north american monsoon precipitation reconstructed from ... › wordpressua.uark.edu › ... ·...
TRANSCRIPT
-
North American monsoon precipitation reconstructed from tree-ringlatewood
Daniel Griffin,1,2 Connie A. Woodhouse,1,2 David M. Meko,1 David W. Stahle,3
Holly L. Faulstich,1,2 Carlos Carrillo,4 Ramzi Touchan,1 Christopher L. Castro,4
and Steven W. Leavitt1
Received 30 November 2012; revised 18 January 2013; accepted 21 January 2013.
[1] The North American monsoon is a major focus ofmodern and paleoclimate research, but relatively little isknown about interannual- to decadal-scale monsoonmoisture variability in the pre-instrumental era. This studydraws from a new network of subannual tree-ring latewoodwidth chronologies and presents a 470-year reconstructionof monsoon (June–August) standardized precipitation forsouthwestern North America. Comparison with anindependent reconstruction of cool-season (October–April)standardized precipitation indicates that southwesterndecadal droughts of the last five centuries were characterizednot only by cool-season precipitation deficits but also byconcurrent failure of the summer monsoon. Monsoondrought events identified in the past were more severe andpersistent than any of the instrumental era. The relationshipbetween winter and summer precipitation is weak, at best,and not time stable. Years with opposing-sign seasonalprecipitation anomalies, as noted by other studies, wereanomalously frequent during the mid to late 20th century.Citation: Griffin, D., C. A. Woodhouse, D. M. Meko, D. W.Stahle, H. L. Faulstich, C. Carrillo, R. Touchan, C. L. Castro,and S. W. Leavitt (2013), North American monsoon precipitationreconstructed from tree-ring latewood, Geophys. Res. Lett., 40,doi:10.1002/grl.50184.
1. Introduction
[2] The North American monsoon is a fundamental climatecomponent in the Southwest that provides moisture relief andmodulates ecosystem structure, wildfire, agriculturalproductivity, public health, and water resources supply anddemand [Ray et al., 2007]. Interannual monsoon moisturevariability, which is pronounced over the southwestern UnitedStates, has been the focus of numerous studies using instru-mental observations [e.g., Higgins and Shi, 2000; Seageret al., 2009; Turrent and Cavazos, 2009; Arias et al., 2012].The North American monsoon is also a research focus for
paleoclimatology [e.g., Asmerom et al., 2007; Barron et al.,2012], but relatively little is known about pre-instrumentalmonsoon variability at interannual to decadal timescales.Understanding the plausible range of monsoon variability iscritical because model projections of monsoon response toanthropogenic greenhouse gas forcing remain unclear[Castro et al., 2012; Bukovsky and Mearns, 2012; Cookand Seager, 2013]. The American Southwest has a richhistory of dendroclimatology, but the vast majority of tree-ringreconstructions from the region reflect only cool-season orannual-scale moisture variability. This is epitomized by theNorth American Drought Atlas, which, for the Southwest,principally reflects the influence of cool-season precipitationon annual water balance [St. George et al., 2010].[3] Annual tree-ring records do not reflect monsoon
moisture variability, but subannual chronologies createdfrom the summer-forming “latewood” component of treerings can contain strong, monsoon-specific precipitation sig-nal [see Griffin et al., 2011, and references therein].However, to date, relatively few latewood width (LW) widthchronologies and monsoon reconstructions have beengenerated for the southwestern United States. Targeting asmall area in southeastern Arizona, Meko and Baisan [2001]used a set of five Pseudotsuga menziesii (PSME) LW chronol-ogies and a nonlinear binary recursive classification model toestimate the probability of dry monsoons for the period1791–1992. Stahle et al. [2009] conducted LW analysis of themultimillennial PSME record from El Malpais [Grissino-Mayer, 1996] and produced a July precipitation reconstruc-tion for northwestern New Mexico, which was interpretedas a record of onset and early monsoon precipitation.Comparing the July estimate with a November–May precip-itation reconstruction, Stahle et al. [2009] found that yearswith winter precipitation extremes tended to have an oppos-ing-sign July moisture anomaly. That phenomenon, alsoidentified in instrumental studies, may be dynamically linkedto a negative land-surface feedback [e.g., Gutzler, 2000; Zhuet al., 2005; Notaro and Zarrin, 2011] and atmosphericteleconnections to Pacific sea surface temperature variability[e.g., Castro et al., 2001].[4] The present study offers several advancements on
prior research. It uses a new multispecies LW chronologynetwork to reconstruct monsoon (June–August) precipitationfor a large region in the Southwest. Following the Stahleet al. [2009] analytic, the paleomonsoon record is comparedwith an October–April precipitation reconstruction developedfrom chronologies of tree-ring earlywood width (EW).These reconstructions provide novel perspective on monsoonpaleoclimatology and reveal instability in the relationshipbetween winter and summer precipitation through time.
1Laboratory of Tree-Ring Research, University of Arizona, Tucson,Arizona, USA.
2School of Geography and Development, University of Arizona,Tucson, Arizona, USA.
3Department of Geosciences, University of Arkansas, Fayetteville,Arkansas, USA.
4Department of Atmospheric Sciences, University of Arizona, Tucson,Arizona, USA.
