Projected future changes in surface marine winds off the west coast of Canada

Transcription

1 Click Here for Full Article JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 114,, doi: /2008jc005123, 2009 Projected future changes in surface marine winds off the west coast of Canada W. J. Merryfield, 1 B. Pal, 2 and M. G. G. Foreman 2 Received 12 September 2008; revised 10 February 2009; accepted 26 March 2009; published 9 June [1] Projected future changes in the seasonal climatology of surface marine winds at locations off the west coast of Canada having long records of buoy wind observations are evaluated, drawing on output from 18 climate models. The ensemble mean modeled wind climatology for a late 20th century control period is found to reproduce that of the buoy observations reasonably well, except where local orographic influences not resolved by the models are strong. Over the 21st century, ensemble mean summertime upwelling favorable winds increase in speed by 5 10% and rotate clockwise by 5, statistically significant changes. By contrast, 21st century changes in ensemble mean wintertime downwelling favorable winds are not statistically significant. These trends are too weak to be detectable in the 20th century buoy observations. Citation: Merryfield, W. J., B. Pal, and M. G. G. Foreman (2009), Projected future changes in surface marine winds off the west coast of Canada, J. Geophys. Res., 114,, doi: /2008jc Introduction [2] Surface marine winds along the west coast of Canada exert a pronounced influence on the oceanography and ecosystems of the region. Although other forcing influences are also significant, including the bifurcation of the eastward North Pacific Current into the southward California Current and the northward Alaska Current, and freshwater forcing from coastal rivers and the estuarine circulation of the Strait of Juan de Fuca [Freeland et al., 1984], marine winds dominate the seasonality of the coastal surface currents. In autumn and winter generally southerly winds are associated with a northward flowing, downwelling coastal current regime, whereas in spring and summer northwesterly winds give rise to a southward flowing, upwelling regime [e.g., Murphree et al., 2003]. This cycle strongly modulates primary production and other aspects of the coastal ecosystem and biogeochemical cycling [Ianson and Allen, 2002]. [3] Long-term observations of the coastal winds in this region have been undertaken via a network of coastal meteorological buoys since the late 1950s [Cherniawsky and Crawford, 1996]. Gaps and irregularities in data have been removed using an empirical-statistical downscaling procedure applied to large-scale predictors derived from the NCEP reanalysis [Faucher et al., 1999]. The resulting data set consists of 6 hourly winds at 13 buoy sites for the 40 year period [4] North Pacific climate is characterized by pronounced interdecadal variations [e.g., Mantua et al., 1997; Zhang et al., 1997], which are correlated with major changes in marine 1 Canadian Centre for Climate Modelling and Analysis, University of Victoria, Victoria, British Columbia, Canada. 2 Institute of Ocean Sciences, Sidney, British Columbia, Canada. Published in 2009 by the American Geophysical Union. ecosystems [Peterson and Schwing, 2003]. This variability has tended to obscure anthropogenically induced climate change in this region, although climate models project that the anthropogenic influence on SST will become comparable to the interdecadal variability during the first half of the 21st century [Overland and Wang, 2007]. Other projected anthropogenic shifts include a deepening and northward shift of the Aleutian Low, accompanied by a northward shift in north Pacific storm tracks [Yin, 2005; Salathé, 2006], and a shallowing of the ocean mixed layer (W. J. Merryfield, manuscript in preparation, 2009). [5] The aim of this paper is to examine the likely impact of anthropogenic climate change on surface marine winds off the Canadian west coast, motivated by the possibility that such changes could have significant impacts on coastal ecosystems. The approach is to examine such changes in an ensemble of climate model simulations representing 21st century climate change under a projected anthropogenic forcing scenario. The models and procedure used to obtain modeled winds at the buoy locations are described in section 2, and the modeled winds are validated