Identifying high-wind features within extratropical cyclones using a probabilistic random forest

<strong class="journal-contentHeaderColor">Abstract.</strong> Strong winds associated with extratropical cyclones are one of the most dangerous natural hazards in Europe. These high winds are mostly associated with five mesoscale dynamical features: the warm (conveyor belt) jet (WJ); the cold (conveyor belt) jet (CJ); cold frontal convection (CFC); strong cold-sector winds (CS); and, in some cases, the sting jet (SJ). The timing within the cyclone's life cycle, the location relative to the cyclone core and further characteristics differ between these features and, hence, likely also their associated forecast errors. In Part 1 of this study (Eisenstein et al., 2022b), we introduced the objective and flexible identification tool RAMEFI (RAndom-forest-based MEsoscale wind Feature Identification), which distinguishes between WJ, CFC and CS as well as CJ and SJ combined. RAMEFI is based on a probabilistic random forest trained on station observations of 12 storm cases over Europe. Being independent of spatial distribution, RAMEFI can also be applied to gridded data. Here, we use RAMEFI to compile a climatology over 19 extended winter seasons (October&ndash;March, 2000&ndash;2019) based on high-resolution regional reanalyses of the German Consortium for Small-scale Modelling (COSMO) model over Europe. This allows the first-ever long-term objective statistical analysis of the mesoscale wind features, their occurrence frequency, geographical distribution and characteristics. For Western and Central Europe, we demonstrate that the CS is prominent in most winter storms, while CFC is the least common cause of high winds, both in terms of frequency and affected area. However, probably due to convective momentum transport, CFC is on average the cause of the highest gusts after the CJ, and has the highest gust factor. As expected, CFC high-wind areas show high levels of humidity and overcast conditions. In contrast, CS is characterised by sunnier conditions interspersed by patchy cumulus clouds leading to a broader cloud cover distribution than for other features. The WJ produces the weakest winds on average, but affects a larger area than CJ. Central Europe is more strongly affected by WJ and CFC winds, while the CJ usually occurs farther north over the North and Baltic Seas, northern Germany, Denmark and southern Scandinavia. System-relative composites show that WJ and CFC tend to occur earlier in the cyclone life cycle than CJ and CS. Consistently, CS is the most common cause of high winds over Eastern Europe, where cyclones tend to occlude, represented by a narrowing warm sector and weakening cold front. The WJ mostly occurs within the southeastern quadrant of a cyclone bordering with the narrow CFC in the west. However, the location of CFC varies greatly between cases. The CS occurs in the southwestern quadrant, while the CJ appears closer to the cyclone centre, sometimes stretching into the southeastern quadrant. This objective climatology largely confirms previous, more subjective investigations but puts these into climatological context. It allows a more detailed analysis of feature properties, and provides a solid foundation for model assessment and forecast evaluation in future studies.