Corresponding author: D. Griffin, Laboratory of Tree-Ring Research,University of Arizona, 1215 E. Lowell Street, Tucson, AZ 85721, USA.([email protected])
©2013. American Geophysical Union. All Rights Reserved.0094-8276/13/10.1002/grl.50184
1
GEOPHYSICAL RESEARCH LETTERS, VOL. 40, 1–5, doi:10.1002/grl.50184, 2013
-
2. Methods and Results
[5] This study focuses on monsoon region 2 (NAM2), asdefined by the North American Monsoon ExperimentForecast Forum (Figure 1) [Gochis et al., 2009]. NAM2 isa region of spatially coherent variability in monsoonprecipitation that approximates regions identified by otherstudies. Interannual monsoon variability is strong over NAM2,which includes critical waterways, major metropolitan areas,and more than 550 km of the United States–Mexico border.NAM2 is a target for seasonal climate forecasts and has beenthe focus of dynamically downscaled regional climate modelprojections [Castro et al., 2012]. Monthly precipitation dataare from a 0.5� gridded product developed by NOAA thatextends across North America and uses a terrain-sensitivealgorithm to interpolate between available instrumental sta-tion records for the period 1895–2010. NAM2 grid point datawere averaged to produce time series of monthly precipita-tion and correlation analysis was used to assess relationshipsto the subannual tree-ring chronologies. October–April (coolseason) and June–August (monsoon) were selected as theseasons for reconstruction, and precipitation totals forthese seasons were converted to standardized precipitationindices (SPI).[6] Tree-ring data are from a new network of subannual
chronologies developed using EW and LW measurementson collections from more than 50 sites in the southwesternUnited States and Baja California (Figure 1). EW and LWindices were standardized with the protocol described byGriffin et al. [2011]. Chronologies more than 450 years longfrom NAM2 that exhibited significant (p
-
[8] The monsoon and cool-season SPI reconstructions areuncorrelated for the 1539–2008 common period (r=0.02;p=0.67) and the 1896–2008 instrumental period (r=0.06;p=0.53), although the monsoon and cool-season SPI instru-mental data exhibit weak negative correlation (r=-0.19;p=0.04). Wavelet analysis indicates that both reconstructionsare dominated by high-frequency variability (Figure S5 inSupporting Information). Low-frequency covariability be-tween the two seasonal reconstructions is visually evident(Figure 3a and b). Cross-wavelet squared coherence, analo-gous to correlation between the reconstructions as a functionof time and frequency, is variable and time unstable at periodsless than 32 years (Figure S5 in Supporting Information). Atlonger periods, the reconstructions generally exhibit in-phasecoherence. Potential sources of the low-frequency reconstruc-tion coherence include long-term changes in tree physiologyrelated to root and crown mass [Fritts, 2001], standardizationof the ring-width time series, or true low-frequency coherencebetween the seasonal climate regimes. Diagnosing the source(s) of this coherence is considered important but also beyondthe brief scope of the present study.
3. Seasonal Precipitation History for NAM2
[9] Several persistent periods of anomalous SPI stand outin the instrumental records (Figure 2). Both seasons hadnegative SPI near the turn of the 20th century and again inthe early 21st century. Precipitation was consistently aboveaverage in both seasons during the early 20th centurypluvial, as recently noted by Cook et al. [2011]. Thesepersistent variations are tracked remarkably well by thetree-ring reconstructions. Monsoon SPI was negative from1972 to 1982, the longest run of dry monsoons in theinstrumental record. This episode is not particularly wellmatched by the LW reconstruction, possibly because itcontained several notably wet May and June months thatwould have differentially benefitted summer tree growth.
Visual comparison indicates that cool-season SPI containsmore middle- to low-frequency variability than monsoonSPI and that the EW-based reconstruction tracks thesefrequency components well.[10] The reconstructions offer an expanded perspective, re-
vealing seasonal drought events more persistent and severethan those of the instrumental era (Figure 3). The period ofmost persistent summer drought was 1882–1905, when 19of 24 years had negative monsoon SPI. In the 1890s and early1900s, this monsoon failure was complemented by severe andpersistent cool-season drought. This dual-season droughtevent factors importantly into the socioenvironmental historyof the region. Arrival of the Southern Pacific Railroad in theearly 1880s ushered an expansion of the Arizona cattle popula-tion from ~40,000 to 1.5 million individuals [Sheridan, 1995].By the early 1890s, however, an estimated 50–75% of the herdstarved and perished from drought-induced rangeland failure[Sayer, 1999]. Cattle overgrazing triggered landscape-scalevegetation change that remains evident today. Environmentalconsequences of the 1890s drought were also severe innorthern Mexico, as discussed by Seager et al. [2009]. Iron-ically, this decades-long drought was immediately followedby the well-known early 20th century pluvial, which in-cluded the highest frequency of dual-season wetness in theentire reconstruction (Figure S4 in Supporting Information).[11] Other dual-season drought events are evident in the
early 1820s and the 1770s. Yet more severe was the 17thcentury “Puebloan Drought,” an event Parks et al. [2006]implicate as one factor leading to the Pueblo Revolt of1680. In northwestern New Mexico, Stahle et al. [2009]found that this drought included below-average precipitationduring both the cool season and July. For NAM2, the periodfrom 1666 to 1676 included a seven-year run of drier-than-average monsoons and six years with below-average SPI inboth seasons. According to the smoothed record, this wasthe most extreme monsoon drought episode of the last 470years. This result, coupled with the finding by Stahle et al.
Figure 3. Time-series graphs of reconstructed SPI for the cool season (a) and monsoon (b) with a 5-year cubic spline plot-ted in black. Vertical gray bars denote years with opposing-sign SPI anomalies and the black line represents a centered 30-year running count of these events (c).
