High-wind climate changes inferred from CMIP5 model runs

High-wind climate changes inferred from CMIP5 model runs Andreas Sterl,1 Renske de Winter2 Several papers investigate wind changes in the north-east A...
Author: Adam Green
4 downloads 4 Views 5MB Size
High-wind climate changes inferred from CMIP5 model runs Andreas Sterl,1 Renske de Winter2 Several papers investigate wind changes in the north-east Atlantic and European regions. Observational studies find large variability on decadal time scales [e.g., Alexandersson et al., 2000; Wang et al., 2009, 2011], but only small trends. Under climate change, trends are also small and regionally confined [e.g., Debernard, Røed, 2008; Kjelstr¨ om et al., 2011; Nikulin et al., 2011]. Position and sign of the trends are strongly model-dependent [Kjelstr¨ om et al., 2011, their Figures 12+13; Nikulin et al., 2011, their Figure 8]. In this paper we investigate whether new results from CMIP5 (Coupled Model Intercomparison Project, phase 5; Taylor et al. [2012]) give rise to a modification of these findings.

The CMIP5 multi-model dataset is investigated for changes in wind climate, especially in high-wind conditions. For twelve models daily-mean winds were available when starting this research. For these models we investigate changes of the annual-maximum daily-mean wind speed between 2051-2100 under the rcp85 forcing and the hisorical period 1951-2000. Results differ widely between the models, with some showing increases and others decreases in the same region. However, and most importantly, the detected changes are often not significant at the 95% level. The ensemble-mean change displays a general weakening with the exception of the Southern Ocean and tropical land areas. Our results do not back fears of a worsening wind climate. However, one caveat is that all models employed are of relatively coarse resolution (roughly between 1◦ and 3◦ ) and thus do not resolve tropical cyclones. Changes in the number or intensity of tropical cyclones may lead to different results.

2. Models used We use results from twelve CMIP5 models that had wind data (10-m wind, U10 ) with daily-mean resolution available at the time we produced this paper. These models are listed in Table 1. Their horizontal resolution varies roughly between 1◦ and 3◦ . For our own EC-Earth model [Hazeleger et al., 2012; Sterl et al., 2012] we have 3-hourly data available. We use them to investigate the impact of time resolution on the results.

1. Introduction Wind, and especially strong wind, is an important parameter for safety considerations. On sea it affects the safety of ships, marine constructions, or the coastline, either directly or via the generation of waves and storm surges. Over land it can cause destruction of buildings or forests. Its potential change in a changing climate is therefore of special interest. As wind damage is mainly caused by sustained high wind speeds we here focus on the highest daily-mean winds. When investigating storms a distinction between tropical and extratropical storms has to be made. They have different development mechanisms and spatial scales. Tropical storms gain their energy from latent heat release. Moist air is transported upward in a very narrow channel (the “eye”) that cannot be resolved by present-day climate models. Extratropical storms are generated by baroclinic instabilities and are well-resolved in climate models. As this paper uses output from climate models it is mainly concerned with extratropical storms. Investigating changes in storm climate under a SRES A1b climate change scenario, Bengtsson et al. [2006] find “no indications in this study of more intense storms in the future climate, either in the Tropics or extratropics”, but regional changes in intensity and location of the storm tracks. Especially, they find a poleward shift of the main extratropical storm tracks. These findings are confirmed by Bengtsson et al. [2009].

3. Time resolution 3.1. Hourly to daily time scales We are interested in possible changes of the most extreme wind speeds. Naturally, wind speeds averaged over a short period can reach higher values than those averaged over a longer period. Except for EC-Earth we only have daily-mean output. Using EC-Earth output at 3-hourly and daily resolution we now investigate whether trends over the 21st century differ between the two resolutions. This is done in Figure 1, where we compare the linear trends amx of annual-maximum 3-hourly (U10 ) and annual-maximum amx(dm) daily-mean (U10 ) winds, respectively. Clearly, trends in both cases hardly differ, and we can safely assume that trends for both time resolutions are equivalent and that it is suffient to investigate the annual maxima of daily means. Also clear from Figure 1 is that the trends are only significant in small areas which are irregularly distributed over the Globe. Taking into account the fact that when testing for 95% significance one expects 5% of the points to prove significant just by chance, and that this fraction increases when taking field significance into account [Livezey and Chen, 1983; Sterl et al., 2007], the results presented in Figure 1 are compatible with the statement that no significant changes in extreme wind speeds occur in the EC-Earth rcp85 run. An exception is the Arctic Ocean. Here vanishing sea ice leads to strong air-sea interaction that destroys the stably stratified boundary layer [Bintanja et al., 2011, their Fig. 7], so that the upper-level winds can reach the surface.