<strong class="journal-contentHeaderColor">Abstract.</strong> Strong winds associated with extratropical cyclones are one of the most dangerous natural hazards in Europe. These high winds are mostly associated with five mesoscale dynamical features: the warm (conveyor belt) jet (WJ); the cold (conveyor belt) jet (CJ); cold frontal convection (CFC); strong cold-sector winds (CS); and, in some cases, the sting jet (SJ). The timing within the cyclone's life cycle, the location relative to the cyclone core and further characteristics differ between these features and, hence, likely also their associated forecast errors. In Part 1 of this study (Eisenstein et al., 2022b), we introduced the objective and flexible identification tool RAMEFI (RAndom-forest-based MEsoscale wind Feature Identification), which distinguishes between WJ, CFC and CS as well as CJ and SJ combined. RAMEFI is based on a probabilistic random forest trained on station observations of 12 storm cases over Europe. Being independent of spatial distribution, RAMEFI can also be applied to gridded data. Here, we use RAMEFI to compile a climatology over 19 extended winter seasons (October&ndash;March, 2000&ndash;2019) based on high-resolution regional reanalyses of the German Consortium for Small-scale Modelling (COSMO) model over Europe. This allows the first-ever long-term objective statistical analysis of the mesoscale wind features, their occurrence frequency, geographical distribution and characteristics. For Western and Central Europe, we demonstrate that the CS is prominent in most winter storms, while CFC is the least common cause of high winds, both in terms of frequency and affected area. However, probably due to convective momentum transport, CFC is on average the cause of the highest gusts after the CJ, and has the highest gust factor. As expected, CFC high-wind areas show high levels of humidity and overcast conditions. In contrast, CS is characterised by sunnier conditions interspersed by patchy cumulus clouds leading to a broader cloud cover distribution than for other features. The WJ produces the weakest winds on average, but affects a larger area than CJ. Central Europe is more strongly affected by WJ and CFC winds, while the CJ usually occurs farther north over the North and Baltic Seas, northern Germany, Denmark and southern Scandinavia. System-relative composites show that WJ and CFC tend to occur earlier in the cyclone life cycle than CJ and CS. Consistently, CS is the most common cause of high winds over Eastern Europe, where cyclones tend to occlude, represented by a narrowing warm sector and weakening cold front. The WJ mostly occurs within the southeastern quadrant of a cyclone bordering with the narrow CFC in the west. However, the location of CFC varies greatly between cases. The CS occurs in the southwestern quadrant, while the CJ appears closer to the cyclone centre, sometimes stretching into the southeastern quadrant. This objective climatology largely confirms previous, more subjective investigations but puts these into climatological context. It allows a more detailed analysis of feature properties, and provides a solid foundation for model assessment and forecast evaluation in future studies.

During 2–3 December 2012, the Black Sea and east coast of Romania were affected by a rapidly deepening Mediterranean cyclone. The cyclone developed a bent-back front along which short-lived (2–4 h) strong winds up to 38 m s−1 were recorded equatorward of the cyclone center. A mesoscale model simulation was used to analyze the evolution of the wind field, to investigate the physical processes that were responsible for the strong winds and their acceleration, and to investigate the relative importance of the stability of the boundary layer to those strong winds. The origin of the air in the wind maximum equatorward of the cyclone center was twofold. The first was associated with a sting jet, a descending airstream from the midlevels of the cloud head and the lower part of the cyclonic branch of the warm conveyor belt. The sting jet started to descend west of the cyclone center, ending at the frontolytic tip of the bent-back front. The second was a low-level airstream associated with the cold conveyor belt that originated northeast of the cyclone center and traveled below 900 hPa along the cold side of the bent-back front, ending behind the cold front. Both airstreams were accelerated by the along-flow pressure gradient force, with the largest accelerations acting on the sting-jet air before entering into the near-surface strong-wind area. The sensible heat fluxes destabilized the boundary layer to near-neutral conditions south of the cyclone center, facilitating downward mixing and allowing the descending air to reach the surface. Mesoscale instabilities appeared to be unimportant in the sting-jet formation.

An idealised modelling study of sting‐jet cyclones is presented. Sting jets are descending mesoscale jets that occur in some extratropical cyclones and produce localised regions of strong low‐level winds in the frontal fracture region. Moist baroclinic lifecycle (LC1) simulations are performed with modifications to produce cyclones resembling observed sting‐jet cyclones.A sting jet exists in the idealised control cyclone with similar characteristics to the sting jet in a simulation of windstormGudrun(a confirmed sting‐jet case). Unlike in windstormGudrun, a low‐level layer of strong moist static stability prohibits the descent of the strong winds from above the boundary layer to the surface in the idealised case. Conditional symmetric instability (CSI) exists in the cloud head and dissipates as the sting jet leaves the cloud head and descends. The descending, initially moist, sting‐jet trajectories consistently have negative or near‐zero saturated moist potential vorticity but moist static stability and inertial stability, consistent with CSI release; the moist static stability becomes negative during the period of most rapid descent, by which time the air is relatively dry implying conditional instability release is unlikely.Sensitivity experiments show that the existence of the sting jet is robust to changes in the initial state, and that the initial tropospheric static stability significantly impacts the descent rate of the sting jet. Inertial and conditional instability are probably being released in the experiment with the weakest initial static stability. This suggests that sting jets can arise through the release of all three instabilities associated with negative saturated moist potential vorticity. While evaporative cooling occurs along the sting‐jet trajectories, a sensitivity experiment with evaporation effects turned off shows no significant change to the wind strength or descent rate of the sting jet implying that instability release is the dominant sting‐jet driving mechanism.