against 20th century observations in section 3. Projected 21st century changes are detailed in section 4 and compared with observed 20th century trends in section 5. Results are discussed and conclusions drawn in section Models and Analysis Procedure [6] The climate model output used for this study was obtained from the World Climate Research Programme s (WCRP s) Coupled Model Intercomparison Project phase 3 (CMIP3) multimodel data set, assembled at the Program for Climate Model Diagnosis and Intercomparison (PCMDI) to inform the IPCC Fourth Assessment. Of the models represented in this archive, 18 met selection criteria based on the availability of monthly mean surface (10 m) wind output for a baseline period under the Climate of the 20th 1of10

2 Table 1. Climate Models Used in This Study and Their Atmospheric Resolutions a Alphabetical Code Institution/Model Atmospheric Resolution Horizontal Grid Dimensions, Longitude Latitude a BCCR/BCM2.0 T63L b CCCMA/CGCM3.1(T47) T47L c CCCMA/CGCM3.1(T63) T63L d CCSR/MIROC3.2(med) T42L e CNRM/CM3 T63L f CSIRO/Mk3.5 T63L g GFDL/CM L h GFDL/CM L i GISS/AOM 4 3 L j GISS/EH 5 4 L k GISS/ER 5 4 L l INM/CM L m IPSL/CM L n MIUB/ECHO-G T30L o MPI/ECHAM5 T63L p MRI/CGCM2.3.2 T42L q UKMO/HadCM L r UKMO/HadGEM L a Horizontal resolution is described by spectral truncation or grid box dimensions as appropriate, and vertical resolution is described by the number of model levels, e.g., L31. Century scenario, as well as for two 20-year periods, and , under the SRES A1B scenario for 21st century radiative forcing. These models are listed along with their grid dimensions in Table 1. [7] Because model grids do not generally coincide with buoy positions, two methods have been used to obtain model winds at the locations of the buoys. The first is simply to select the nearest-neighbor grid value (taking an average in the event of a tie ). The second consists of bilinear interpolation between the four grid values surrounding the buoy location. Monthly time series were constructed for each model and buoy for each of the three 20-year periods considered, and from these values monthly time series specific to each period were constructed. Land fractions specific to each buoy location were also calculated as an indication of the relation of the buoy locations to modeled coastlines. Figure 1. Locations of 13 weather buoys off the Canadian west coast considered in this study. 2of10

3 Table 2. List of Buoys and Their Positions Buoy Latitude Longitude Classification N W Far offshore N W Far offshore N W Near offshore N W Inshore N W Near offshore N W Hecate Strait N W Far offshore N W Inshore N W Inshore N W Near offshore N W Near offshore N W Near offshore N W Near offshore [8] The 13 buoy sites along the west coast of British Columbia, Canada are illustrated in Figure 1, and their geographical positions are listed in Table 2. In order to condense the analysis the buoy locations are classified as far offshore, further than 200 km from outer coast, near offshore, within 200 km of outer coast, and inshore, occupying the Inside Passage between coastal islands and the British Columbia mainland. In addition, buoy 183, which has an atypical seasonal climatology subject to strong orographic influences in Hecate Strait, is considered separately. 3. Validation Against 20th Century Observations [9] The credibility of model projections of future climate depends largely on the models ability to simulate observed properties of the present climate. The skill of models in representing surface marine winds hinges on their ability to represent the large-scale atmospheric circulation of the North Pacific, including its seasonality. However, an important limitation even of models that represent present-day large-scale climate of the North Pacific skillfully is that their typical atmospheric grid resolutions of km render them unable to represent smaller-scale orographic influences that can affect marine winds in coastal and shelf waters. [10] With such considerations in mind, this section examines the ability of the models to represent the observed seasonal climatology of marine winds at the buoy locations. The control period was selected as representative of late 20th century climate, and because the reconstructed wind data of Faucher et al. [1999] includes and is free of gaps during this period. [11] Figure 2 compares observed and modeled seasonal climatologies for , for each of the four geo- Figure 2. Seasonal surface wind climatologies for regional groupings of buoys defined in Table 2 for : (a) far offshore, (b) near offshore, (c) inshore, and (d) Hecate Strait. Thick vectors indicate buoy observations, whereas thin vectors indicate multimodel averages obtained at the buoy positions through nearest neighbor (dashed) and bilinearly interpolated (solid) values. Annual mean climatologies are shown at right, with axes indicating 1 m s 1 zonal and meridional wind components. 3of10

