‘account for the observed differences in concentration and utilization between rising and high water? The total carbon input to the river reach between Iquitos ‘and Manaus was 21.1 x 10" g sec™! at rising water and 11.5 x 10° g sec“? at high water, of which 60 percent was from the tributaries and 40 percent was from upriver (Fig. 1). The output from the reach was 17.1 x 10° g sec”! at rising water, of which 30 percent was oxida- tion. OF the high-water output of 11.1 x 10° g sec~', essentially all was down- Table 1. Organic matter (asthe percentage of total seston), POC, DOC, respiration, and river discharge (5) forthe rising- water eruise, February to March 1977 (transect TI) and the high-water cruise, May to June 1977 (transect 12). Thales" distance i above mouth Gm) 3464 xa 3167 Quebrada Cayarum 3022 Leticia 21 Belem 2798 ‘Sto Paulo da Olivenca 2m Uniao 2664 Santo Ant6nio do Ted 2588 Tonantina 279 uta 2423 Foz do Marmaria 2321 Jurua 2209 Piranhas 28 Tete os Montivendu 1868. naa 1670 1594 1323, Manacapura 403 Solimoes im Obidos 500 Tapaios 168 Xingu 88 Mouth 0 “Deepest points ofa iver channel joined by a ine. 21 MARCH 1980 Organic oc boc Respiration Discharge matter (%) om em) (mgm hour) (a see") 1 oP nh RP 1 RP OT R a9 6a 37427 8,000 27 167 76 196 39 36. a 186 45 a1 16 59 028 80 38 2. 52,000 70,000 70 162 360 40. m0 20 S58 102018 $9,000 80,000 mo rg 4s 024 RO US 41 13) 4039 pO 13s 34 10 Be 20 wa 0m 22 168 as 19 102 80 210 04 10 230.24 81,000 110,000 nas 39 89 2s 99 021 Ra 1s 48 0.16 96,000 130,000 100 82 60 408 90 20 0.22 170,000 230,000 103 6 10 021 "7 017 neriver transport. Of the 19.2 X 10° g sec”! input to the lower reach at high water, bout 60 percent was from upriver and 40 percent was from the ibutaries whereas 18.9 x 10° g sec-* was lost to downriver transfer. (ur results reinforce the initial hy- potheses ofthis study. (i) With respect to sources, our data do not directly demon- strate inputs from hardwood forests and floodplains. However, POC concentra- tions and oxidation were considerably higher at rising than at high water, which suggests that labile organics were being. entrained and utilized. These organics ‘could also consist of degradation prod: ucts of macrophytes G, 8) and inputs from primary production in. bordering varzea lakes (J, 9). Phytoplankton pro- duction in the main river is negligible (Wo). Erosion in the Andean highlands supplies some POC (17). The DOC in these tropical waters is chemically simi- lar to that of soil humic acids and prob- ably is of terrestrial origin (2). (ii) With respect to the utilization rate, the oxida- tion of organic carbon was relatively constant for a particular flow regime throughout all reaches of the river, but both the respiration rates and the relative utilization rates were much greater dur- ing the rising-water period than during high water, presumably a reflection of ‘ereased substrate availablity. (il) With respect to downstream export, Gibbs (11) has suggested that a downstream de- ‘crease in the concentration of total sus- pended solids during the wet season re- flected the dilution of montane waters by less concentrated tropical rivers. Our re- sults showed that the downstream de- crease at rising water was due about ‘equally to oxidation and dilution, where a dilution was the major factor at high water. These observations suggest that retention and consequent oxidation are the major factors reducing sestonic con- centrations in the river and providing @ uniform downstream export of organics. ‘This research has provided two points on the hydrograph, We hypothesize that the annual sequence of transport and oxida- tion of carbon is as shown in Fig. 2. These results have a significant bear- ing on the interpretation of calculated losses of organic carbon from Amazonia and inputs to the ocean. Ifthe calculated export of carbon (Fig. 1) is used to