Extra-tropical windstorms are one of the costliest natural hazards affecting Europe, and windstorms that develop a sting jet are extremely damaging. A sting jet is a mesoscale core of very high wind speeds that occurs in Shapiro–Keyser type cyclones, and high-resolution models are required to adequately model sting jets. Here, we develop a low-cost methodology to automatically detect sting jets, using the characteristic warm seclusion of Shapiro–Keyser cyclones and the slantwise descent of high wind speeds, within pan-European 2.2 km convection-permitting climate model (CPM) simulations. The representation of wind gusts is improved with respect to ERA-Interim reanalysis data compared to observations; this is linked to better representation of cold conveyor belts and sting jets in the CPM. Our analysis indicates that Shapiro–Keyser cyclones, and those that develop sting jets, are the most damaging windstorms in present and future climates. The frequency of extreme windstorms is projected to increase by 2100 and a large contribution comes from sting jet storms. Furthermore, extreme wind speeds and their future changes are underestimated in the global climate model (GCM) compared to the CPM. We conclude that the CPM adds value in the representation of extreme winds and surface wind gusts and can provide improved input for impact models compared to coarser resolution models.

High winds are one of the key forecast challenges across southeast Wyoming. The complex mountainous terrain across the region frequently results in strong gap winds in localized areas, as well as more widespread bora and chinook winds in the winter season (October–March). The predictors and general weather patterns that result in strong winds across the region are well understood by local forecasters. However, no single predictor provides notable skill by itself in separating warning-level events from others. Random forest (RF) classifier models were developed to improve upon high wind prediction using a training dataset constructed of archived observations and model parameters from the North American Regional Reanalysis (NARR). Three locations were selected for initial RF model development, including the city of Cheyenne, Wyoming, and two gap regions along Interstate 80 (Arlington) and Interstate 25 (Bordeaux). Verification scores over two winters suggested the RF models were beneficial relative to current operational tools when predicting warning-criteria high wind events. Three case studies of high wind events provide examples of the RF models’ effectiveness to forecast operations over current forecast tools. The first case explores a classic, widespread high wind scenario, which was well anticipated by local forecasters. A more marginal scenario is explored in the second case, which presented greater forecast challenges relating to timing and intensity of the strongest winds. The final case study carefully uses Global Forecast System (GFS) data as input into the RF models, further supporting real-time implementation into forecast operations.

