S2S / TIGGE Workshop ECMWF April 219 Stratospheric influences on subseasonal predictability of European energy-industry-relevant parameters Dominik Büeler 1,2 / Remo Beerli 3,2, Heini Wernli 2, Christian M. Grams 1,2 1) Institute of Meteorology and Climate Research, Department Troposphere Research, KIT, Germany 2) Institute for Atmospheric and Climate Science, ETH Zurich, Switzerland 3) AXPO Solutions AG, Switzerland e t i b o t ast fails 9 E e h 21 h c t r a M m 8 o st fr arket Analyst, 1 a e B e Th as M roley, G Alex F KIT The Research University in the Helmholtz Association www.kit.edu
Motivation Polar vortex weather regimes wind power REPORTS A 1998-1999 Northern Annular Mode.3 Weak Vortex Regimes e.g., Baldwin & Dunkerton, 21, SCI; Tripathi et al., 215, ERL; Charlton-Perez et al., 218, QJRMS 1 5 hpa 4 km Nov Dec Jan Feb Mar A 2 - Apr Fig. 1. Time-height development of the northern annular mode during the winter of 1998 1999. The indices have daily resolution and are nondimensional. Blue corresponds to positive values (strong polar vortex), and red corresponds to negative values (weak polar vortex). The contour interval is.5, with values between!.5 and.5 unshaded. The thin horizontal line indicates the approximate boundary between the troposphere and the stratosphere. -1 B Strong Vortex Regimes Composite of 18 Weak Vortex Events hpa 2.7.8 Between ERAI NAO and ERAI m wind speed Between ERAI NAO and ERAI 1.5m air temperature Between G5 NAO and G5 m wind speed (48 samples) Between G5 NAO and G5 1.5m air temperature (48 samples) Between G5 EM NAO and G5 EM m wind speed (2 samples) Between G5 EM NAO and G5 EM 1.5m air temperature (2 samples) km -9-6 B - Lag (Days) 6 9 2 Composite of Strong Vortex Events hpa 2-1 km -9-6 - Lag (Days) 6 9 Fig. 2. Composites of time-height development of the northern annular mode for (A) 18 weak vortex events and (B) strong vortex events. The events are determined by the dates on which the -hpa annular mode values cross 3. and "1.5, respectively. The indices are nondimensional; the contour interval for the color shading is.25, and.5 for the white contours. Values between!.25 and.25 are unshaded. The thin horizontal lines indicate the approximate boundary between the troposphere and the stratosphere. Stratospheric and tropospheric annular mode variations are sometimes independent of each other, but (on average) strong anomalies just above the tropopause appear to favor tropospheric anomalies of the same sign. Opposing anomalies as in December 1998 (Fig. 1) are possible, but anomalies of the same sign dominate the average (Fig. 2). To examine the tropospheric circulation after these extreme events, we define weak and strong vortex regimes as the 6-day periods after the dates on which the!3. and "1.5 thresholds were crossed. Our results are not sensitive to the exact range of days used and do not depend on the first few days after the events. We focus on the average behavior during these weak vortex regimes and 582 4 Oct.6 e.g., Clark et al., 217, ERL; Brayshaw et al., 211, RE.5 Figure 1. Correlation between GloSea5 ensemble mean and ERA Interim sea level pressure for DJF compiled from 2 yea simulation. Mask (white areas) applied to correlations not significantly greater than zero at % level. 2.4 strong vortex regimes, as characterized by the normalized AO index (22). The average value (8 days) during weak vortex regimes is!.44, and ".35 for strong vortex regimes (18 days). The large sample sizes contribute to the high statistical significance of these averages (23). During the weak and strong vortex regimes the average surface pressure anomalies (Fig. 3) are markedly like opposite phases of the AO (11) or NAO (14), with the largest effect on pressure gradients in the North Atlantic and Northern Europe. The probability density functions (PDFs) of the daily normalized AO and NAO indices (24) during weak and strong vortex regimes are compared in Fig. 4. More pronounced than the shift in means are differences in the shapes of Fig. 3. Average sea-level pressure anomalies (hpa) for (A) the 8 days during weak vortex regimes and (B) the 18 days during strong vortex regimes. the PDFs, especially between the tails of the curves. Values of AO or NAO index greater than 1. are three to four times as likely during strong vortex regimes than weak vortex regimes. Similarly, index values less than!1. are three to four times as likely during weak vortex regimes than strong vortex regimes. Values of the daily AO index greater than 1. and less than!1. are associated with statistically significant changes in the probabilities of weather extremes such as cold air outbreaks, snow, and high winds across Europe, Asia, and North America (25). The observed circulation changes during weak and strong vortex regimes are substantial from a meteorological