Grays Harbor/Johns River snags - WA133 Additional information David Yamaguchi Reference: Yamaguchi, D.K., B.F. Atwater, D.E. Bunker, B.E. Benson, and M.S. Reid. 1997. Tree-ring dating the 1700 Cascadia earthquake. Nature, Vol. 389, pp. 922 - 923, 30 October 1997. Supplementary information for "Tree-ring dating the 1700 Cascadia earthquake" published in Nature, Scientific Correspondence David K. Yamaguchi, Dept. of Environmental Health, University of Washington, Box 354695, Seattle, Washington 98195, USA (TrRingzRUs@aol.com, fax 1-206-616-4875) Brian F. Atwater, U.S. Geological Survey at Dept. of Geological Sciences, University of Washington, Box 351310, Seattle, Washington 98195, USA (atwater@u.washington.edu, fax 1-206-553-8350) Daniel E. Bunker, Boyd E. Benson & Marion S. Reid Test of validity of Long Island series. To check for missing rings/1,2 at Long Island (46.42 N, 123.95 W), we compared its series with one previously obtained from 4-8 old-growth redcedar about 200 km to the north near Ozette/3 (about 48.2 N, 124.7 W). (The exact size of the Ozette sample is unknown to us.) The two series overlap between 993 and 1978. Although the correlation is weak (r = .17, t = 5.3, P < 0.0000001; see Statistical significance), weak correlation is common between moist-site trees. We found no reason to doubt that the Long Island chronology is complete. Correlation between Long Island and snags. The dating is supported by alignment of narrow rings, which are especially useful in tree-ring dating/1,2; statistical comparisons with alternative matches/4; strong matching of ring patterns among snags at each estuary; and best correlation nearest Long Island, at Willapa Bay (Table S1). (Similarity in tree growth declines with distance/5.) To convert ring-width measurements into indices we removed long-term trends/6 by sequential fitting of decay curves and 128-year cubic splines, with division of measurements by curve values after each fitting, followed by time-series modeling/7. Signals were enhanced by arithmetically averaging detrended series among samples from different radii of individual trees or, at Long Island only, among trees. With one exception, each reported correlation is the strongest obtained along the entire 994-year Long Island series (or along its local analog, below). (The correlation for the exception, CP-790, is the fourth highest.) Each correlation also corresponds to the position that gives the highest within-estuary correlation (Table S1). All samples are chain-sawed wedges except for cores from the trunk of CP-GF2. All samples correlated directly with Long Island are from trunks except for roots of PX-J6; combined trunk series from this snag contain only 124 rings, too few for firm dating. Test of tree-ring dating by high-precision radiocarbon dating. From one snag trunk (PX-782), we obtained a high-precision radiocarbon age--241±11 C-14 yr BP (QL-4411)--on rings pattern-matched to 1645-1655. With calibration to calendric time/8 and a probably generous laboratory error multiplier of 1.6 (ref. 9), the 95-percent confidence interval corresponding to this C-14 age is 1648-1671. Statistical significance of correlations. The probability (P) of obtaining each correlation by chance was calculated from three interim statistics. First, we converted each r value to a Student's t value that weights the correlation (r) for ring overlap (n, in years; commonly equal to sample length): t = r /{sq rt of [(1-r**2)/(n-2)]}. Next, we obtained an alpha probability for the t value using a calculator (Hewlett-Packard 