&amp;lt;p&amp;gt;Storm surges are driven by low air pressure and strong winds in tropical (TC) and extratropical cyclones (ETC). Coastal flooding is often caused by this type of extreme weather with large socio-economic impacts in densely populated and low-lying coastal areas. Recent examples of coastal disasters include typhoon Hagibis that made landfall in Japan, Hurricane Dorian which devastated the northwestern Bahamas, and extratropical cyclone Xaver that affected northern Europe. Each of these storms generated dangerous storm surges, reaching 6m in some parts of the Bahamas during Hurricane Dorian with approximately 100 fatalities as a result. Economic losses are estimated at 10 billion U.S. Dollars for both typhoon Hagibis and hurricane Dorian.&amp;lt;/p&amp;gt;&amp;lt;p&amp;gt;To inform flood risk management and develop effective adaptation strategies it is important to have accurate information on return periods of extreme sea levels. To date, there exists no global database with return periods of extreme sea levels that fully includes TCs. Global databases of extreme sea levels are typically based on historical climate simulations covering multiple decades. While this is sufficient for ETCs, TCs will be underestimated in such databases. This because TCs have generally low probabilities and affect only a small stretch of coastline, compared to ETCs. A climate reanalysis covering multiple decades includes too few TCs to perform an extreme value analysis. To resolve this, previous studies at local scale have used synthetic TC tracks generated by a statistical model to estimate the probabilities of extreme sea levels.&amp;lt;/p&amp;gt;&amp;lt;p&amp;gt;The aim of this research is to develop a global database of extreme sea levels that include both ETCs and TCs. For ETCs, we force the hydrodynamic Global Tide and Surge Model (GTSM) with ERA5 10-meter wind speed and air pressure data to calculate the return periods of extreme sea levels based on the period 1979-2017. Since ERA5 includes all storms, we filter out extreme sea levels caused by TCs. Preliminary results show that GTSM forced with ERA5 atmospheric data performs well for ETCs. For TCs, we force GTSM with synthetic TC tracks that correspond to 10.000 year of TC statistics. The synthetic tracks of TCs are obtained from the STORM model (Bloemendaal et al., in review) based on the International Best Track Archive for Climate Stewardship (IBTrACS) TC database. With STORM it is possible to statistically extend the ~38-year observed dataset to a 10.000-year synthetic dataset. The synthetic dataset preserves the climatological statistics as found in the original dataset. Finally, we will merge the TC and ETC related return periods to create a global extreme sea level database.&amp;lt;/p&amp;gt;

In this paper South Africa is divided into strong wind climate zones, which indicate the main sources of annual maximum wind gusts. By the analysis of wind gust data of 94 weather stations, which had continuous climate time series of 10 years or longer, six sources, or strong-wind producing mechanisms, could be identified and zoned accordingly. The two primary causes of strong wind gusts are thunderstorm activity and extratropical low pressure systems, which are associated with the passage of cold fronts over the southern African subcontinent. Over the eastern and central interior of South Africa annual maximum wind gusts are usually caused by thunderstorm gust fronts during summer, while in the western and southern interior extratropical cyclones play the most dominant role. Along the coast and adjacent interior annual extreme gusts are usually caused by extratropical cyclones. Four secondary sources of strong winds are the ridging of the quasi-stationary Atlantic and Indian Ocean high pressure systems over the subcontinent, surface troughs to the west in the interior with strong ridging from the east, convergence from the interior towards isolated low pressure systems or deep coastal low pressure systems, and deep surface troughs on the West Coast.

Extra-Tropical Cyclones are major systems ruling and influencing the atmospheric structure at mid-latitudes. They are characterised by strong winds and heavy precipitation, and can cause considerable storm surges potentially devastating for coastal regions. The availability of historical observations of the extreme events caused by intense ETCs are rather limited, hampering risk evaluation. Increasing the amount of significant data available would substantially help several fields of analysis influenced by these events, such as coastal management, agricultural production, energy distribution, air and maritime transportation, and risk assessment and management.Here, we address the possibility of generating synthetic ETC atmospheric fields of mean sea level pressure, wind speed, and precipitation in the North Atlantic by training a Generative Adversarial Network (GAN). The purpose of GANs is to learn the distribution of a training set based on a game theoretic scenario where two networks compete against each other, the generator and the discriminator. The former is trained to generate synthetic examples that are plausible and resemble the real ones. The input of the generator is a vector of random Gaussian values, whose domain is known as the &amp;#8220;latent space&amp;#8221;. The discriminator learns to distinguish whether an example comes from the dataset distribution. The competition set by the game-theoretic approach improves the network until the counterfeits are indistinguishable from the originals.To train the GAN, we use atmospheric fields extracted from the ERA5 reanalysis dataset in the geographic domain with boundaries 0&amp;#176;- 90&amp;#176;N, 70&amp;#176;W - 20&amp;#176;E and for the period 1st January 1979 - 1st January 2020. We analyse the generated samples&amp;#8217; histograms, the samples&amp;#8217; average fields, the Wasserstein distance and the Kullback-Leibler divergence between the generated samples and the test set distributions. Results show that the generative model has learned the distribution of the values of the atmospheric fields and the general spatial trends of the atmosphere in the domain. To evaluate better the atmospheric structure learned by the network, we perform linear and spherical interpolations in the latent space. Specifically, we consider four cyclones and compare the frames of their tracks to those of the synthetic tracks generated by interpolation. The interpolated tracks show interesting features consistent with the original tracks. These findings suggest that GANs can learn meaningful representations of the ETCs&amp;#8217; fields, encouraging further investigations to model the tracks&amp;#8217; temporal evolution.