1 Royal Netherlands Meteorological Institute (KNMI), De Bilt, Netherlands 2 Royal Netherlands Meteorological Institute (KNMI), De Bilt, Netherlands, and Physical Geography Research Institute, Utrecht University, Netherlands

3.2. Monthly time scales In Figure 2 we show linear trends of monthly-mean wind speeds for four months from the same EC-Earth run as before. For the monthly means the change patterns appear to

Copyright 2012 by the American Geophysical Union. 0094-8276/12/$5.00

1

be more organized than those for the shorter time scales. Changes occur in both directions, i.e., increasing and decreasing. The patterns are compatible with the poleward shift of the main circulation zones that has been noted before [e.g., Bengtsson et al., 2006]. They show no signs of a generally worsening wind climate.

4. Multi-model ensemble 4.1. Comparison with ERA-Interim amx(dm)

To assess the quality of the U10 values as simulated by the twelve CMIP5 models we compare them with results from ERA-Interim. As observed 10-m winds have been assimilated in ERA-Interim [Dee et al., 2011] we assume that they are close to observations. Figure 3 depicts the interamx(dm) model standard deviation of U10 , showing that modamx(dm) eled U10 differ widely between models. The largest differences occur near topographic features like the Andes or the Himalayas. The large inter-model variation around Antarctica and in the Arctic is mainly caused by one model (MRI-CGCM3) that exhibits a large deviation from ERAInterim in these regions (see Figure S1). Supporting Figure S1 shows that the models both under- and overestimate the ERA-Interim values. There is no simple relation between the horizontal resolution of a model (Table 1) and its bias. For instance, CanEM2 and IPSL-CMA5-MR have comparable resolutions (2.8◦ and 2.5◦ , respectively), but they represent the two extremes in bias, with CanEM2 overstimating amx(dm) U10 by several m/s and IPSL-CMA5-MR underestimating by a comparable amount. On the other hand, ECEarth and the two GFDL models exhibit the closest correspondence to ERA-Interim, but differ in resolution by a factor of 2. 4.2. Simulated change amx(dm)

We now investigate the change in mean U10 values between the periods 2051-2100 (rcp85 forcing) and 19512000. The mean of the twelve model differences is shown in Figure 4, and the results for the individual models are given in Figure S2 (supporting material). For rcp45 forcing the patterns are similar, but with smaller amplitudes and therefore smaller areas of significance (not shown), indicating that the strength of the change scales with the forcing. The results differ considerably between the models (Figure S2), especially when significance is concerned. For some of the models several members are available. Change patterns agree between members (not shown), indicating that they reflect robust model physics. The ensemble-mean change amx(dm) (Figure 4) shows a reduction of U10 nearly everywhere, with the exception of large parts of the Southern Ocean and some tropical land areas. For the North Atlantic none of the models simulates a significant increase of wind speeds, and the ensemble-mean exhibits a significant decrease for most parts of the basin. Figure S2 shows that there is no simple relation between a models’ resolution and the simulated change. Relatively high-resolution models like EC-Earth and MIROC5 show quite different change patterns, while EC-Earth and the two relatively low-resolution GFDL models exhibit a close agreement. A comparison between the two supporting figures shows that there is also no correspondence between the simulated change patterns and the closeness to the ERA-Interim values.

5. Summary and Conclusion We have looked for changes in extreme wind speeds (annual maximum of daily means) in twelve CMIP5 models.

The models do not agree on location or size of changes in this variable. The ensemble-mean displays a decrease over most areas, but an increase in the Southern Hemisphere storm track, some tropical land areas as well as the Arctic Ocean. For the North Atlantic none of the models simulates a significant increase of wind speeds, and the ensemble-mean exhibits a decrease that is significant in most parts of the basin. Our results do not support any fears of a worsening wind climate. The simulated patterns of change do not depend on the horizontal resolution of the model used. However, all models used have a relatively coarse resolution of roughly between 1◦ and 3◦ . Therefore none of them can resolve tropical cyclones. There is a consensus in the existing literature (see the review article by Knutson et al. [2010] and references therein) that tropical cyclones will get more intense in a warmer climate and will develop over larger regions. In a high-resolution EC-Earth run Haarsma et al. (in preparation) find that in a warmer climate hurricanes in the North Atlantic may affect maximum wind speeds as far northward as 60◦ N. In this paper we only deal with wind speed, excluding changes in wind direction from our analysis. However, changes in wind direction, even if not accompanied by changes in wind speed, may have a notable effect on the height of waves or storm surges as the wind fetch (distance along which the wind can affect the sea) changes. In a second paper (de Winter et al., in preparation) we provide a more detailed analysis of wind changes, including directions, for the North Sea. Acknowledgments. We acknowledge the World Climate Research Programme’s Working Group on Coupled Modelling, which is responsible for CMIP, and we thank the climate modeling groups (listed in Table 1 of this paper) for producing and making available their model output. For CMIP the U.S. Department of Energy’s Program for Climate Model Diagnosis and Intercomparison provides coordinating support and led development of software infrastructure in partnership with the Global Organization for Earth System Science Portals.