viewpoint and can be anticipated by observing the stratosphere. These results imply a measure of predictability, up to 2 months in advance, for AO/ NAO variations in northern winter, particularly for extreme values that are associated with unusual weather events having the greatest impact on society. Since the NAO and AO are known to modulate the position of surface cyclones across the Atlantic and Europe, we examine the tracks of surface cyclones with central pressure less than 19 OCTOBER 21 VOL 294 SCIENCE www.sciencemag.org NOAA NOAA EnergyWay State of the stratospheric polar vortex (SPV) as a direct source of subseasonal predictability for European energy industry?.8.6.4.2..2.4.6.8 Figure 2. Correlations between NAO on winter near-surface wind speed (left column) and temperature (right column). Ob (ERA Interim) relationships are shown in the top row. Middle row shows ensemble member relationships in hindcasts. Botto shows ensemble mean relationships in hindcasts. Mask (white areas) applied to correlations not significant at % level. 2 2 April 219 ECMWF the wind speed at m (in m s!1), following the Seasonal means of power density were produ Meteorology Climate Research (IMK-TRO) gridpoint by averaging over power d approach of Manwell et alinstitute (2) inof which wind powerandeach is primarily a function of the volume throughput of air computed using the daily-mean output fr driving the blades of a turbine. r is the air density, GloSea5 hindcasts and 6-hourly means from
Data Statistical forecast Strength of SPV (Δ Z@15hPa) 6-9 N from ERA-Interim Daily, DJF, 1985 214 Wind power generation for every European country Renewables.ninja dataset (Staffel & Pfenninger, 216, ENE; www.renewables.ninja) Daily month-ahead average, DJF, 1985 214 Beerli et al., 217, QJRMS 3 2 April 219 ECMWF
Beerli Results et al. Simple 3-categorical statistical forecast Stratos days ahead Weaker Stronger Figure 8. (a) The RPSS of three-categorical statistical fo SPV SPV value in the bins indicated on the x-axis. (b) Same as (a), et al.,of217, QJRMS Figure 7. Beerli The RPSS three-categorical statistical forecasts of month-aheadconfidence interval for the RPSS values derived by the boo average wind electricity generation as a function of lead time for eight European countries. The lead time on the x-axis indicates the start of the forecast day period. For instance, the RPSS at a lead time of 15 days shows the skill of forecasts for wind electricity generation averaged over 15 44 days ahead. The(their figure 7) for month-ahead temperature fo shaded colours show the confidence interval for the RPSS values derived by the cities ofinstitute Europe. For instance, our (IMK-TRO) RPSS for m 4 2 April 219 ECMWFdescribed Dominikin Büeler dominik.bueeler@kit.edu of Meteorology and Climate Research bootstrapping approach the text for Sweden (blue), Germany (red) electricity generation in Sweden is about.4 and Spain (yellow). How does this mechanism influence the skill of subseasonal numerical weather models?
Data Numerical forecast Subseasonal ECMWF model (www.s2sprediction.net) 2 reforecasts / week, DJF, 1995 217 11 ensemble members Fields calculated for each reforecast Strength of SPV = (Δ Z@hPa) 6-9 N At forecast initial time (Δ m wind) European Countries (Δ 2m temperature) European Countries (Δ precipitation) European Countries Average over 1 month lead time 5 2 April 219 ECMWF
Results Regional model skill pattern m wind 2m temperature Precipitation Anomalies after % strongest SPV states anomalies after % weakest SPV states 6 2 April 219 ECMWF
Results Model skill for m wind Anomalies after 2% strongest SPV states Anomalies after 2% weakest SPV states S2S ERA S2S ERA 7 2 April 219 ECMWF
Results Model skill for 2m temperature Anomalies after 2% strongest SPV states Anomalies after 2% weakest SPV states S2S ERA S2S ERA 8 2 April 219 ECMWF
Results Model skill for 2m temperature Anomalies after 2% strongest SPV states Anomalies after 2% weakest SPV states S2S ERA S2S ERA 9 2 April 219 ECMWF
Conclusions R. Beerli et al. ronger than normal polar event, but there are no n of a stratospheric signal e findings of Limpasuvan contrast, for weak polar the composite mean φpc en stratospheric warmings 21; Limpasuvan et al., 5 days prior to the events bit a surface signal that is The positive φpc in the re the weak polar-vortex heric precursor of SSWs, ous previous studies (e.g. suvan et al., 24). Given tential height anomalies of the SSWs documented in hese weakest polar-vortex SSWs. r stratospheric circulation electricity generation in C15 is far from its climahe lower stratosphere and on is a result of long-lived ions, which occur due to and the stratosphere. We hese results for the precity generation in Europe. city forecasts based on the n estigate the predictive skill p between the state of the wind