48G) having a built-in, continuous t distribution. Last, we adjusted that probability for the number of reasonable alternative dates evaluated, m, from P = 1 - (1-alpha)**m (refs. 4, 10). Note that probability estimation depends heavily on selecting an appropriate value for exponent m. This value rests on inferred endpoints for dating. In the case of trunk samples, we used AD 1192 and 1720 as endpoints for the date of the outer ring in each sample. These are based on an estimated minimum overlap of 200 years for detecting significant correlation at the 993 start of the Long Island series, and on radiocarbon evidence for death by 1720 (ref. 1 of printed Scientific Correspondence). (An exception is CH-764, for which short sample length set the minimum overlap.) Root samples were more tightly constrained in time by the outer-rings dates of eroded trunk samples (Table 1 of Scientific Correspondence). For both trunks and roots, probabilities of P < 10**(-10) were obtained from the approximation, P = m alpha (ref. 4). A time-series model was fitted to the previously detrended Ozette/3 series before correlation with Long Island (m = 1 for the special case of comparing modern series such as these). Correlation among adjacent snags. In these matches, we assumed minimum ring overlaps of 150 years for detecting correlation. This minimum is based on relationships among n, r, and P. Correlation between trunk and root. We traced distinctive rings in vertical slices that extend from trunk to root in seven trees (T in Table S1). In addition, we matched long series of ring patterns between trunk and root in four trees (Tables 1 and S1). Flexible 60-year cubic splines/7, along with time-series models, were used to detrend the root series, which are less regular than those in trunks. Comparison with season of death recorded by Sitka spruce. The complete 1699 rings of the redcedar roots contrast with the incomplete outer rings of roots of Sitka spruce killed by the same submergence at Copalis River and at Willapa Bay/6. The difference probably reflects temporary survival of spruce for a few months to years after submergence. Sitka spruce tolerates more flooding than does redcedar/11. Archival of ring-width data. Ring-width measurements will be deposited in the International Tree-Ring Data Bank (http//:www.ngdc.noaa.gov/paleo.treering.html). Acknowledgements. We are grateful to the persons noted in Table 1, to Erich Eipert for loans of sanders, and to Minze Stuiver for advice on dating. We thank Alan Nelson, John Clague, David Stahle, Katherine Reed, and a shy seismologist for critical reviews. Support was provided by the U.S. Geological Survey, through the National Earthquake Hazards Reduction Program, and by the National Research Council (research associateship to D.K.Y.). 1. Stokes, M. A. & Smiley, T. L. An Introduction to Tree-Ring Dating (Univ. Chicago, 1968; reprinted by Univ. Arizona Press, 1996). 2. Schweingruber, F. H. Tree Rings, Basics and Applications of dendrochronology, Boston, Reidel (1988). 3. Jozsa, L. A., Parker, M. L., Johnson, S. G. & Bramhall, P. A. Ozette Dendrochronological Studies (ed Gleeson, P.F.) 27-56 (Laboratory of Anthropology, Wash. State Univ., Pullman, Wash., 1983). 4. Yamaguchi, D. K., Can. J. For. Res. 22, 1215-1221 (1992). 5. Cropper, J. C., & Fritts, H. C. Tree-Ring Bull. 42, 3-9 (1982). 6. Atwater, B. F., & Yamaguchi, D. K. Geology 19, 706-709 (1991). 7. Cook, E. R., & Briffa, K. R. in Methods of Dendrochronology, Applications in the Environmental Sciences, (eds Cook, E. R., & Kairiukstis, L. A.) 97-162 (Kluwer, Boston, 1990). 