References Alexandersson, H., H. Tuomenvirta, T. Schmidth, and K. Iden (2000), Trends of storms in NW Europe derived from an updated pressure data set, Climate Res., 14, 7173. Bengtsson, L., K.I. Hodges, and E. Roeckner (2006), Storm tracks and climate change, J. Clim., 19, 3518-3543. Bengtsson, L., K.I. Hodges, and N. Keenlyside (2009), Will extratropical storms intensify in a warmer climate? J. Clim., 22, 2276-2301. Bintanja, R., E.C. van der Linden, and W. Hazeleger (2011), Boundary layer stability and Arctic climate change: a feedback study using EC-Earth Clim. Dyn., online first, doi: 10.1007/s00382-011-1272-1. Debernard, J.B. and L.P. Røed (2008), Future wind, wave and storm surge climate in the Northern Seas: a revisit, Tellus, 60A, 427-438, doi: 10.1111/j.1600-0870.2008.00312.x. Dee, D.P., S.M. Uppala, A.J. Simmons, P. Berrisford, P. Poli, S. Kobayashi, U. Andrae, M.A. Balmaseda, G. Balsamo, P. Bauer, P. Bechtold, A.C.M. Beljaars, L. van de Berg, J. Bidlot, N. Bormann, C. Delsol, R. Dragani, M. Fuentes, A.J. Geer, L. Haimberger, S.B. Healy, H. Hersbach, E.V. H´ olm, L. Isaksen, P. K˚ allberg, M. K¨ ohler, M. Matricardi, A.P. McNally, B.M. Monge-Sanz, J.-J. Morcrette, B.-K. Park, C. Peubey, P. de Rosnay, C. Tavolato, J.-N. Th´ epaut, and F. Vitart (2011), The ERA-Interim reanalysis: configuration and performance of the data assimilation system, Q.J.R. Meteorol. Soc., 137, 553597, doi: 10.1002/qj.828.

Knutson, T.R., J.L. McBride, J. Chan, K. Emanuel, G. Holland, C. Landsea, I. Held, J.P. Kossin, A.K. Srivastave, and M. Sugi (2010), Tropical cyclones and climate change, Nature Geosci., 3, 157-163, doi: 10.1038/NGEO779. Hazeleger, W., X. Wang, C. Severijns, S. S ¸ tefˇ anescu, R. Bintanja, A. Sterl, K. Wyser, T. Semmler, S. Yang, B. van den Hurk, T. van Noije, E. van der Linden, and K. van den Wiel (2012), EC-Earth V2: description and validation of a new seamless Earth system prediction model, Clim. Dyn., online first, doi: 10.1007/s00382-011-1228-5. Kjelstr¨ om, E., G. Nikulin, U. Hansson, G. Strandberg, and A. Ullerstig (2011), 21st century changes in the European climate: uncertainties derived from an ensemble of regional climate model simulations, Tellus A, 63, 24-40, doi: 10.1111/j.16000870.2010.00475.x. Livezey, R.E., and W.Y. Chen (1983), Statistical field significance and its determination by Monte Carlo techniques, Mon. Weath. Rev., 111, 46-59. Nikulin, G., Kjelstr¨ om, E., U. Hansson, G. Strandberg, and A. Ullerstig (2011), Evaluation and future projections of temperature, precipitation and wind extremes over Europe in an ensemble of regional climate simulations, Tellus A, 63, 41-55, doi: 10.1111/j.1600-0870.2010.00466.x.