GRIFFIN ET AL.: NA MONSOON PRECIPITATION RECONSTRUCTION
3
-
[2009], indicates that monsoon failure during the 17thcentury Puebloan Drought was robust and widespread acrossthe Southwest.[12] The late 16th century “megadrought” is evident in
both the monsoon and cool-season SPI reconstructions forNAM2 (Figure 3). This event, among the most severedroughts of the last millennium, drove landscape-scaleecosystem change across the Southwest [e.g., Swetnam andBetancourt, 1998]. From 1566 to 1579, 11 of 14 monsoonswere drier than average. Monsoon SPI rebounded to averageor near average in the 1580s, but cool-season droughtpersisted through that decade. From 1566 to 1587, 21 of22 years were reconstructed to include at least one seasonof below-average SPI and 10 of those years had negativeSPI in both seasons. July dryness was previously noted innorthwestern New Mexico [Stahle et al., 2009], but this isthe first time that persistent monsoon failure has beenconnected with the megadrought in the NAM2 region.Several exceptionally dry monsoons in the late 1560s revealthat the megadrought may have first manifest as a warm-season phenomenon. This idea is lent some support byspatial analysis of the megadrought through time (Figure 6in Stahle et al. [2007]), which mapped the 1560s droughtcenter over northwestern Mexico, where the monsoonprovides more than 70% of annual precipitation.[13] These reconstructions offer a novel opportunity to
assess the relationship between cool-season and monsoonprecipitation in NAM2. Years with opposing-sign SPIanomalies, plotted in Figure 3c, appear to be randomlydistributed through time. There are brief periods when theseevents appear to congregate (e.g., 1960s and 1720s) andothers when they were rare (i.e., the decadal drought andwet events described above). A 30-year running countillustrates some temporal variability and indicates that themid to late 20th century was characterized by a relativelyhigh frequency of such events. These results are congruentwith instrumental data analysis by Gutzler [2000] and byZhu et al., who found that the “linkage is strong from1965–1990 and weak otherwise” [2005]. Opposing-signyears were also relatively frequent in the mid-18th and early17th centuries. In contrast, the period centered near the turnof the 20th century contained relatively few of these events.Wavelet analysis reveals no time-stable coherence atsubdecadal to decadal timescales, indicating that theNAM2 seasonal precipitation relationship is dominated bynoise (Figure S3 in Supporting Information).
4. Summary
[14] This study provides a high-quality, precisely dated,seasonally resolved geochronology for the NAM2 region.The reconstructions represent a plausible range of interannual-to decadal-scale variability in multiseason precipitation forthe past 470 years. These precipitation estimates are suitablefor comparison with paleoproxy, historical, modern, andprojected climate data and should be useful for a more preciseinterpretation of the social and environmental history in theAmerican Southwest. Decadal drought events in this regionover the past five centuries (e.g., 1570s, 1660s, 1770s,1820s, 1890s, and early 2000s) were characterized not onlyby cool-season precipitation deficits but also by consistentfailure of the summer monsoon. Monsoon drought events inthe past were more persistent (e.g., late 19th century) and
extreme (e.g., mid-17th century) than any during the instru-mental era to date. With the exception of the ongoing early21st century drought, the instrumental era lacks the persistentdual-season drought episodes common to centuries past. TheNAM2 seasonal precipitation relationship is weak at bestand is not time stable. The mid to late 20th century experi-enced a relatively high frequency of years with opposing-signprecipitation anomalies. These results indicate that the instru-mental era, especially the later half of the 20th century, maynot be the ideal period for characterizing NAM2 monsoonprecipitation climatology, its relationship to the cool-seasonclimate regime, and potential connections to the coupledocean-atmosphere system.
[15] Acknowledgments. This research was funded by NSF P2C2award #0823090 and NOAA award #NA08OAR4310727. D. Griffin wassupported by Environmental Protection Agency STAR award #FP917185.The authors thank R. Vose and R. Heim for access to the gridded climatedata, M. Losleben and numerous students for assistance in subannual tree-ring chronology development, and two anonymous reviewers for commentsthat improved the article.
ReferencesArias, P. A., R. Fu, and K. C. Mo (2012), Decadal variation of rainfallseasonality in the North American monsoon region and its potentialcauses, J. Climate, 25, 4258–4274, doi:http://dx.doi.org/10.1175/JCLI-D-11-00140.1
Asmerom, Y., V. Polyak, S. Burns, and J. Rassmussen (2007), Solar forcing ofHolocene climate: New insights from a speleothem record, southwesternUnited States, Geology, 35, 1–4, doi:10.1130/G22865A.1
Barron, J. A., S. E. Metcalfe, and J. A. Addison (2012), Response of theNorth American monsoon to regional changes in ocean surface tempera-ture, Paleoceanography, 27, PA3206, doi:10.1029/2011PA002235
Bukovsky, M. S., and L. O. Mearns (2012), Projections of climate changefor the North American monsoon from the NARCCAP regional climatemodels, Paper presented at the 92nd American Meteorological SocietyAnnual Meeting, New Orleans, LA.
Castro, C. L., T. B. McKee, and R. A. Pielke Sr. (2001), The relationship ofthe North American Monsoon to tropical and north Pacific sea surfacetemperatures as revealed by observational analyses, J. Climate, 14,4449–4472, doi: http://dx.doi.org/10.1175/1520-0442(2001)0142.0.CO;2.
Castro, C. L., H. Chang, F. Domiguez, C. Carrillo, J Schemm, and H.H.Juang (2012), Can a regional climate model improve the ability to fore-cast the North American Monsoon?, J. Climate, 25, 8212–8237, doi:http://dx.doi.org/10.1175/JCLI-D-11-00441.1.
Cook, B. I., and R. Seager (2013), The response of the North AmericanMonsoon to increased greenhouse gas forcing, J. Geophys. Res.,doi:10.1002/jgrd.50111, in press.
Cook, B. I., R. Seager, and R. L. Miller (2011), On the causes and dynamicsof the early twentieth-century North American Pluvial, J. Climate, 24,5043–5060, doi:10.1175/2011JCLI4201.1.
Fritts, H. C. (2001), Tree Rings and Climate, 2nd ed., Blackburn Press,Caldwell, New Jersey, 567 pp.
