Quantifying projected impacts under 2°C warming
Download 62.35 Kb.
|
- Navigate this page:
- 1. Introduction
- 2. Description of the experiments
- 3. Band-pass variance
- 4. Strong winds
- 5. Conclusion
- 6 References
 
282746 IMPACT2C Quantifying projected impacts under 2°C warming
Report on the role of global modeling on storm changes over Europe
October 2014 
Executive Summary
Index Quantifying projected impacts under 2°C warming 1 1. Introduction 4 2. Description of the experiments 4 3. Band-pass variance 5 4. Strong winds 9 5. Conclusion 16 6 References 17
1. IntroductionMost WP2 results are obtained from the EuroCORDEX initiative. It consists of regional climate simulations at 12 km resolution over a common domain. The EuroCORDEX domain is wide enough to ensure a relative independence of the climate in Europe of the lateral boundary forcing from the CMIP5 GCMs: is covers from Greenland to Red Sea, including the whole European continent and the whole Mediterranean basin. Thus, the large-scale warming is imposed by the GCM, but a large fraction of precipitation, the soil moisture evolution and the regional distribution of the warming are little dependent on the lateral conditions. Sea surface temperature in the eastern Atlantic, North Sea, Baltic Sea, and Mediterranean Sea play a role in precipitation and are probably the major contribution of the GCM to the regional climate response. The winter storms are however potentially impacted by the western lateral boundary. Indeed, the atmospheric perturbation start they life outside of the domain, above the warm Atlantic waters in the western part of the basin. If the climate response of the strength and location of those systems is poorly treated because of the low resolution of the GCM or the numerical tricks in the lateral sponge layer, the high resolution description of the storms in EuroCORDEX may be altered. To investigate the possible drawbacks in EuroCORDEX, the MeteoF contribution to EuroCORDEX has been completed by global high resolution (12 km) simulations with a version of the atmospheric model. The aim is to compare CMIP5, EuroCORDEX and a high resolution experiment without lateral constraint, as far as winter Atlantic storms is concerned, in the framework of a +2°C warming with respect to pre-industrial. A simple way to identify Atlantic storms is to count the surface pressure minima day by day in the simulation. This method is very difficult to automatize in long simulations. It has been early given up. Blackmon et al. (1984) proposed to consider the band-pass filtered variance of 500 hPa geopotential height to identify the mean storm track. We will use this simple and efficient method in this study. A more sophisticated method would be to identify the trajectory of the individual storms by an automatic tracking procedure (Bengtsson at al., 2006). The impact of the winter storms over Europe are sea level surges, heavy precipitation, and strong winds. We will only examine this latter point here. Storm surges require specific modelling. Precipitation has been widely studied in WP2, and winter precipitation over Europe is very model dependent. Section 2 describes the experiments used or specifically run for the study, section 3 examines the storm track intensity and location with 500 hPa height variance, section 4 shows the response in intensity and frequency of strong winds, and section 5 summarizes the results. 2. Description of the experimentsThe analysis framework is clearly defined. We have a reference period (1971-2000), a +2°C period (depending on the scenario), a +1.5°C period (idem) and a stabilization period (2071-2100 in a stabilized concentration scenario). We consider here CNRM-CM5 (Voldoire et al., 2013) contribution to CMIP5 with 3 scenarios RCP8.5, RCP4.5 and RCP2.6. In RCP8.5 we will just consider the +2°C period (2031-2060) as the +1.5°C period is too close to it. In RCP4.5, we consider the +1.5°C period (2023-2052) and the +2°C period (2044-2073). In RCP2.6, we consider the 2071-2100 period, which corresponds to a +1.8°C global warming. The periods will be named respectively S85, S45a, S45b and S26, the reference being named REF. The CNRM-CM5 model will be referred to as CMIP, whereas the EuroCORDEX contribution will be referred to as EUC. The atmospheric component of CMIP is ARPEGE (Déqué, 2010) and EuroCORDEX is run with the limited area model ALADIN (Colin et al., 2010). In fact these two models share the same code, the same equations and the same physical parameterizations. They differ only by the spatial discretization, and the presence of a lateral boundary relaxation in ALADIN. However ARPEGE can also be driven like an EuroCORDEX standard limited area model, by introducing a 6-hour e-folding time nudging outside the EuroCORDEX domain for temperature, moisture and velocity. We use a configuration of ARPEGE with variable resolution: 12 km over Europe, 12 to 25 km in the North Atlantic, 160 km in the South Pacific (this maximum resolution is the resolution of ARPEGE in CNRM-CM5). Two simulations are considered, one driven (named DHR) and one without nudging (named FHR). We have thus 5 periods and 4 models, which yields 20 30-year simulations. The uncoupled simulations (EUC, DHR and FHR) use the same sea surface temperature as the coupled one (CMIP). All model use the same physics. FHR differs from CMIP by the coupling and the resolution. DHR differs from FHR by the lateral driving outside Europe. EUC differs from DHR by the discretization grid (similar resolution, but one is Lambert and the other is polar stereographic) and the spectral technique (one is bi-Fourier, the other is Legendre). We have thus a cascade of models between CMIP and EuroCORDEX. Each step helps us to identify the impact of the modification. In particular, the role of lateral condition can be cleanly analyzed by the comparison DHR versus FHR.