4 Figure 3. Wintertime (JFM) mean surface winds from buoy observations (vectors), individual models (alphabetical codes from Table 1), and multimodel mean (large circles) for regional groupings of buoys. The error bars represent standard deviations of the multimodel ensemble. graphical classifications indicated in Table 2. In this plot, the observed monthly mean climatologies constructed from the buoy records are represented by the thick arrows, whereas multimodel ensemble means are shown both for the nearestneighbor calculations (dashed) and bilinear interpolations (thin solid). Annual mean climatological wind vectors are indicated at far right, where the axes denote longitudinal and latitudinal wind speeds of 1 m s 1. In the case of the far offshore buoys, the models capture the weak reversal in the north-south wind component that occurs in July and August. Comparable agreement is seen for the near-offshore buoys, which exhibit a stronger wind reversal that encompasses June. The directions of the modeled annual mean wind vectors are in close agreement with observations for both classes of offshore buoys, although the modeled wind speeds are 20 40% larger than observed. Some of this disagreement is attributable to the difference in height between the 10 m model winds and the buoy winds, which are sampled near 5 m [Faucher et al., 1999]. Although in principle it is possible to approximately correct for such differences [Cherniawsky and Crawford, 1996], such a correction has not been attempted here because we are mainly interested in fractional 21st century changes in wind speed, which to a first approximation should be independent of height. [12] Figures 2c and 2d compare modeled and observed climatological winds at near-coastal locations. For the inshore buoys, the modeled wind directions in seasons other than summer are consistently biased clockwise, and the annual mean wind direction shows a weak westerly component, similar to that for the near-offshore locations, rather than the easterly component and generally alongshore orientation that is observed. These differences are likely attributable to the coarsely resolved coastal orography in the models; nonetheless, the model ensemble means again capture the summer weakening and reversal of the alongshore wind component reasonably well. Perhaps not surprisingly, the model ensemble means perform particularly poorly at the location of the Hecate Strait buoy (Figure 2d), where local orographic influences dominate in summer, resulting in generally easterly rather than westerly winds. 4of10

5 Figure 4. As in Figure 3 but for JJA. [13] Figures 3 and 4 illustrate scatter among the individual models for the regional groupings of buoys indicated in Table 2 and for averages over all the buoys (the Hecate Strait buoy is not considered as a separate region but is included in the All classification). Shown are the seasonal mean winter (JFM) and summer (JJA) winds according to the buoy observations (vectors), the individual models (symbols defined in Table 1), and the multimodel mean (larger circle); the error bars represent standard deviations of the multimodel ensemble. The multimodel means are again seen to reasonably represent the observed seasonality of the winds within the limitations imposed by model resolution, whereas individual model biases are typically larger, as is frequently the case for climate model ensembles [e.g., Lambert and Boer, 2001]. As for individual months, wind speeds at offshore locations are biased high. Winter wind directions (Figure 3) are relatively accurate far offshore, but exhibit a northerly bias at nearshore and inshore locations, likely due to underrepresentation of the orographic influence of the southeast-to-northwest oriented coastline. (The progressive anticlockwise rotation from far offshore to inshore locations is however partially captured suggesting that at least some