Gochis, D., J. Schemm, W. Shi, L. Long, W. Higgins, and A. Douglas(2009), A Forum for Evaluating Forecasts of the North AmericanMonsoon, Eos Trans. AGU, 90(29), 249, doi:10.1029/2009EO290002.
Griffin, D, D. M. Meko, R. Touchan, S. W. Leavitt, and C. A. Woodhouse(2011), Latewood chronology development for summer-moisturereconstruction in the U.S. Southwest, Tree-Ring Res., 67, 87–101,doi:10.3959/2011-4.1.
Grinsted, A., J. C. Moore, and S. Jevrejeva (2004), Application of thecross wavelet transform and wavelet coherence to geophysical time series,Nonlin. Processes Geophys., 11, 561–566, doi:10.5194/npg-11-561-2004.
Grissino-Mayer, H. D. (1996), A 2129-year reconstruction of precipitationfor northwestern New Mexico, USA, in Tree Rings, Environment, andHumanity, Radiocarbon, pp. 191–204, Department of Geosciences,University of Arizona, Tucson.
Gutzler, D. S. (2000), Covariability of spring snowpack and summer rainfallacross the Southwest United States, J. Climate, 13, 4018–4027, doi:http://dx.doi.org/10.1175/1520-0442(2000)0132.0.CO;2.
Higgins, R. W., and W. Shi (2000), Dominant factors responsible forinterannual variability of the summer monsoon in the southwesternUnited States, J. Climate, 13, 759–776, doi:http://dx.doi.org/10.1175/1520-0442(2000)0132.0.CO;2.
GRIFFIN ET AL.: NA MONSOON PRECIPITATION RECONSTRUCTION
4
http://doi:http://dx.doi.org/10.1175/JCLI-D-11-00140.1http://doi:http://dx.doi.org/10.1175/JCLI-D-11-00140.1http://doi:http://dx.doi.org/10.1175/JCLI-D-11-00441.1http://doi:http://dx.doi.org/10.1175/JCLI-D-11-00441.1http://dx.doi.org/10.1175/1520-0442(2000)0132.0.CO;2http://dx.doi.org/10.1175/1520-0442(2000)0132.0.CO;2http://dx.doi.org/10.1175/1520-0442(2000)0132.0.CO;2http://dx.doi.org/10.1175/1520-0442(2000)0132.0.CO;2
-
Meko, D. M., and C. H. Baisan (2001), Pilot study of latewood-width of con-ifers as an indicator of variability of summer rainfall in the North Americanmonsoon region, Int. J. Climatol., 21, 697–708, doi:10.1002/joc.646
Notaro, M., and A. Zarrin (2011), Sensitivity of the North Americanmonsoon to antecedent Rocky Mountain snowpack, Geophys. Res. Lett.,38, L17403, doi:10.1029/2011GL048803.
Parks J. A., J. S. Dean, and J. L. Betancourt (2006), Tree rings, drought, andthe pueblo abandonment of south- central New Mexico in the 1670s, in:Environmental Change and Human Adaptation in the Ancient AmericanSouthwest, pp. 214–227, University of Utah Press, Salt Lake City, UT.
Ray, A. J., G. M. Garfin, M. Wilder, M. Vasques-Leon, M. Lenart, andA. C. Comrie (2007), Applications of monsoon research: opportunitiesto inform decision making and reduce regional vulnerability, J. Climate,20,1608–1627, doi:http://dx.doi.org/10.1175/JCLI4098.1.
Sayer, N. (1999), The cattle boom in southern Arizona: towards a criticalpolitical ecology, J. Southwest, 41, 239–271.
Seager, R., M. Ting, M. Davis, M. Cane, N. Naik, J. Nakamura, C. Li, E. Cook,and D. W. Stahle (2009), Mexican drought: an observational modeling andtree ring study of variability and climate change, Atmósfera, 22, 1–31.
Sheridan, T.E. (1995), Arizona: A History, University of Arizona Press,Tucson, Arizona, 434 pp.
Stahle, D. W., F. K. Fye, E. R. Cook, and R. D. Griffin (2007), Tree-ringreconstructed megadroughts over North America since A.D. 1300, Cli-matic Change, 83, 133–149, doi:10.1007/s10584-006-9171-x
Stahle, D. W., M. K. Cleaveland, H. Grissino-Mayer, R. D. Griffin,F. K. Fye, M. D. Therrell, D. J. Burnette, D. M. Meko, and J. Villanueva-Diaz (2009), Cool- and warm-season precipitation reconstructions overwestern New Mexico, J. Climate, 22, 3729–3750, doi:http://dx.doi.org/10.1175/2008JCLI2752.1.
Swetnam, T. W., and J. L. Betancourt (1998), Mesoscale disturbance andecological response to decadal climatic variability in the AmericanSouthwest. J. Climate, 11(89), 3128–3147, doi:http://dx.doi.org/10.1175/1520-0442(1998)0112.0.CO;2.
St. George, S., D. M. Meko, and E. R. Cook (2010), The seasonality ofprecipitation signals encoded within the North American Drought Atlas.Holocene, 20, 983–988, doi:10.1177/0959683610365937
Turrent, C., and T. Cavazos (2009), Role of the land-sea thermal contrast inthe interannual modulation of the North American Monsoon, Geophys.Res. Lett., 36, L02808, doi:10.1029/2008GL036299.
Zhu, C., D. P. Lettenmaier, and T. Cavazos (2005), Role of antecedent landsurface conditions on North American monsoon rainfall variability, J.Climate, 18, 3104–3121, doi:http://dx.doi.org/10.1175/JCLI3387.1.