8. Stuiver, M. & Becker, B. Radiocarbon 35, 35-67 (1993). 9. Stuiver, M. & Pearson, G. W. Radiocarbon 28, 805-838 (1986). 10. Yamaguchi, D. K. Can. J. For. Res. 24, 427-429 (1994). 11. Brink, V. C. Ecology 35, 94-95 (1954). ------------------------------------------------------------------------ Table S1 Dating of redcedar snags ------------------------------------------------------------------------ Trunk Root --------------------------- ------------------------------- L e Outer Mean Outer e pre- r with ring g served Correlation other Height against Estu- t ring with Long snags relative bark Correlation ary h (yr Island at es- to high yr with trunk & tree (yr) AD) (r, t, log P) tuary tide (m) AD) (r, n, t) ------------------------------------------------------------------------ Copalis River CP-789 347 1677 .21 3.9 -1.52* .35 +0.1 to -0.2 >1680 T CP-790 327 1632 .16 3.0 -0.24* .38 -- -- -- CP-791 283 1680 .28 4.9 -3.39 .43 +0.2 to -0.3 1708 T CP-793 325 1664 .28 5.3 -4.11 .40 -- -- -- CP-794 271 1599 .26 4.4 -2.29 .44 -- -- -- CP-GF2 255C 1675 .21 3.4 -0.74* .31 -0.7 to -1.2 1699L .57 255 11 ------------------------------------------------------------------------ Grays Harbor CH-764 150C 1682 .38 5.0 -3.30 .27 -- -- -- JN-560 295 1674 .26 4.6 -2.68 .26 -0.2 to -0.6 1699L T JN-561 263C 1678 .44 7.9 -10.54 .33 ca -0.5 1699L .35 227 5.6 JN-566 227 1685 .30 4.7 -2.89 .28 -- -- -- ------------------------------------------------------------------------ Willapa Bay BN-915 286 1684 .32 5.7 -5.00 .42 -- -- -- PXnt778 357C 1682 .37 7.6 -10.15 .41 -0.1 to -0.3 >1670 T PX-779 289 1677 .42 7.8 -10.48 .41 -0.4 to -0.8 -- -- PX-782 239C 1681R .52 9.5 -15.53 .63 -0.2 to -0.6 1699L .42 239 7.1 T PX-783 340 1675 .30 5.9 -5.59 .43 -0.2 to -0.7 1699L .63 194 11 T PX-J6 208 1699 .23 3.3 -0.54# .29 -0.8 to -1.2 1699 -- ------------------------------------------------------------------------ Columbia River GR-578 336 1676 .21 3.8 -1.39@ .46 -- -- -- GR-580 222 1658 .36 5.8 -5.17 .50 -- -- -- GR-582 220C 1691 .30 4.7 -2.92 .59 -- -- -- GR-776 244 1675 .22 3.5 -0.82@ .50 -- -- -- GR-777 378C 1671 .27 5.4 -4.52 .41 -0.7 to -0.9 1699L .28 62 ns T ------------------------------------------------------------------------ Notes. Sample length is number of rings used in correlation analysis; a few additional, innermost rings were lost to fitting of time-series models. Statistics and letter codes follow text above or Table 1 of Scientific Correspondence. Specific tree locations are in Table S2. Trunks: C, length combined from 2-3 trunk samples on different radii. R, dating checked by radiocarbon. Roots: Series are averaged from 3-10 radii. High tide denotes present mean higher high water, as estimated from depth below present tidal wetland surface (all sites), from leveling tied to one (CP-789, CP-791, CP-GF2, JN-560, GR-777) or three (PX) tidal cycles, and from tide tables. Limiting (">") dates are from roots retaining bark solely on radii toward which tapering rings disappear. n, ring overlap; ns, not significant at P < 0.01. * Correlations with local master series derived from CP-791, -793, and -794: CP-789, r =.32 (n = 347, t = 6.3, log P = -6.92); CP-790, r = .25 (n = 303, t = 4.4, log P = -2.77); CP-GF2, n = 255, t = 4.7. # Correlations with local master from trunks of all other dated Willapa Bay snags: n = 181, t = 4.1. @ Correlations with local master from Gr-580, -582, and -777: GR-578, r = .48, n = 336, t = 10 (log P = -17.82); GR-776, r = .41, n = 244, T = 6.9 (log P = -8.28). ------------------------------------------------------------------------ Table S2 Locations of dated snags ------------------------------------------------------------------------ Estu- 7.5-minute ary Stream quadrangle Tree Latitude N Longitude W ------------------------------------------------------------------------ Copa- -- Moclips CP-789 47 07.74' 124 09.61' lis R. Moclips CP-790 47 07.77' 124 09.68' Moclips CP-791 47 07.76' 124 09.56' Moclips CP-793 47 07.76' 124 09.70' Moclips CP-794 47 07.76' 124 09.70' Moclips CP-GF2 47 07.83' 124 09.71' Grays Chehalis R. Central Park CH-764 46 56.83' 123 43.67' Har- Johns R. Hoquiam JN-560 46 52.89' 123 57.01' bor Johns R. Hoquiam JN-561 46 52.87' 123 57.01' Johns R. Hoquiam JN-566 46 52.89' 123 57.06' Wil- Bone River Bay Center BN-915 46 39' 123 53' lapa S.Fk. Palix R. Nemah PX-nt778 46 34.31' 123 53.92' Bay S.Fork Palix R. Nemah PX-779 46 34.36' 123 53.91' S.Fork Palix R. Nemah PX-782 46 34.30' 123 53.95' S.Fork Palix R. Nemah PX-783 46 34.29' 123 53.96' S.Fork Palix R. Nemah PX-J6 46 34.30' 123 53.93' Colum- Seal Slough Rosburg GR-578 46 19.36' 123 40.08' bia R. Seal Slough Rosburg GR-580 46 19.86' 123 39.79' Crooked Creek Rosburg GR-582 46 17.53' 123 39.60' Grays River Rosburg GR-776 46 19.30' 123 39.97' Seal Slough Rosburg GR-777 46 19.62' 123 39.97' ------------------------------------------------------------------------ COFECHA File_name Date Time_span Series Spline Seg Lag N_Seg Probs Prob% XCorr MnSens C/Var MnMsmt MnStDv MnACorr wa133.rwl 25May06 1378 1685 4 32 50 25 39 11 28.21 .428 .177 39.64 1.622 .964 .880 Run GRAY Title: Grays Harbor/Johns River snags PART 5: CORRELATION OF SERIES BY SEGMENTS: Grays Harbor/Johns River snags 14:43 Thu 25 May 2006 Page 5 ----------------------------------------------------------------------------------------------------------------------------------- Correlations of 50-year dated segments, lagged 25 years Flags: A = correlation under .3281 but highest as dated; B = correlation higher at other than dated position Seq Series Time_span 1400 1425 1450 1475 1500 1525 1550 1575 1600 1625 1650 1449 1474 1499 1524 1549 1574 1599 1624 1649 1674 1699 --- -------- --------- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- ---- 1 JR560 1378 1674 .16B .37 .44 .45 .45 .29B .17B .30A .73 .59 2 JR5611 1431 1678 .67 .69 .49 .68 .53 .37 .47 .66 .63 .51 3 JR5612 1413 1670 .28A .55 .48 .36 .61 .67 .31A .06B .47 .62 4 JR566 1457 1685 .44 .31A .41 .40 .13B .13B .37 .38 .31A Av segment correlation .22 .53 .51 .40 .54 .47 .25 .24 .56 .56 .41 PART 7: DESCRIPTIVE STATISTICS: Grays Harbor/Johns River snags 14:43 Thu 25 May 2006 Page 8 ----------------------------------------------------------------------------------------------------------------------------------- Corr //-------- Unfiltered --------\\ //---- Filtered -----\\ No. No. No. with Mean Max Std Auto Mean Max Std Auto AR Seq Series Interval Years Segmt Flags Master msmt msmt dev corr sens value dev corr () --- -------- --------- ----- ----- ----- ------ ----- ----- ----- ----- ----- ----- ----- ----- -- 1 JR560 1378 1674 297 10 4 .399 1.28 3.54 .649 .873 .161 2.63 .375 -.038 1 2 JR5611 1431 1678 248 10 0 .537 1.45 7.00 1.197 .897 .201 2.66 .395 -.022 1 3 JR5612 1413 1670 258 10 3 .441 1.75 7.26 1.211 .923 .175 2.74 .366 -.009 1 4 JR566 1457 1685 229 9 4 .326 2.10 4.65 .844 .821 .172 2.80 .460 -.002 1 --- -------- --------- ----- ----- ----- ------ ----- ----- ----- ----- ----- ----- ----- ----- -- Total or mean: 1032 39 11 .428 1.62 7.26 .964 .880 .177 2.80 .396 -.019