ENVIRONMENTAL CHANGES IN THE SOUTHERN CANADIAN ROCKIES FROM MULTIPLE TREE-RING PROXIES
Emma Watson1, Brian Luckman2, Greg Pederson3 and Rob Wilson4
1Meteorological Service of Canada, Environment Canada 2Department of Geography, University of Western Ontario 3U.S. Geological Survey and Big Sky Institute, Montana State University 4School of Geosciences, Edinburgh University
Peyto Glacier in 1966 taken by W.E.S. Henoch (NHRI - Canadian Glacier Information Centre). Introduction
• study of glacier fluctuations had traditionally provided many of our ideas about the climate history of the last millennium in the Canadian Rockies
• Paleoclimate signal in glacier records is complex, incomplete and biased to large events
• we describe tree-ring based research of the late Holocene climate of the area
• in particular we detail the development of continuous records of temperature, precipitation, glacier mass balance and streamflow from tree-ring chronologies sampled in different environments which can be compared with the glacial record Moraine dating is available from 66 glacier forefields PEYTO in the Canadian Rockies
Luckman, 2000 Summary of Little Ice Age (LIA) glacier events in the Canadian Rockies
Periods of advance:
• 1150-1350 (advances through forest- calendar dated logs) • selected preservation of glacier record between 14th and 17th centuries • widespread advances early 18th and throughout 19th century May-August Maximum temperatures Columbia Icefield, Canadian Rockies 950-1995 Based on regional ringwidth and maximum tree-ring density chronologies
Update to Luckman (1997) using: more chronologies from wider area (i.e. better replication and more regionally representative); different predictand (original Apr-Aug mean) and RCS standardization of MXD data • RCS on average cooler, shows more low frequency trend, 1690s most extreme cold period reconstructed
Anomalies from 1901-1980 mean Comparison with northern Hemisphere temperature reconstructions
Standardized to the 1000-1980 period. Tree ring chronologies and precipitation records Instrumental and estimated precipitation Smoothed annual precipitation reconstructions for the southern Canadian Cordillera Widespread wet periods Widespread dry periods Using tree-rings to study glacier mass balance
• mass balance is a time series that represents the difference between accumulation and ablation on a glacier on an annual basis - mass balance records are short (~30 yrs)
• because tree-ring chronologies are responsive to similar climate controls as those that influence mass balance, they may be useful for studying and reconstructing mass balance
• Developing seasonal mass balance reconstructions can help us understand the climate parameters responsible for changes in glacier length (e.g. do the contributions of winter and summer balance vary back through time or during important intervals (e.g. the LIA max?) Peyto Glacier, Rocky Mountains, Alberta
Peyto Glacier in 1966 taken by W.E.S. Henoch (NHRI - Canadian Glacier Information Centre).
Area estimates • Outlet glacier from Wapta Icefield 1887 17.15 km2 Peyto Glacier in 1966 taken by W.E.S. Henoch • contributes flow to the Mistaya River (NHRI - Canadian Glacier Information2 Centre). catchment and the North Saskatechewan 1966 13.5 km River Basin 1993 11.81 km2 • 34% decrease in Bw Impact of 1976 PDO shift on glacier mass balance in the Pacific Northwest Relationships between measured mass balance and instrumental climate (1966-1995)
• meteorological data from stations in Banff, Jasper, Lake Louise, Golden, Carway, Valemont • agrees with other findings for continental glaciers (e.g. Yarnal, 1984)
Winter balance (Bw) – October-April precip Summer balance (Bs) – June-Aug temps
500
450 Banff precipitation Banff temperature 17
400 C)
) 15
350 o m p ( m
300 m p ( i
c 13 ug t e 250 A il pr un-
pr 200 J A
11 n - t a
150 e Oc m 100 y = 0.1296x + 93.774 y = -0.0009x + 11.18 9 50 R = 0.70 R = -0.48 2 R2 = 0.49 R = 0.23 0 7 0 500 1000 1500 2000 2500 -3000 -2500 -2000 -1500 -1000 -500 0 Bw (mwe) Bs (mwe)
350 18 Jasper precipitation Jasper temperature 17 300
16 C) ) 250 o m p ( m 15 m p ( t i 200 g c u e 14 A il pr
150 un- pr
13 J A n - t a 100 e Oc 12 m y = 4.8817x + 256.81 y = -0.0014x + 11.723 50 R = 0.65 R = -0.83 11 R2 = 0.42 R2 = 0.61 0 10 0 500 1000 1500 2000 2500 -3000 -2500 -2000 -1500 -1000 -500 0 Bw (mwe) Bs (mwe) Reconstruction Strategy
• goal = reconstruct seasonal and net mass balance from tree-ring chronologies