GRIFFIN ET AL.: NA MONSOON PRECIPITATION RECONSTRUCTION
5
http://dx.doi.org/10.1175/JCLI4098.1http://dx.doi.org/10.1175/2008JCLI2752.1http://dx.doi.org/10.1175/2008JCLI2752.1http://dx.doi.org/10.1175/1520-0442(1998)0112.0.CO;2http://dx.doi.org/10.1175/1520-0442(1998)0112.0.CO;2http://dx.doi.org/10.1175/1520-0442(1998)0112.0.CO;2http://dx.doi.org/10.1175/1520-0442(1998)0112.0.CO;2http://dx.doi.org/10.1175/JCLI3387.1
-
1
Auxiliary Materials: Griffin et al. North American monsoon precipitation reconstructed 1
from tree-ring latewood. 2
1. Study Area 3
[1] This study is focused on North American monsoon region 2 (NAM2), one of eight regions 4
identified using EOF analysis of instrumental station data by researchers of the North American 5
Monsoon Experiment [Gochis, 2009; NAME, 2013]. NAM2 is similar in domain to monsoon 6
regions identified in other studies, including Comrie and Glenn [1998], Gutzler [2004], and Zhu 7
et al. [2005], suggesting this regional delineation is relatively robust. NAM2 extends from 30 to 8
35.25 degrees north and from 107.75 to 113.25 degrees west (Figure 1). The region is 9
physiographically diverse, ranging from the continental divide west to the low deserts beyond 10
Phoenix, and from the complex ecosystems of the Arizona-Sonora border region north to the 11
heavily forested Mogollon Rim and the southern Colorado Plateau. NAM2 contains many 12
environmentally and socially significant waterways, including the Bavispe, Gila, Magdalena, 13
Salt, Santa Cruz, San Pedro, Sonora, and Verde rivers. NAM2 population centers include the 14
Phoenix-Tucson metropolitan corridor, Flagstaff, Prescott, Nogales, and Silver City. 15
16
2. Tree-Ring Data 17
[2] Tree-ring data in this study come from a new systematic network of “earlywood” width 18
(EW) and “latewood” width (LW) chronologies developed for the southwestern U.S. and Baja 19
California, Mexico (Figure 1). Annual growth rings of many southwestern conifer trees exhibit a 20
clear anatomical transition from the earlywood produced in spring to the denser latewood 21
produced in summer. The width of these sub-annual components varies independently, and 22
several studies have demonstrated that latewood width can be used as a proxy for evaluating the 23
-
2
paleoclimatic history of the North American monsoon [see references in Griffin et al., 2011]. 24
Field sampling from 2009–2011 was used to update and augment archived collections of 25
Pseudotsuga menzieszii (PSME) and Pinus ponderosa (PIPO) tree ring samples from 52 sites 26
across the southwestern U.S. and Baja California (Figure 1). Twenty-six sites fall within the 27
NAM2 region. Field sampling explicitly targeted young to middle aged trees (100–250 years) to 28
produce age-stratified chronologies, a strategy recommended by prior research [Meko and 29
Baisan 2001; Stahle et al., 2009] to maximize monsoon precipitation signal. The new samples 30
were prepared and cross-dated using classical methods [Douglass, 1941; Stokes and Smiley, 31
1996]. For the first time on the new or archived samples, EW and LW were measured to the 32
nearest 0.001mm. Calendar-year dating was verified using COFECHA [Holmes, 1983]. 33
34
[3] Numerical tree-ring chronologies were computed with standard methods [Cook and 35
Kariukstis 1990]. Sub-annual EW and LW measurement time series were detrended using a 36
cubic smoothing spline [Cook and Peters, 1981] with a frequency response of 0.5 at a 37
wavelength of 100 years. The tree-ring index time series exhibited low-order persistence 38
theorized to be of biological origin [Fritts, 2001]. Pooled autocorrelation was modeled and 39
removed to produce so-called “residual” indices [Cook, 1985]. LW indices were “adjusted” to 40
remove the interannual EW dependence by computing the residual of LW index regressed onto 41
EW index. Isolating the variability unique to LW can enhance monsoon precipitation signal in 42
the final chronologies [Meko and Baisan, 2001]. Site-level EW and adjusted LW chronologies 43
were computed as the Tukey robust bi-weight mean index for each year [Cook, 1985]. Further 44
detail on the strategies and methods of chronology development were provided by Griffin et al. 45
[2011]. 46
-
3
47
3. Instrumental data 48
[4] Instrumental precipitation data comes from a 0.5{degree sign} gridded climate product 49
developed by Russ Vose and Richard Heim of the U.S. National Oceanic and Atmospheric 50
Administration (NOAA). Similar to the Parameter Regression on Independent Slopes Model 51
(PRISM) [Daly et al., 2008], the NOAA data were developed for North America using a terrain-52
sensitive algorithm to interpolate between all available monthly station records for the period 53
1895–2010. Additional detail is provided by Castro et al. [2012]. Monthly precipitation was 54
averaged for grid points within NAM2. Figure S1 uses box plots to illustrate the distribution of 55
monthly precipitation values for the 1896–2008 period common with the tree-ring data. The 56
region experiences a bi-modal precipitation distribution [e.g., Sheppard et al., 2002]. May is the 57
driest month and represents the climatological transition from the cool season to the summer 58
monsoon. In NAM2, the monsoon is strongest from July through September. On average, 59
precipitation during those months comprises 39% of the 335.95 mm annual total for NAM2. 60
However, NAM2 occupies a transition zone for monsoon onset and the southeastern part of the 61
region experiences onset in late June [see Liebmann et al., 2008], as is more common in the core 62
monsoon region over northwestern Mexico. On average, precipitation during the monsoon 63