esti- mate annual losses, then 6 x 10° year" are exported to the South Atlantic ‘Ocean, However, the data used here are from surface samples, which may not be representative of the sediment sus- pended through the full depth of the wa ter column, Curtis et al. (/3) found that the velocity-weighted concentration of 9 particles smaller than 53 um on the high- Water cruise was about twice the surface concentration; thus, they cautioned that use of surface samples alone could yield significant errors and suggested that ve- locity-weighted, depth-integrated values should be used. If the relation hypothe- sized in Fig. 2 holds, the amount of car- bbon oxidized within the river is about 50 percent of the amount exported. Fur- thermore, these data do not include or- ganic carbon > 1 mm in size, which ‘could be a significant fraction ofthe total @). The effective carbon efflux, as the sum of depth- and velocity-weighted ex- port, oxidation, and carbon fractions > 1 ‘mm, could thus approach three times the efflux measured directly. The effective carbon output from the Amazon could exceed 10" g year Current estimates of the total organic carbon input to the oceans from rivers are about 10" g year~ Q). If the trans port and oxidation patterns observed in the Amazon hold for other major rivers, the contribution of river ecosystems 10 slobal carbon fluxes could be seriously ‘underestimated, We estimate that the to- tal effective organic carbon efflux in the ‘world's rivers might exceed 10" g year~* (14), Although controversy persists as to the magnitude of carbon release from the disruption of the terrestrial biosphere chars im? /se 108) Carton fx sme» 105 a i Fig 2. Hypothesized annual sequence of wa: ter discharge (Solid tine) and carbon transport (Gashed line) atthe mouth and carbon oxids- tion (dotted fine) throughout the Amazon Ri fet. Oxidation of organic arbon increases Fap- idly with rising flood waters atthe beginning of the wet season, as labile organics, partly fecomposed during the dry season, are en tained from the riparian habitats. Such as- ending ood states ate also periods when rv- ferine waters of Andean origin are recharged ‘with inorganic nutrients, which in tun stimu: late primary production in floodplain waters. During periods of high discharge, oxidation decreases through the dilution and depletion fof labile substances. Oxidation increases ‘gain to-a secondary maximum a6 flows sub ‘ie and organc-rich water from the primary production and biological degradation in the Aoodplains drains back into the iver. At any fone time, however, the overall oxidation of ‘organic matter is ‘constant throughout the main river because ofa balance between Up. Stream inputs, entrainment, and adequate Fesidence time for downstream utilization. ‘The net result is that export from the river i constant over the hydrologic cycle. (2 10" to 18 X 10 g year", it has ‘been assumed that all such losses are to the atmosphere (15). Carbon inputs to drainage waters ater such land-use prac- tices as clear-cutting, burning, or plow- ing can, however, be accelerated sever- alfold (16). Our estimate of riverine car- bon efflux is large enough to suggest that, since rivers are recipients of terrestrial drainage waters, they could provide a significant pathway for carbon released through human activities JerrRey E, RICHEY Fisheries Research Institute University of Washington, Seautle 98195 James T. BROCK Department of Biology, Idaho State University, Pocatello 83201 Roper J. NAIMAN Woods Hole Oceanographic Institution, Woods Hole, Massachusetts 02543 Ronerr C. WIssMaR, Fisheries Research Institute, University of Washington Ronert F. STALLARD Department of Earth and Planetary Sciences, Massachusetts Institute of Technology, Cambridge 62139 References and Notes trol Verh th 3 973) Bt Fka, Vem fer Wink Rel @, We Sch Medina Ed: (Springer Verag, New