• ablation related to SUMMER temperatures therefore use summer temperature reconstruction
• winter accumulation more difficult because there are no winter-sensitive chronologies available for the area
• winter climate in Alaska and the southern Canadian Cordillera are both influenced by conditions in the North Pacific
Typical winter season climate anomalies during positive PDO years
• therefore, use tree-ring chronologies from Alaska to study WINTER mass balance at Peyto Location and length of records used in the study
Lat. N Long. W Elev. (m) Prov./State Length Monthly Climate Records Banff Precipitation1 51 11 115 34 1389 Alb. 1895-1995 Banff Temperature1 51 11 115 34 1389 Alb. 1895-2001 Jasper Precipitation1 52 53 118 04 1061 Alb. 1936-1995 Jasper Temperature2 52 53 118 04 1061 Alb. 1916-1994
Mass Balance Peyto3 51 41 116 32 2140-3180 Alb. 1966-1997
Tree-Ring Data Miners Well (MW)4 60 00 141 41 650 Alaska 1428-1995 Athabasca (ATHA)5 52 13 117 14 2000 Alb. 869-1994 Waterton (WA)6 49 28 113 34 1200 Alb. 1673-1996 6 Lytton (LY) 50 14 121 35 258 B.C. 1468-1996 Reconstructed versus measured mass balance at Peyto Glacier
The predicted net mass balance series is the difference between the winter and summer series
It is not a separate reconstruction. Mass balance reconstructions (1673-1994) for Peyto Glacier based on tree-ring data
• 1966-1995 winter accumulation below long-term mean
• 1966-1995 summer ablation is greater than the long-term mean
• 1966-1995 net mass balance is well below the long-term mean
• the LIA maximum at Peyto glacier is estimated between ca. 1836 and 1841 based on dating of killed and damaged trees along the western trimline (Luckman, 1996)
mean balances 1673-1883 +70 mm w.e./yr. 1883-1993 -317 mm w.e./yr.
• the decrease in Bn since the 1880s corresponds well with Wallace’s (1995) estimate that Peyto has lost 70% of its volume over the past 100 years Moving correlations between the seasonal and net mass balance series
Correlations calculated using a 31-year window (plotted on central year). The dashed horizontal lines denote statistical significance (p<0.05). Correlations over the full period (1673-1994) are given in parenthesis beside each legend entry.
• over the full period correlations are highest between Bn and Bs • Bn-Bw and Bn-Bs correlations are variable but always significant • Bw-Bs are not significantly correlated over the full period but correlate positively during the early 1700s and 1800s (higher accumulation and reduced ablation) and again near the end of the record (lower accumulation and greater ablation) The LIA moraine record for the Canadian Rockies and reconstructed mass balance at Peyto Glacier
Dated moraines in the Canadian Rockies (66 glaciers, 25 year intervals Luckman 2000)
• The reconstructed periods of positive mass balance during the early 18th, early and late 19th centuries immediately precede or coincide with regional periods of moraine development in the Canadian Rockies.
•The correspondence between these totally independent proxy climate records provides strong mutual verification of the regional climate history.
• some glaciers in the Premier range built small moraines in the 1970s (Luckman et al., 1997) Exploring relationships with Pacific SSTs
• winter mass balance reconstruction filtered to yield time series of high (<8 years) and low (>8 years) frequency variability • Nov-Mar SSTs regressed on the mass-balance reconstructions (1870-1994) previous studies have associated changes in glacier mass balance with decadal-Interdecadal variability in the Pacific Ocean • the mass balance reconstructions allow relationships with SSTs to be explored over a much longer period than the short measured balance records (1966-present) permit
Winter balance – high frequency Winter balance – low frequency
ENSO-like pattern Resembles pattern of decadal-interdecadeal variability identified previously (e.g. Zhang et al. 1997) • Luckman (2000) noted that at the most northerly sites (Mount Robson area and Premier Ranges), most glaciers have 18th century moraines at their downvalley limits while few glaciers farther south formed moraines during this period (e.g. Kananaskis area)
• only 19th century moraines have been identified in Waterton Lakes National Park and Glacier National Park, Montana
1300 1400 1500 1600 1700 1800 1900 Mass Balance reconstructions in the Canadian and northern U.S. Rockies
• How do recent mass balance reconstructions for Peyto Glacier and Glacier National Park, Montana compare?