Figure 1 shows the standard deviation in the reference period for the 4 models, and for ERA-interim. The model underestimates the north-eastward extension of the storm tracks, the free high resolution model, being slighly better than the constrained or the low resolution versions. Figure 2 shows the difference between the standard deviations in the S85 scenario and in the reference. The response of the CMIP model is an increase in variance over Europe. The response of FHR is more standard with respect to Bengtsson et al. (2006), with a northward shift of the storm track. The two driven models DHR and EUC are more in agreement with FHR than with CMIP, with a weaker signal. This shows that the model resolution plays a greater role than the lateral forcing, which is a good news for Eurocordex. Figure 3 is the equivalent of Figure 2 for the RCP4.5 scenario. As far as the free simulations are concerned (CMIP and FHR), we have the same patterns as in the RCP8.5 scenario, with some differences in the amplitude. The two forced models are still in better agreement with FHR than with CMIP, but now, the response is enhanced, whereas it was reduced in S45. Figure 4 has some parenthood with Figure 3, because we consider the same simulations earlier, at +1.5°C stage. We have here a weak and probably little significant signal, because CMIP response is larger here than 20 years later. Figure 5 exhibits also a weak signal, with a great similarity between the two driven simulations.
4. Strong windsThe winter storms are associated with strong winds (e.g. Lothar and Martin in December 1999). In hydrostatic weather forecast models, strong near surface winds are underestimated for numerical stability reasons. To identify severe storms, forecasters used altitude wind during the past century. Nowadays, models compute “gust winds” with an appropriate parameterization. Here we will use the former approach, as it is less model dependent. Here again, we analyse, for our 20 30-year time series, the November to March period. A first question is the intensity of the most severe storms. Will it increase in our scenarios ? To answer this, a simple, but not highly robust, diagnostic is the maximum wind in each season, averaged over the 30 years. Storms are large-scale systems, so using 500 hPa geostrophic wind is not too unreallistic, except that the velocity is larger at circa 5000m height than near the surface. But as we consider a relative change with respect to a reference climate, the drawback is minor. Figure 6 shows the maximum wind in the reference period. ERA-interim data yield about 50 m/s over Europe. The order of magnitude of the maximum surface wind in Europe (for an average year) is 30 m/s that 500 hPa value is a little less than twice the surface value. The various versions of the models underestimate the strong winds in the eastern Atlantic as they have underestimated the synoptic variance. Here again, the free high resolution FHR is the closest to reality. Figures 7 to 10 show the relative change of the maximum wind in percentage in the 4 scenarios, with respect to the reference value. Since the seasonal maximum is far from a Gaussian variable, the average of 30 independent values poorly obey the rule of great numbers, an the difference is affected by sampling error. When dividing by the reference value, the noise is enhanced. Let us focus on Europe in these noisy maps to look for some information if any. In S85 (fig. 7) CMIP increases the maximum wind, FHR decreases it, and the two driven simulations increase in the North, decrease in the South. In S45b (fig. 8) CMIP increases the maximum wind in mid-Europe but decreases it in Scandinavia, FHR has a similar behavior, but the two driven simulations show an opposite behavior (increase the in North Europe and decrease in mid Europe). The two weaker scenarios S45a and S26 show a weaker signal over Europe, but without consistency between the models. The only stable feature we have is a similar response in the two driven simulations. For the rest of the story, we cannot conclude about an increase or a decrease in the maximum wind.