orographic influence exists the models.) Wind speeds and directions in summer are represented reasonably well in the ensemble mean, indicating that orographic influences may be emerging more realistically than in JFM, although again there is some scatter among the individual models (Figure 4). [14] Ability to represent observed 20th century winds was also considered for a subset of eight models common to the 10-model subset determined by Overland and Wang [2007] to represent 20th century North Pacific SST variability reasonably well. (Specifically these are models b, c, d, g, h, n, p and q in Table 1.) For JFM the mean winds and intermodel scatter for this subset are nearly the same as for the full set in Figure 3. For JJA the mean model wind vectors for the subset are rotated counterclockwise with respect to those in Figure 4 although they remain within the standard deviations of the full set; in this case the intermodel scatter is somewhat smaller than for the full set of models. 4. Projected 21st Century Changes [15] The preceding section showed that, although individual models exhibit biases in wind speed and direction at 5of10

6 Figure 5. Future changes in climatological (left) JFM and (right) JJA surface winds, averaged at faroffshore buoy locations, between the base period and (top) and (bottom) The letters correspond to individual models as in Table 1, and the bulls-eyes correspond to multimodel means. The larger numerals indicate the numbers of models residing in the corresponding half planes, and the smaller numerals indicate the numbers of models in the corresponding quadrants. Bold letters indicate p < the buoy locations, the seasonalities of the modeled winds are generally in accord with late 20th-century observations. In this section, future trends in the modeled winds in response to human-induced climate change are examined. Two intervals, and , are compared to the baseline period. The A1B scenario representing an intermediate projected rate of increase in greenhouse gas concentrations [Intergovernmental Panel on Climate Change [(IPCC), 2007] is considered. [16] In order to factor out model biases, changes in wind climatologies are assessed in terms of fractional changes in wind speed and changes in wind direction. To improve statistical confidence, these changes are computed for winter (JFM) and summer (JJA) averages, with buoys grouped according to the classification in Table 2. For changes obtained from individual models, statistical significance is assessed by applying a difference in means t test using sample variances computed from the yearly values of these seasonal and regional averages. For changes in the multimodel ensemble means, variances instead are based on the spread of the ensemble, which is a more appropriate (and much larger) indicator of uncertainty than the interannual variance of the ensemble mean because differences and uncertainties in model formulation enter. [17] Changes in speed and direction of winter and summer climatological winds are shown for the far offshore buoys in Figure 5, the near offshore buoys in Figure 6, and the inshore buoys in Figure 7. (The Hecate Strait results resemble those for the inshore buoys and are not shown.) In each plot, values for the individual models are coded as in Table 1; dark symbols are used if p < 0.05 for changes either in wind speed or direction, p indicating the probability that such a result would be realized under a null hypothesis of no change. The bulls-eyes in Figures 5 7 indicate changes in the ensemble mean winds, numerical values for which, along with 1s confidence intervals and p values, are listed in Table 3. The larger numerals at the top and bottom of each plot in Figures 5 7 indicate how many models exhibit wind speed increases vs wind speed decreases, whereas those on the right and left sides similarly count models showing positive (clockwise) and negative (anticlockwise) changes in direction. The smaller numerals in each corner denote the number of model wind changes in the corresponding quadrants. [18] From Figures 5 7 and Table 3 it is evident that there are no statistically significant changes in JFM ensemble mean winds between and , whereas the individual model changes exhibit considerable scatter and generally also are not statistically significant. 6of10

7 Figure 6. As in Figure 5 but for near-offshore buoy locations. Figure 7. As in Figure 5 but for inshore buoy locations. 7of10