months of June through September contributes 49% of the annual total for NAM2. 64
65
[5] Exploratory analysis used the Seascorr software [Meko et al., 2011] to produce Pearson 66
coefficients for NAM2 EW and LW data correlated with monthly precipitation (Figure S2) and 67
seasonal precipitation (not shown) for the period 1896–2008. Most of the EW chronologies were 68
significantly correlated (p
-
4
that correlation, May is the driest month of the year, it falls outside of the cool season, and its 70
inclusion in seasonal precipitation totals did not improve the calibration with the EW 71
chronologies. Consequently, October through April was selected as the period for cool-season 72
precipitation calibration. Most of the adjusted LW chronologies were significantly correlated 73
with July precipitation and several were also significantly correlated with June and August 74
precipitation. These results are congruent with findings by Fritts [2001], who continuously 75
observed Pinus ponderosa growth for several years in southern Arizona and found sensitivity to 76
monsoon moisture variations. The results are also in line with more recent findings that the LW 77
chronologies are most sensitive to moisture variability in the early- to mid-monsoon season with 78
some sites remaining sensitive through the month of August [Meko and Baisan, 2001; Stahle et 79
al., 2009; Griffin et al., 2011]. No LW chronologies were significantly correlated with 80
September precipitation in the current year, suggesting that radial tree growth is completed 81
before or during that month. As such June through August was selected as the period for 82
monsoon precipitation calibration. On average, June through August precipitation represents 83
80% of that from the classic monsoon months of June through September. Precipitation totals 84
were converted to Standardized Precipitation Indices (SPI) [Mckee et al., 1993] for the 7-month 85
period ending in April (cool season) and 3-month period ending in August (monsoon), with the α 86
and β parameters of the gamma distribution computed for the 1896–2008 period common to the 87
tree-ring and instrumental data. 88
89
4. Tree-ring reconstructions of seasonal SPI 90
[5] EW chronologies were used to reconstruct cool-season SPI and adjusted LW chronologies 91
were used to reconstruct monsoon SPI. Chronologies were only considered as candidate 92
-
5
predictors if they exhibited highly significant correlation (p < 0.01) with the reconstruction 93
targets over the full common period (1896–2008) and its split halves (1896–1952; 1953–2008). 94
Forward stepwise multiple linear regression was used to develop the SPI reconstructions, and 95
predictors were only allowed to step in when their coefficients were significantly different than 96
zero (p < 0.05). Exploratory regression analysis involved four categories of predictors: 1) site-97
level chronologies over 350 years long (a majority of NAM2 chronologies); 2) site-level 98
chronologies over 450 years long (i.e. those covering the 1559–1582 megadrought event [Stahle 99
et al., 2007]); 3) principal-component scores of the 350+ year-long chronologies; 4) principal-100
component scores of the 450+ year long chronologies. Regression model skill was assessed with 101
standard validation methods, including the sign test [Fritts, 2001], leave-one-out cross validation 102
[Michaelsen, 1987], and the reduction of error statistic [Cook et al., 1999]. Verification results 103
indicated that the site-level predictors and PC-score predictors produced very similar regression 104
results. The site-level chronologies produced marginally higher coeffecients of determination 105
and improved results in the sign test (frequency of cases when the instrumental and reconstructed 106
indices were of the same sign relative to the zero mean). The monsoon estimates based on 350-107
year chronologies captured more instrumental variance than those with the 450-year 108
chronologies (adjusted R2 = 0.53 and 0.45, respectively). To maximize length and simplify 109
presentation, the reconstructions using site-level chronologies over 450 years long were 110
presented and analyzed. Seven EW chronologies and four LW passed the correlation and length 111
screening criteria and were submitted as candidate predictors in forward stepwise linear 112
regression (Table S1). Information on final SPI calibration models are presented in Table S2. 113
114
[6] Both reconstructions used three predictors representing PSME and PIPO chronologies well 115
-
6
distributed across the NAM2 region (Figure 1). All predictor coeffecients were positive. The 116
leading predictors in the reconstructions were PSME EW and LW chronologies from the same 117
site (FSM) in the Santa Rita Mountains near Tucson. EW and LW chronologies from a PIPO site 118
near Flagstaff (WMP) also entered as predictors in each reconstruction. Time series and scatter 119
plots illustrate the relationship between observed and reconstructed SPI (Figure 2). All 120
reconstruction predictors, predictands, and residuals met regression assumptions of normality 121
and zero-autocorrelation (Table S3). The calibration and verification statistics indicated that 122
reconstructions have significant predictive skill, even by southwestern standards. The cool-123
season reconstruction captures more instrumental variance than does the monsoon 124
reconstruction, but this is not surprising because summer precipitation exhibits greater spatial 125
heterogeneity than winter precipitation [Mock, 1996] and cool-season precipitation dominates 126