A second question is the frequency of the severe storms. It is possible that in a warmer climate the strongest storms are not stronger than in the reference climate, but are more frequent. A simple index consists of counting the days with wind above a fixed threshold. As we work at 500 hPa and not at the surface, a reasonable threshold, in terms of statistical robustness, is 40 m/s. Figure 11 shows the number of “storms” per Winter. In ERA-interim this threshold allows a large number of events, in particular in the West Atlantic, but we are interested in Europe and in model simulations (our only insight into a +2°C climate), so 40 m/s is adapted. Another solution would be to use a local wind threshold, e.g. a quantile. Let us do simple ! The events we count here occur about 3-4 times per extended Winter (NDJFM) in the model over Europe and 4-6 times in the real world. These are not centennial storms, which would require thousands of years of simulation for a robust frequency estimation. As far as climate change is concerned (figures 12 to 15) the spatial patterns are as noisy as those of the maximum winds. But here the relative changes reach ±30% instead of ±5%. One relatively stable feature amongst the figures is the decrease over Scandinavia with the free high resolution. Another stable feature, unfortunately in disagreament with the first one, is the increase in the frequency of events over North Europe with the two driven models
5. ConclusionCaveat emptor ! We do not claim here to describe the response of storms over Europe, because a single model is considered. Our purpose was to investigate the role of the lateral boundary forcing in the model response, given that the physical processes originate from outside of the EuroCORDEX domain. As far as the 2-6 day 500 hPa height variance is concerned, introducing a lateral constraint modifies the storm track response compared. to the original CMIP simulation. But this modification is welcome, because the new response is closer to the response of a global model with the same resolution as EuroCORDEX (at least in the EuroCORDEX domain). Another good news is the excellent agreement between the original EuroCORDEX simulation with ALADIN strongly driven (10 min characteristic time) in a narrow band (about 100 km wide) around the domain, and the exotic EuroCORDEX with ARPEGE, a global variable resolution model softly driven (6 hour characterisic time) over a large part of the globe. The maximum wind in a given season is a variable difficult to estimate, and 30 years are insufficient to establish whether this maximum will increase or decrease in a warmer climate; We have considered here very large-scale phenomena with altitude geostrophic wind. One can imagine how difficult it could be with surface winds at 12 km resolution with local orographic effects. The number of strong wind events is a more stable parameter, because it is based on several events per year instead of a single. We can see here a clearer response with an increase by about 20% of the frequency over northern Europe. This is consistent with the northward displacement of the storm tracks. But, over Scandinavia, this is contradicted by the decrease we get with the free high resolution model. So we should apply the same caution as with the storm intensity. 6 ReferencesBengtsson, L., K.I. Hodges and E. Roeckner, 2006: Storm Tracks and Climate Change. J. Climate, 19, 3518–3543 Blackmon, M.L., Y.H. Lee and J.M. Wallace, 1984: Horizontal structure of 500 mb height fluctuations with long, intermediate and short time scales. J. Atmos. Sci., 41, 961-979 Colin, J., M. Déqué, R. Radu, and S. Somot, 2010: Sensitivity study of heavy precipitations in Limited Area Model climate simulation: influence of the size of the domain and the use of the spectral nudging technique. Tellus A, 62, 591-604, Déqué, M., 2010 : Regional climate simulation with a mosaic of RCMs. Meteorologische Zeitschrift, 19, 259-266 Doblas-Reyes F.J. and M. Déqué, 1998: A flexible bandpass filter design procedure applied to midlatitude intraseasonal variability. Mon Wea Rev 126, 3326-3335 Voldoire, A. E. Sanchez-Gomez, D. Salas y Mélia, B. Decharme, C. Cassou, S.Sénési, S. Valcke, I. Beau, A. Alias, M. Chevallier, M. Déqué, J. Deshayes, H. Douville, E. Fernandez, G. Madec, E. Maisonnave , M.-P. Moine, S. Planton, D. Saint-Martin, S. Szopa, S. Tyteca, R. Alkama, S. Belamari, A. Braun, L. Coquart and F. Chauvin, 2013. The CNRM-CM5.1 global climate model : description and basic evaluation , Clim. Dyn., 40(9-10):2091-2121 Directory: data data -> Iea shcp/pvps working Group on pv/t solar Systems data -> Jaeyong song data -> By Fel (aka James Galloway) Table of Contents Chapter 1 data -> Vitaly Goldshteyn data -> Hurricane Katrina, a climatological Perspective October 2005, Updated August 2006 Introduction data -> 1997 Atlantic Tropical Storms data -> Introduction data -> The Impact of Saharan dust aerosols on tropical cyclones using wrf-chem: Model framework and satellite data constraint technique data -> Earth-Atmosphere Interactions: Tropical Storm and Hurricane Activity in the Caribbean and Consequent Health Impacts data -> Meeting Of September 16, 2010 today’s meeting Download 62.35 Kb. Share with your friends:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