8 Table 3. Changes in Ensemble Mean Model Winds at Buoy Positions JFM JJA Region Epoch Dmag (%) (p) Ddir (deg) (p) Dmag (%) (p) Ddir (deg) (p) Far offshore ± 3.0 (>0.2) 1.9 ± 2.7 (>0.2) 1.6 ± 1.3 (>0.2) 1.9 ± 1.4 (0.18) ± 5.0 (>0.2) 2.8 ± 3.1 (>0.2) 4.5 ± 2.9 (0.13) 4.2 ± 1.7 (0.02) Near offshore ± 3.0 (>0.2) 1.2 ± 1.8 (>0.2) 4.0 ± 1.9 (0.06) 2.6 ± 1.2 (0.04) ± 4.9 (>0.2) 1.2 ± 3.1 (>0.2) 9.0 ± 3.0 (0.008) 4.6 ± 1.8 (0.02) Inshore ± 2.9 (>0.2) 1.2 ± 2.0 (>0.2) 2.8 ± 1.9 (0.17) 2.6 ± 1.3 (0.06) ± 4.9 (>0.2) 0.3 ± 3.2 (>0.2) 7.9 ± 3.2 (0.03) 4.8 ± 2.1 (0.03) All ± 3.2 (>0.2) 1.5 ± 1.9 (>0.2) 2.9 ± 1.7 (0.10) 2.4 ± 1.2 (0.06) ± 5.1 (>0.2) 1.2 ± 3.7 (>0.2) 7.3 ± 2.9 (0.02) 4.7 ± 1.7 (0.02) [19] The JFM wind changes between and tend to be larger, and are statistically significant (p < 0.05) for approximately one third of the models. However, there is little consistency between the individual model changes, and Table 3 indicates that although the ensemble means exhibit a wind speed increase of around 5%, these changes are not significant because of the large interannual variability in winter winds. The ensemble mean wind directions exhibit slight counterclockwise rotations that also are not statistically significant. [20] A much clearer picture emerges for the JJA wind changes. Between and , although only a few of the individual model changes are significant, the ensemble means exhibit wind speed increases of approximately 2 4% and clockwise rotations of about 2 that are statistically significant in some instances. [21] The JJA wind changes between and show similar though considerably stronger trends consisting of wind speed increases ranging from 4.5% to 9% and clockwise rotations between 4 and 5, with p < 0.05 in most cases. The strength of this signal is reflected in substantial intermodel consistency, with 16 models showing clockwise rotations at the near offshore buoy locations (probability of chance occurrence 10 3 ; see Table 4) and 17 of the 18 models showing clockwise rotations at the inshore buoy locations (probability of chance occurrence 10 4 ). Approximately two thirds of the models jointly exhibit clockwise rotations and wind speed increases in these regions. 5. Comparison With Observed 20th Century Trends [22] To ascertain whether the 21st century trends detected in the climate model output are apparent in the buoy observations, the latter were averaged by season and region in a similar way to the models. Linear trends were then computed for the period during which the reconstructed buoy observations are free of gaps. These trends along with p values obtained from a two-tailed t test are listed in Table 5, where trends are scaled in units of (% change) century 1 for wind speed and ( of clockwise rotation) century 1 for wind direction to facilitate comparison with the minus changes listed in Table 3. [23] It is apparent in comparing Tables 5 and 3 that the observed trends are generally much larger than those inferred from the climate model ensemble means. This is not unexpected, because averaging over the ensemble of models greatly reduces the sampling error associated with unforced climate variability. Indeed the minus changes for the individual models indicated in Figures 5 7, which can exceed ±30% in wind speed and ±20 in wind direction, are closer in magnitude to the trends in Table 5, although the observed trends still exceed those of any single model in most instances. [24] The large sampling error associated with the 38 years of buoy observations is evident in the low statistical significance of most observed trends. Thus, although there are some similarities in the signs of the observed trends with those of the multimodel means, including generally increasing wind speeds along with anticlockwise rotation in JFM, and clockwise rotation in JJA, this comparison must be regarded as inconclusive. 6. Discussion and Conclusions [25] This study has examined trends in 21st century surface marine winds at locations off the west coast of Canada where long records of buoy observations are available for validating model results and evaluating observed 20th century trends. Output from an ensemble of 18 climate models under the Climate of the 20th Century and A1B radiative forcing scenarios has been examined. [26] In the winter season, when marine winds are generally southerly and downwelling favorable, the multimodel ensemble mean exhibits statistically insignificant changes consisting of an approximately 5% intensification and slight counterclockwise rotation. The relative weakness of this response might at first seem contrary to the robust modeled changes in wintertime North Pacific circulation, character- Table 4. Probability That by Chance Occurrence at Least N 2 of N 1 + N 2 Models Exhibit Changes of the Same Sign N 1 N 2 p of10