annual water balance and tree growth in the region [St. George et al., 2010]. 127
128
5. Analyses 129
[7] Runs analysis was used to identify reconstructed events with consecutive years above (wet) 130
or below (dry) the zero mean. The cool-season reconstruction contains wet and dry events more 131
persistent than those identified in the instrumental record while the monsoon instrumental record 132
contains a wet event and a dry event more persistent than any in the reconstruction. Plotting 133
these events through time highlights periods of persistent drought and wetness in the 134
reconstructions (Figure S1). Runs analysis is sensitive to values that fall near the zero mean, such 135
that one year of slightly above average precipitation causes a break in the drought run (e.g., late 136
19th century in Fig. S1b). To highlight the most persistent SPI anomalies, five-year cubic 137
smoothing splines were fit and plotted on the long reconstructions (Figure 3). Five-year spline 138
-
7
values were ranked to identify the non-overlapping events of greatest magnitude. 139
140
[8] The reconstructions were analyzed to determine when both seasons were wet, dry, or when 141
one season was wet and the other was dry (Figure S2a). Results indicate periods of time when 142
both seasons were consistently wet (e.g., early 20th century) or dry (e.g., 1570s). Running 30-143
year counts were calculated for each case (Figure S2b) and for cases when the seasonal moisture 144
anomalies were of opposing sign (Figure S2c). To evaluate the reconstructions in the time-145
frequency domain, they were submitted to cross-wavelet analysis [Grinstead et al., 2004]. 146
Results reveal limited time-restricted concentrations of spectral power at various wavelengths for 147
both the cool-season and monsoon SPI reconstructions (Figure S3ab) and some concentrations 148
significant and common to both (Figure S3c). Coherence in the reconstructions is limited and 149
highly time restricted at periods less than about 32 years while at longer periods, in-phase 150
coherence is generally evident (Figure S3d). 151
152
-
8
References Cited 153
Castro, C. L, H. Chang, F. Domiguez, C. Carrillo, J Schemm, and H.H. Juang (2012) Can a 154
regional climate model improve the ability to forecast the North American Monsoon?, J. 155
Climate, doi:http://dx.doi.org/10.1175/JCLI-D-11-00441.1, in press. 156
157
Comrie, A. C., and E. C. Glenn (1998), Principal components-based regionalization of 158
precipitation regimes across the southwest United States and northern Mexico, with an 159
application to monsoon precipitation variability, Climate Res., 10, 201–215. 160
161
Cook, E. R., D. M. Meko, D. W. Stahle, and M. K. Cleaveland (1999), Drought reconstructions 162
for the continental United States, Journal of Climate, 12, 1145–1162, doi: 163
http://dx.doi.org/10.1175/1520-0442(1999)0122.0.CO;2. 164
165
Cook, E. R. (1985), A Time-Series Approach to Tree-Ring Standardization, Ph.D. Dissertation, 166
University of Arizona, Tucson, Arizona. 167
168
Cook, E. R., and L. A. Kariukstis (1990), Methods of Dendrochronology: Applications in the 169
Environmental Sciences, Klewer, Boston. 170
171
Cook, E. R., and K. Peters (1981), The smoothing spline: A new approach to standardizing forest 172
interior tree-ring width series for dendroclimatic studies, Tree-Ring Bulletin, 41, 45–53. 173
174
-
9
Daly, C., M. Halbleib, J. I. Smith, W. P. Gibson, M. K. Doggett, G. H. Taylor, J. Curtis, and P. 175
A. Pasteris (2008), Physiographically-sensitive mapping of temperature and precipitation across 176
the conterminous United States. Int. J. Climatol., 28, 2031-2064, doi:10.1002/joc.1688. 177
178
Douglass, A.E. (1941) Crossdating in dendrochronology, Journal of Forestry, 39, 825-828. 179
180
Gochis, D., J. Schemm, W. Shi, L. Long, W. Higgins, and A. Douglas (2009), A Forum for 181
Evaluating Forecasts of the North American Monsoon, Eos Trans. AGU, 90(29), 249, 182
doi:10.1029/2009EO290002. 183
184
Fritts, H. C. (2001), Tree Rings and Climate, 2nd ed., Blackburn Press, Caldwell, New Jersey. 185
186
Griffin, D, D. M. Meko, R. Touchan, S. W. Leavitt, and C. A. Woodhouse (2011), Latewood 187
chronology development for summer-moisture reconstruction in the U.S. Southwest, Tree-Ring 188
Research, 67, 87–101, doi: 10.3959/2011-4.1. 189
190
Grinsted, A., J. C. Moore, and S. Jevrejeva (2004), Application of the cross wavelet transform 191
and wavelet coherence to geophysical time series, Nonlin. Processes Geophys., 11, 561-566, 192
doi:10.5194/npg-11-561–2004. 193
194
Gutzler, D. S. (2004) An index of interannual precipitation variability in the core of the North 195
American Monsoon region, J. Climate, 17, 4473-4480, doi:http://dx.doi.org/10.1175/3226.1. 196
197
-
10
Holmes, R. L., (1983), Computer-assisted quality control in tree-ring dating and measurement. 198
Tree-Ring Bulletin, 44, 69–78. 199
200
Liebmann, B., I. Bladé, N. A. Bond, D. Gochis, D. Allured, and G. T. Bates (2008), 201
Characteristics of North American summertime rainfall with emphasis on the monsoon. J. Clim., 202
21, 1277–1294, doi: 10.1175/2007JCLI1762.1 203
204
McKee, T. B., N. J. Doesken, and J. Kleist (1993), The relationship of drought frequency and 205
duration of time scales, Paper presented at the Eighth Conference on Applied Climatology, 206
American Meteorological Society, Jan17-23, 1993, Anaheim, CA., pp.179–186. 207
208
Meko, D. M., and C. H. Baisan (2001), Pilot study of latewood-width of conifers as an 209
indicator of variability of summer rainfall in the North American monsoon region, 210