York, Fond nels: HSS Gaol, Rend Bre A ec Pea 2, L RuSedeiland G. E- Likens, in Measurements eee alee shee: Ei ertee cra Sar fecarch and” Developmen Admission, Wanhngion: DC. 1978p. 3 3 imte 6, we Mla, W, Cam See ermal es = 1 kes een oe oe a Sais cetera the be were pny sired and oaygsn com ‘centations were monitored hourly wih a Ye Sie Pcie rei SEE Seat ara Sea wiewuse Siroraee Pexciome Si eivie ateneniae ae pea icererarie a ip ironic aan ee tesiatihacta tl, ELS nonce Nature (London) 278, i6t (1979. . eee OO noe eae arora ate rs UE Sescenae ‘Percents carbon), yeling upriver t0 dow. Reacnsarty ethcue a Be cedealte eased Lay Seales SNe ae ES) eotaaae aciasa ee SCIENCE, VOL. 207London) 218, 937 (968) found dey season (BOs 92 to 35 m= downriver 7. We atsume that Boge oxiton of organic hres marl assoited with POC Ne Sve uulzaton rates are calculated ts POC spe SS agg per eae ter For ‘toy Tabi 4, CFL Willams and WJ. Junk, Biogeographic Fb Ure 9. Ek: Baber, J. Comp. Blochom. Physiol. in i. Bamond, i preparation ® 1 ASB, cin Cmachin. Ate 36 12 RYC( Beak, JH. Reaer, EM, Perdue, bi SLC 13, WF Cory BH, Meade. . Now ice, ER Shove, Narure (London) 38, Sa cin, 14, We estimate the POC lod inthe 2 largest iv. fom the daa otal suspend sols of Sl tartan See tao ec cat Eieceernat ee ee Biteacteta ett PSRs ees the ooeans. As forthe Amazon, coretons for ‘Ge ppled Enrganiecarbo tacos oie Bow corte {he ocean, this estimate woul be lcreased 19 ara er, ea Ei ater areer ge Hela, 1P5! Spi, Ek Ree 16, 1PE. Hobie and GE Likens, Lins, Ocean. dggr MT (o79) D, Le Lahand HBC. iBacerdbiogi a aks. v7, We'tane AH Devel for auatance with te ‘spin dt Caren snd BG fro, {helt v pn dtc andthe laut Nacional Pesquisas Amazonas for thei courtesy and co ‘pertion fom the Nation Science Fonds. the Suppor by grants DEB TERN and BMS.ISU0SES: Condibutn Nov S13 fom he College of Fisheries, Unversiy of Washington, fu Eontrpttn No. 9 irom the River Con tinuam poet. 20 Augst 179; revised 6 December 1979 Alaskan Seismic Gap Only Partially Filled by 28 February 1979 Earthquake Abstract, The Saint Elias, Alaska, earthquake (magnitude 7.7) of 28 February 1979 is the first major earthquake since 1900 t0 occur along the complex Pacific-North American plate boundary between Yakutat Bay and Prince William Sound. This ‘event involved complex rupture on a shallow, low-angle, north-dipping fault beneath the Chugach and Saint Elias Mountains. The plate boundary between Yakutat Bay ‘and Prince William Sound had been identified as a seismic gap, an area devoid of ‘major earthquakes during the last few decades, and was thought to be a likely site for 4a future major earthquake. Since the Saint Elias earthquake fills only the eastern ‘quarter ofthe gap, the remainder of the gap to the west isa prime area for the study of precursory and coseismic phenomena associated with large earthquakes. On 28 February 1979 an earthquake with surface wave magnitude (M,) 7.7 (1) ‘occurred beneath the Chugach and Saint Elias Mountains about 130 km northwest of Yakutat Bay, Alaska. Earlier, the re- sion between the 1958 Fairweather earthquake (M, 7.9), which broke the Fairweather fault as far north as Yakutat Bay, and the 1964 Prince William Sound earthquake (M, 8.4), which ruptured the Aleutian megathrust from about Kodiak Island to Kayak Island, had not been the site of a major earthquake since 1899 and 1900, when four events of M, 8.5, 7.8, 8.4, and 8.1 occurred within 13 months @). Although instrumental control for the epicenters of 1899 and 1900 is almost nonexistent, felt reports and observed uplifts place at least three of them be- tween Yakutat Bay and Kayak Island @, 43). Absence of recent major earthquakes identifies this zone as a seismic gap 4, 5), aregion of greater potential for major earthquakes than the adjoining regions that have ruptured more recently ‘The 28 February 1979 earthquake