Does the timing of the LIA maximum coincide? Mass Balance reconstructions for Peyto Glacier and Glacier National Park (GNP)
For a detailed comparison of these mass balance records please see the poster by G. Pederson, E. Watson, B. Luckman, D. Fagre, S. Gray and L. Graumlich titled: Tree-Ring Based Estimates of Glacier Mass Balance in the Northern Rocky Mountains for the Past 300 Years Exploring the controls of pre-instrumental streamflow
How have changes in precipitation (winter and summer) and glacier wastage affected streamflow?
Can this be addressed using tree-ring based reconstructions?
Bow at Banff streamflow (1912-1996) • flows from Bow glacier in Wapta Icefield across the semi-arid Canadian Prairies • unregulated upstream
Banff precipitation • precipitation-sensitive Douglas-fir chronologies exist and have been used to reconstruct annual (July-June precipitation)
Peyto Glacier Mass balance • outlet glacier from Wapta Icefield • regionally representative • separate summer and winter balances available which represent summer melting Data used in this study and winter accumulation (i.e. snowpack) What controls streamflow? Instrume ntal Correlations with climate-related Bo w Ap r-Au g parameters over the instrumental period Com m on p e riod (1966- 1994 1) • does not correlate highly with summer In strum ental precipitation Banf f P re ( Apr-A ug) 0. 15 •streamflow correlates most highly with Banf f P re ( Oct-A pr) 0. 68 winter precipitation at Banff (r=0.68) Peyto Bw 0. 60 Peyto B s 0. 09 • Bw correlates with winter precipitation at Banff (r = 0.60) and Bow streamflow and can therefore be used as a surrogate for winter snowpack
•flow is snowmelt dominated
• 76% of flow is concentrated in the months April-August Developing a physically realistic streamflow reconstruction for the Bow at Banff
• Decided to model summer flow • Predictors = precipitation-sensitive chronologies from Cranbrook, Jasper and Waterton AND winter Balance (as a surrogate for winter snowfall)
Summary characteristics of the reconstruction Reconstruction Record/Season Length 1 N calib. calib. calib. ver. 2 SE period R Adj. R2 r2 % of mean
Bow at Banff streamflow/Apr-Aug 1619-1995 4 1912-1995 0.60 0.33 0.29 25 Summer precipitation reconstruction for Banff
• April-August based on Douglas-fir chronology from Banff • totally independent of streamflow reconstruction
Summary characteristics of the reconstruction Reconstruction Record/Season Length 1 N calib. calib. calib. ver. 2 SE period R Adj. R2 r2 % of mean
Banff precipitation/Apr-Aug 1308-1995 2 1896-1994 0.60 0.35 0.31 43 How realistic is the streamflow Reconstructed reconstruction? Bow Apr-Aug
Common period (1966-1994) • instrumental streamflow correlates most highly with winter precipitation at Banff Reconstructed (r=0.68; 1966-1994) and this relationship is Banff Pre (Apr-Aug) 0.24 also identified in the reconstructed series. Peyto Bw 0.53 Peyto Bs 0.11
Maximum Paired Interval (1912-1996)
Reconstructed Banff Pre (Apr-Aug) 0.28 Peyto Bw 0.41 Peyto Bs 0.01 Simple comparison of reconstructions
Winter precipitation (snowpack)
• difficult to establish to what extent coherent departures Summer melting are a result of causal relationships or pure coincidence
• However, the majority of low and high streamflow Summer precipitation events over past 350 years are related to changes in snowpack (33%) or snowpack and summer precipitation (44%)
Summer streamflow • remaining 23% of pronounced high and low streamflow events appear to be related to changes in summer precipitation alone Moving correlations between streamflow and inputs
• Relationships between the variables vary considerably through time
• Correlations between streamflow and summer balance (glacier melting) over the full period of record are not significant (p>0.05) -- small component of total flow; outside the seasonal window; proportion of flow in individual months related to glacier wastage varies by year.
• suggests that total volume of flow may not change as much as the timing of flow and that such changes in timing cannot be detecting using these tree-ring data Conclusion
• Tree-ring chronologies from different species and environments can be used to reconstruct temperature and precipitation and more complex variables such as glacier mass balance and streamflow in the Canadian Rockies
• Intercomparison of independently-derived proxy records from within and outside the region are useful for validating results and identifying the spatial scales of climate events (e.g. droughts)
• Future attempts to integrate and compare reconstructions must try to reconcile differences in the climate signal in the proxies and the timing of changes in the environmental variable of interest
• A better understanding of the relationships between glacier mass balance, streamflow and the major driving forces of climate variability (e.g. ENSO, PDO etc ) will be useful in predicting how these phenomena may respond to future climate changes.