Int. J. Climatol., 21, 697–708, doi:10.1002/joc.646 211
212
Meko, D. M., R. Touchan, and K. J. Anchukaitis (2011), Seascorr: a MATLAB program for 213
identifying the seasonal climate signal in annual tree-ring time series, Comput.Geosci., 37, 1234–214
1241, doi: 10.1016/j.cageo.2011.01.013. 215
216
Michaelsen, J. (1987), Cross-validation in statistical climate forecast models, J. Clim. Appl. 217
Meteorol. 26, 1589−1600. doi:http://dx.doi.org/10.1175/1520-218
0450(1987)0262.0.CO;2. 219
220
-
11
Mock , C. J. (1996), Climatic controls and spatial variations of precipitation in the western 221
United States. J. Climate, 9, 1111–1125, doi: http://dx.doi.org/10.1175/1520-222
0442(1996)0092.0.CO;2. 223
224
NAME, North American Monsoon Experiment, (2013), website: 225
http://www.eol.ucar.edu/projects/name/documentation/NAME_Forecast_Forum.pdf, last 226
accessed 11 January, 2013. 227
228
Stahle, D. W., F. K. Fye, E. R. Cook, and R. D. Griffin (2007), Tree-ring reconstructed 229
megadroughts over North America since A.D. 1300, Climatic Change, 83,133–149, 230
doi:10.1007/s10584-006-9171-x 231
232
Stahle, D. W., M. K. Cleaveland, H. Grissino-Mayer, R. D. Griffin, F. K. Fye, M. D. Therrell, D. 233
J. Burnette, D. M. Meko, and J. Villanueva-Diaz (2009), Cool- and warm-season precipitation 234
reconstructions over western New Mexico, J. Climate, 22, 3729–3750, 235
doi:http://dx.doi.org/10.1175/2008JCLI2752.1. 236
237
Sheppard, P. R., A. C. Comrie, G. D. Packin, K. Angersbach, and M. K. Hughes, (2002), The 238
climate of the U.S. Southwest. Climate Research 21, 219–238. 239
240
Stokes, M.A. and T.L. Smiley, (1996), An Introduction to Tree-Ring Dating, 2nd ed., University 241
of Arizona Press, Tucson, Arizona. 242
243
-
12
St. George, S., D. M. Meko, and E. R. Cook (2010), The seasonality of precipitation signals 244
encoded within the North American Drought Atlas. The Holocene, 20, 983–988, 245
doi:10.1177/0959683610365937 246
247
Zhu, C., D. P. Lettenmaier, and T. Cavazos (2005), Role of antecedent land surface conditions 248
on North American monsoon rainfall variability, J. Climate, 18, 3104–3121, doi: 249
http://dx.doi.org/10.1175/JCLI3387.1. 250
251
-
13
Table S1. 252
Metadata for tree-ring chronologies considered as candidate predictors for SPI reconstructions 253
EWa LWb
Site
Code Site Name Species ITRDB Codec Lat. Long.
Elev.
(m)
Inner
Date
Outer
Date
Yes No BFM Black Mountain Fir PSME NM565 33.35 -108.26 2489 1306 2008
Yes No BPM Black Mountain Pine PIPO NM566 33.35 -108.26 2489 1494 2008
Yes Yes FSM Florida Saddle PSME AZ504 31.72 -110.84 2385 1493 2008
Yes Yes SCM Santa Catalina High PSME AZ501 32.44 -110.78 2746 1539 2009
Yes No WKM Wahl Knoll PSME n/a 33.99 -109.38 2969 1460 2008
Yes Yes WMP Walnut Canyon Pine PIPO AZ049, AZ547 35.17 -111.51 1995 1531 2010
Yes Yes WPU Webb Peak PSME AZ558, AZ559 32.71 -109.92 3014 1466 2008
254 aCool-season SPI candidate predictor based on correlation screening and chronology length 255 bMonsoon SPI candidate predictor based on correlation screening and chronology length 256 cInternational Tree-Ring Data Bank reference code (http://www.ncdc.noaa.gov/paleo/treering.html) 257
258
-
14
Table S2. 259
Calibration model equations. 260 Model Equation
Cool-season SPI yhat = -2.700 + (1.393*FSM_EW) + (0.674*BFM_EW) + (0.673*WMP_EW)
Monsoon SPI yhat = -3.275 + (1.487*FSM_LW) + (0.986*WMP_LW) + (0.904*SCM_LW)
261
262
-
15
Table S3. 263
Calibration and verification information. 264
Cool-Season SPI Monsoon SPI
Potential predictors 7 4
Predictors in final equation 3 3
Calibration Period 1896–2008 1896–2008
n (years) 113 113
R2 0.616 0.462
adjusted R2 0.606 0.468
RE 0.589 0.42
Standard Error of Estimate 0.635 0.75
Root Mean Squared Error of
Cross-Validation 0.645 0.77
Regression model F-value 58.36 31.24
F-value probability < 0.00000 < 0.00000
Durbin Watson 1.88 1.96
Portmanteau Q 10.69 8.89
Portmateaau Q probability 0.38 0.54
Sign Test Hits 94/113 84/113
265
266
-
16
Figure S1. 267
Box and whisker plots illustrate the distribution of NAM2 monthly precipitation for the 1896–268
2008 period, centered on the water year of October through September. Boxes capture the inter-269
quartile range and the horizontal black line represents the median value. Upper and lower 270
whiskers capture values +/- 1.5 times the 75th and 25th percentiles, respectively. Dots represent 271
extreme values beyond that range. 272
273
Figure S2. 274
Dot plots illustrate Pearson coefficients for the 26 NAM2 chronologies correlated with monthly 275
precipitation from prior June through current October for the period 1896–2008. Earlywood 276
chronology correlations are shown in A and latewood chronology correlations are shown in B. 277
The horizontal black line represents the 0.05 non-exceedance probability threshold. 278
279
Figure S3. 280
Consecutive year events of positive (blue) and negative (red) SPI for the cool-season SPI (a) and 281
monsoon SPI (b). Notable consecutive year events are annotated. 282
283
Figure S4. Bar plots illustrate the distribution of cases (a) when both seasons were wet (blue), 284
when the cool season was wet and the monsoon was dry (green), vice-versa (orange), and when 285
both seasons were dry (red). Time series illustrate running 30-year counts of each case (b). Time 286
series illustrate running 30-year counts of cases when the seasonal moisture anomalies were of 287
opposing sign (c; wet-dry or dry-wet; black). 288
289
-
17
Figure S5. Continuous wavelet power transform of reconstructed cool-season SPI (a) and 290
monsoon SPI (b), with significant (p
-
griffin_grl_monsoongriffinGRL_revised_compiled_mid