9 Table 5. Trends in Observed Buoy Winds ( ) Region Dmag (%/century) (p) JFM Ddir ( /century) (p) Dmag (%/century) (p) JJA Ddir ( /century) (p) Far offshore 9 (>0.2) 119 (0.03) 49 (0.13) 7 (>0.2) Near offshore 46 (>0.2) 68 (0.15) 31 (>0.2) 27 (>0.2) Inshore 68 (>0.2) 35 (>0.2) 47 (>0.2) 54 (>0.2) All 65 (>0.2) 55 (>0.2) 37 (>0.2) 32 (>0.2) ized by a northward shift and intensification of the Aleutian Low, as reported for example by Salathé [2006]. The associated changes in mean sea level pressure (MSLP) from to for the 18 models considered here are illustrated in Figure 8 (top) [see also Salathé, 2006, Figure 2; IPCC, 2007, Figure 10.9], where the hatching indicates regions where at least 70% or 13 of the 18 models agree as to the sign of the change (this is approximately a 10% significance measure according to Table 4). It is seen that the Canadian west coast lies near a weak local maximum in anomalous MSLP in a region of comparatively mild gradients, in contrast to the pronounced gradients in anomalous MSLP throughout much of the Pacific and northern North America and Asia. The MSLP changes thus appear to be consistent with the relatively weak changes in wintertime geostrophic winds described in section 4. [27] In the summer upwelling season, when climatological winds are primarily northwesterly under the influence of the North Pacific High (NPH) [cf. Bograd et al., 2002, Figure 1b], the projected changes implied by the multimodel ensemble mean consist of an increase in wind speed together with a clockwise rotation of the wind vector. The changes in relative to the base period are statistically significant and amount to 5 10% increases in wind speed and 5 clockwise rotations in wind direction. In addition the projected changes consistently amplify by a factor of 2 3 between and , suggesting a coherent pattern of change. By comparison, Wang et al. [2008] found that most of the climate models they considered show an increase in alongshore, upwelling favorable wind stress at 45 N along the Oregon coast in summer between and These changes appear consistent with a gross northward shift in the pattern of upwelling favorable winds along the North American west coast (J. Fyfe, personal communication, 2008). Such a shift in turn is consistent with modeled changes in JJA MSLP (Figure 8, bottom), which corresponds to a northward shift in the NPH: because summertime coastal marine winds are generally northeasterly equatorward of the NPH, northerly (and strongest) near the latitude of the NPH, and northwesterly poleward of the NPH, a northward shift in this pattern results in a clockwise (southward) shift in summertime Canadian west coast marine winds, consistent with the model results. [28] The projected changes described here may have significant implications for marine ecosystems along the southern British Columbia continental shelf where summer upwelling winds bring nutrients to the surface and drive upper trophic level productivity [Ware and Thomson, 2005]. However several other projected changes need to be considered before the complete picture is understood. Warmer air temperatures can be expected to heat the sea surface and contribute to a stronger stratification that will inhibit upwelling. Climate model analyses along the southern California shelf by Auad et al. [2006] showed that an increase in upwelling wind speed was sufficient to overcome stronger stratification. However, this result may not carry over to British Columbia where water column stratification is largely determined by salinity rather than temperature variations. Projected changes in precipitation and river discharge [Morrison et al., 2002], as well as the air-sea heat flux, will thus be needed to force a regional coastal ocean model in order to provide better estimates of physical changes to the coastal waters off British Columbia. A 15 km regional climate model [Salathé et al., 2008] that covers the southern half of British Columbia and provides a more accurate representation of the mountainous terrain than the global models should also prove to be a useful source of projections. Studies that evaluate this regional climate model against observations and carry out regional ocean model simulations will be reported in future contributions. Figure 8. Multimodel mean sea level pressure changes between and for (top) JFM and (bottom) JJA. Contour interval is 0.25 hpa, with hatching where more than 70% or 13 of the 18 models agree as to the sign of the change. 9of10