oc- curred on the edge of a network of $0, telemetered short-period seismic sta tions operated by the U.S. Geological SCIENCE, VOL. 207, 21 MARCH 1980 Survey in southern Alaska (6) as part of its seismic hazard assessment program. ‘The closest station is about 35 km from the epicenter of the main shock, and ten stations are within 100 km at azimuths between 130° and 320° clockwise from north. Readings of P body waves were also obtained from three new Canadian stations in the southern Yukon Territory at distances of 150 to 200 km. Epicenters determined for the main shock and 102 of the larger aftershocks ‘that occurred within the following 6 days fare shown in Fig. 1A. Only solutions with estimated epicentral standard errors less than 10 km, root-mean-square arriv- altime residuals less than 1 second, and magnitudes 2.5 or larger are shown. Based on the log number versus magni- tude distribution for the aftershocks, the data are probably complete above M, 4.0. Only 42 of the events in Fig. 14 are smaller than M, 3.5. In contrast to the high rate of after- shock activity, Fig. 1B shows the epicen- ters of the 37 events that occurred from 1 September 1978 to just before the main shock. The earthquakes shown were se lected by the same criteria as the after- 0136 8079800521-1351500500 Copyright © 1960 AAAS. shocks. These data are complete, how- ever, above about M;, 2.5 and there are only three events above M, 3.5. No fore- shock sequence is recognized. The pat- tem of seismicity is similar to that ob- served since 1974, when detailed mon- itoring began, with one exception. The cluster of earthquakes near the south- ‘eastern corner of the area outlined by dashes in Fig. 1B occurred during late September in a region without previous high activity. Whether this earthquake swarm is related to the 28 February 1979 event is not known. ‘The focal mechanism for the main shock, as determined from teleseismic and local P-wave first motions, is shown. in Fig. 1A. The steeply dipping plane (strike, N77E; dip, 79°S) is well con- strained, while the gently dipping plane (strike, N105°E: dip, 12°N)is poorly con- strained. From regional geology and tec~ tonics and the aftershock distribution (), the gently dipping plane is inferred 10 be the fault plane and the steeply dipping plane the auxiliary plane. The inferred slip is predominantly reverse dip slip in a north-northwest direction, in close ‘agreement with the direction expected from plate tectonic models @, 9). Aftershocks that occur within 1 day of a large earthquake are often used to in- dicate the extent of the rupture zone (10). Although 6 days of seismic activity are included in Fig. 1A, the distribution of aftershocks during the first day was not substantially different and gives esti- ‘mated upper bounds for the rupture di mensions of 65 by 80 km. If the initial rupture area were limited to the northern twothirds of the indicated aftershock area, which includes the two largest af- tershocks and the concentration of events near the U.S.-Canadian border, the rupture dimensions would be approx- imately 50 by 60 km. In the later ease, the southernmost events would be attrib- uted to secondary faulting triggered by the main shock. Body wave deconvolu- tion suggests thatthe rupturing was com- plex, involving at least three rupturing episodes with a combined rupture length of 50 to 70 km, General constraints on the overall rup- ture process were determined from fun- ‘damental mode Rayleigh waves recorded at the Alaskan stations Palmer (PMR; distance, 425 km; azimuth, 289") and Shemya (SMY, 2795 km, 272°) and the Gy surface wave recorded at Uweka- huna, Hawaii (UWE, 4700 km, 199°) (7). ‘The results of these analyses, which ‘were performed by standard computa- tional techniques (//), are included in ‘Table 1. These results are based on limit- ‘ed data and therefore could not be ade-