Sterl, A., G.J. van Oldenborgh, W. Hazeleger, and G. Burgers (2007), On the robustness of ENSO teleconnections, Clim. Dyn., 29, 469-485, doi: 10.1007/s00382-007-0251-z. Sterl, A., R. Bintanja, L. Brodeau, E. Gleeson, T. Koenigck, T. Schmith, T. Semmler, C. Severijns, K. Wyser, and S. Yang (2012), A look at the ocean in the EC-Earth climate model, Clim. Dyn., online first, doi: 10.1007/s00382-011-1239-2. Taylor, K.E., R.J. Stouffer, G.A. Meehl (2012), An Overview of CMIP5 and the experiment design, Bull. Amer. Meteor. Soc., 93, 485-498, doi:10.1175/BAMS-D-11-00094.1. Wang, X.L., F.W. Zwiers, V.R. Swail, and Y. Feng (2009), Trends and variability of storminess in the Northeast Atlantic region, 1874-2007, Clim. Dyn., 33, 1179-1195, doi: 10.1007/s00382008-0504-5. Wang, X.L., H. Wan, F.W. Zwiers, V.R. Swail, G.P. Compo, R.J. Allen, R.S. Vose, S. Jourdain, and X. Yin (2011), Trends and low-frequency storminess over western Europe, 1878-2007, Clim. Dyn., 37, 2355-2371, doi: 10.1007/s00382-011-1107-0. Andreas Sterl ([email protected]), Renske de Winter, Royal Netherlands Institute of Meteorology (KNMI), P.O. Box 201, NL3730 AE De Bilt, The Netherlands.

a)

b)

amx Figure 1. Linear trends (in (m/s)/10ys) of U10 (a) amx(dm) and U10 (b) of the EC-Earth rcp85 run (2006-2100). Areas in which the trends are significantly different from zero (95% according to a t-test) are denoted by saturated colors enclosed by a black contour. The green contours denote the continents.

Figure 2. Linear trends (in (m/s)/10ys) of monthlymeans for January, April, July and October as indicated in the plot of the EC-Earth rcp85 run (2006-2100). Areas in which the trends are significantly different from zero (95% according to a t-test) are denoted by saturated colors enclosed by a black contour. The green contours denote the continents.

Figure 3. Inter-model standard deviation of mean amx(dm) U10 (in m/s) for the 1951-2000 period. Note that this is identical to the inter-model standard deviation of the bias wrt. ERA-Interim.

amx(dm)

Figure 4. Ensemble-mean difference of U10 (in m/s) between 2051-2100 (rcp85) and 1951-2000. Areas in which the difference is significantly different from zero (95% according to a t-test performed on the 12 individual trends) are denoted by saturated colors enclosed by a black contour. The green contours denote the continents.

Model

longitude latitude # pts ∆λ # pts ∆ϕ CanESM2 128 2.8125 64 2.8125 192 1.875 96 1.875 CSIRO-Mk3-6-0 320 1.125 160 1.125 EC-Earth GFDL-ESM2G 144 2.5 90 2.0 144 2.5 90 2.0 GFDL-ESM2M 192 1.875 144 1.25 HadGEM2-CC HadGEM2-ES 192 1.875 144 1.25 144 2.5 143 1.25 IPSL-CM5A-MR 256 1.40625 128 1.41 MIROC5 128 2.8125 64 2.8125 MIROC-ESM-CHEM MPI-ESM-LR 192 1.875 96 1.875 320 1.125 160 1.125 MRI-CGCM3 Table 1. Models used and their horizontal resolution. ∆λ and ∆ϕ are the increments in longitudinal and meridional direction, repspectively, in degrees. The increments in latitudinal direction are averages as some models use non-constant (e.g., Gaussian) latitudes.

Supplementary Material High-wind climate changes inferred from CMIP5 model runs Andreas Sterl(*), Renske de Winter(*) Royal Netherlands Meteorological Institute (KNMI), De Bilt, Netherlands (**) Also at Physical Geography Research Institute, Utrecht University June 21, 2012

The two supporting figures display results from the twelve CMP5 models used in this study. They supplement Figures 3 and 4 which give an ensemble-mean view of the same results.

amx(dm)

Figure S 1. Difference of mean U10 (in m/s) between each of the 12 CMIP5 models from Table 1 and ERA-Interim. Differences significantly different from zero (95% according to a t-test) are denoted by saturated colors enclosed by a black contour. The green contours denote the continents.

amx(dm)

Figure S 2. Difference of U10 (in m/s) between 2051-2100 (rcp85) and 1951-2000 for the 12 CMIP5 models from Table 1. Areas in which the differences are significantly different from zero (95% according to a t-test) are denoted by saturated colors enclosed by a black contour. The green contours denote the continents.

Suggest Documents