10 [29] Acknowledgments. We acknowledge the modeling groups, PCMDI, and the WCRP s Working Group on Coupled Modelling (WGCM) for their roles in making available the WCRP CMIP3 multimodel data set. Support of this data set is provided by the Office of Science, U.S. Department of Energy. B.P. was funded by the Climate Change Science and Ecosystem Research Initiatives of Fisheries and Oceans Canada. John Fyfe, Jim Christian, and two anonymous reviewers are thanked for suggesting improvements to the manuscript. References Auad, G., A. J. Miller, and E. Di Lorenzo (2006), Long-term forecast of oceanic conditions off California and their biological implications, J. Geophys. Res., 111, C09008, doi: /2005jc Bograd, S., F. Schwing, R. Mendelssohn, and P. Green-Jessen (2002), On the changing seasonality over the North Pacific, Geophys. Res. Lett., 29(9), 1333, doi: /2001gl Cherniawsky, J. Y., and W. R. Crawford (1996), Comparison between weather buoy and Comprehensive Ocean-Atmosphere Data Set wind data for the west coast of Canada, J. Geophys. Res., 101, 18,377 18,389. Faucher, M., W. R. Burrows, and L. Pandolfo (1999), Empirical-statistical reconstruction of surface marine winds along the western coast of Canada, Clim. Res., 11, Freeland, H. F., W. R. Crawford, and R. E. Thomson (1984), Currents along the Pacific coast of Canada, Atmos. Ocean, 22, Ianson, D., and S. E. Allen (2002), A two-dimensional nitrogen and carbon flux model in a coastal upwelling region, Global Biogeochem. Cycles, 16(1), 1011, doi: /2001gb Intergovernmental Panel on Climate Change (IPCC) (2007), Climate Change 2007: The Physical Science Basis Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, edited by S. Solomon et al., 996 pp., Cambridge Univ. Press, Cambridge, U. K. Lambert, S. J., and G. J. Boer (2001), CMIP1 evaluation and intercomparison of coupled climate models, Clim. Dyn., 17, Mantua, N. J., S. R. Hare, Y. Zhang, J. M. Wallace, and R. C. Francis (1997), A Pacific interdecadal climate oscillation with impacts on salmon production, Bull. Am. Meteorol. Soc., 78, Morrison, J., M. C. Quick, and M. G. G. Foreman (2002), Climate change in the Fraser watershed: Flow and temperature predictions, J. Hydrol., 263, Murphree, T., P. Green-Jessen, F. B. Schwing, and S. J. Bograd (2003), The seasonal cycle of wind stress curl and its relationship to subsurface ocean temperature in the Northeast Pacific, Geophys. Res. Lett., 30(9), 1469, doi: /2002gl Overland, J. E., and M. Wang (2007), Future climate of the North Pacific Ocean, Eos Trans. AGU, 88, 178, doi: /2007eo Peterson, W. T., and F. B. Schwing (2003), A new climate regime in northeast pacific ecosystems, Geophys. Res. Lett., 30(17), 1896, doi: / 2003GL Salathé, E. P., Jr. (2006), Influences of a shift in North Pacific storm tracks on western North American precipitation under global warming, Geophys. Res. Lett., 33, L19820, doi: /2006gl Salathé, E. P., Jr., R. Steed, C. F. Mass, and P. H. Zahn (2008), A highresolution climate model for the United States Pacific Northwest, part II: Mesoscale feedbacks and local responses to climate change, J. Clim., 21, Wang, M., J. E. Overland, and N. A. Bond (2008), Climate projections for selected large marine ecosystems, J. Mar. Syst., doi: / j.marsys , in press. Ware, D. M., and R. E. Thomson (2005), Bottom-up ecosystem dynamimcs determine fish production in the Northeast Pacific, Science, 308, Yin, J. H. (2005), A consistent poleward shift of the storm tracks in simulations of 21st century climate, Geophys. Res. Lett., 32, L18701, doi: /2005gl Zhang, Y., J. M. Wallace, and D. S. Battisti (1997), ENSO-like interdecadal variability: , J. Clim., 10, M. G. G. Foreman and B. Pal, Institute of Ocean Sciences, 9860 West Saanich Road, P.O. Box 6000, Sidney, BC V8L 4B2, Canada. W. J. Merryfield, Canadian Centre for Climate Modelling and Analysis, University of Victoria, P.O. Box 3065 STN CSC, Victoria, BC V8W 3V6, Canada. (bill.merryfield@ec.gc.ca) 10 of 10