electricity generation is used as a predictor for s of "CF 31d. We predict gical terciles of "CF 31d served "CF 31d ) using the ue of φpc15, we determine distribution (Pinit ). φpc15 ll within +/ % of this the basis to determine the "CF 31d from all pairs of The following example for ates this approach. On 19 ch is the 18.2th percentile Pinit = P18.2 = 8.8 m). d "CF 31d in the dataset P8.2 = 119.4 m) to the = 53.7 m) to derive the me season are left out to 5 "CF 31d pairs between % of the "CF 31d values are e tercile and 66.3% in the head forecast for "CF 31d 1%/66.3% probability of er tercile. If φpc15 < P, or φpc15 < Pinit+ (and untry, these forecasts are day from 1985 to 214 in "CF 31d i.e. only φpc15 m the current season are ally skilful forecasts. or each winter (DJF) day countries with the highest many, Spain, UK, France, or lagged 31 day windows for wind-power forecasts RPSS for 1 31 days ahead Figure 7. The RPSS of three-categorical statistical forecasts of month-ahead average wind electricity generation as a function of lead time for eight European countries. The lead time on the x-axis indicates the start of the forecast day period. For instance, the RPSS at a lead time of 15 days shows the skill of forecasts for wind electricity generation averaged over 15 44 days ahead. The shaded colours show the confidence interval for the RPSS values derived by the bootstrapping approach described in the text for Sweden (blue), Germany (red) and Spain (yellow). and so on). Additionally, we apply a bootstrapping approach in order to test the sampling sensitivity of the skill scores of these forecasts. In 2 repetitions, we randomly sample 8% of the winters and calculate the RPSS of each repetition. The % and 9% percentiles among these 2 RPSS values are the confidence intervals displayed in shaded colours in Figures 7 and 8. Comparing the RPSS for the eight countries mentioned above (Figure 7) reveals three groups of countries with similar levels of predictability. (1) High predictability of "CF 31d : Sweden and Denmark (in blue in Figure 7), RPSS.2 for lead time. (2) Moderate predictability of "CF 31d : Germany, UK and Poland (in red in Figure 7), RPSS.1 for lead time. (3) No predictability of "CF 31d : Spain, France and Italy (in yellow in Figure 7), RPSS < for lead time. These groups are in line with the findings derived from Figures 2 and 3. Sweden and Denmark are located in the centre of the high (low) wind corridor when φpc15 is strongly negative (positive), which makes it very likely that these countries will indeed experience above (below) normal CF in the following days. The countries with moderate predictability (Germany, UK and Poland) are situated at the southern edge of these high (low) wind corridors, which again makes it likely for them to experience above (below) normal CF, but this is less certain than for the Nordic countries just a subtle change in the synoptic set-up (which is not constrained by φpc15 ) may change the CF outcome. Hence the skill of the statistical forecast for these countries is notably lower than for the Nordic countries. For the Southern European countries (Spain, Italy and France), the RPSS is even negative, which means that the predictions based on the state of the lower stratosphere are worse than simply forecasting the climatological distribution of "CF 31d. These countries are situated well outside the areas with significantly positive or negative wind-speed anomalies shown in Figures 2 and 3. Therefore, there is no sufficiently strong signal related to the stratospheric circulation, which could be exploited to issue skilful forecasts. If the lead time for the forecasts of "CF 31d is increased, the skill levels get gradually lower but, for both high and moderate predictability countries, the statistical forecasts 15 45 days ahead are also better than the climatological reference forecast. The skill that we find here for month-ahead wind electricity generation is slightly higher than the skill levels found by Karpechko (215) Q. J. R. Meteorol. Soc. 143 : 25 36 (217) ological Society Strong spatial variability of statistical and numerical model skill for month-ahead prediction of wind power generation / surface weather in Europe Reason: anomalous SPV states at forecast initial time lead to persistent NAO-like anomaly patterns à model skill for countries located in affected regions tends to be enhanced However, model skill increase much more significant and robust after strongest SPV states than after weakest SPV states (~ SSWs), which even lead to significant skill reduction for certain countries (particularly T@2m) Implications Energy meteorology cannot rely on enhanced predictability after weakest (~ SSWs) but more after strongest SPV states Regional SSW response in S2S models needs to be improved 2 April 219 ECMWF