Influence of extratropical cyclones on air quality in Beijing

The present study investigates changes in the location, frequency, intensity, and dynamical processes of North Atlantic extratropical cyclones with warming consistent with the IPCC Fifth Assessment Report (AR5) representative concentration pathway 8.5 (RCP8.5) scenario. The modeling, analysis, and prediction (MAP) climatology of midlatitude storminess (MCMS) feature-tracking algorithm was utilized to analyze 10 cold-season high-resolution atmospheric simulations over the North Atlantic region in current and future climates. Enhanced extratropical cyclone activity is most evident in the northeast North Atlantic and off the U.S. East Coast. These changes in cyclone activity are offset from changes in eddy kinetic energy and eddy heat flux. Investigation of the minimum SLP reached at each grid point reveals a lack of correspondence between the strongest events in the current and future simulations, indicating the future simulations produced a different population of storms. Examination of the percent change of storms in the storm-track region shows a reduction in the number of strong storms (i.e., those reaching a minimum SLP perturbation of at least −51 hPa). Storm-relative composites of strong and moderate storms show an increase in precipitation, associated with enhanced latent heat release and strengthening of the 900–700-hPa layer-average potential vorticity (PV). Other structural changes found for cyclones in a future climate include weakened upper-level PV for strong storms and a weakened near-surface potential temperature anomaly for moderate storms, demonstrating a change in storm dynamics. Furthermore, the impacts associated with extratropical cyclones, such as strong near-surface winds and heavy precipitation, strengthen and become more frequent with warming.

Background: Daytime convectively-driven boundary layer turbulence produces structures on scales smaller than a few kilometers that can only be resolved in modeling by large-eddy simulations (LES). These structures, when superimposed on larger-scale dynamically- and topographically-driven flows, describe the wind regime experienced by aeolian features on the surface of Mars. Because convective turbulence produces the small percentage of strong wind gusts that may exceed the saltation stress threshold, such activity must be considered in order to properly predict particle entrainment, particularly in an environment in which mean winds are generally much weaker than the saltation stress threshold. We present early results from large-eddy simulations that characterize the daytime turbulent wind distributions under varying local conditions. Method: We have run several large-eddy simulations using the Mars Regional Atmospheric Modeling System (MRAMS) at two locations on Mars, the Viking Lander 1 and Phoenix landing sites. Most simulations were defined at Ls=120o, northern summer, but one set of simulations took place at Ls=300o, northern winter. We have tested a number of initial wind conditions, including cases with no mean wind, a weak wind that is uniform with height, wind speed shear that increases with height, and two cases of strong uniform winds with superimposed speed shear. The resulting friction velocity distributions have been fit to a two-parameter Weibull probability density function (PDF), providing parameters that could help to quantify the influence of convective turbulence on wind distributions. Conclusion: Daytime convective turbulence considerably widens the distribution of friction velocities, increasing the probability that winds will exceed the stress threshold and mobilize sand grains on Mars. The amount of widening increases with insolation as determined by latitude, season, and local time. Although no initial wind is necessary to produce a high speed wind tail that exceeds the local saltation stress threshold, initial winds can add momentum to the model domain that can be mixed down to the surface further increasing the likelihood of strong near-surface winds. In most cases tested thus far, friction velocity fits to a two-parameter Weibull distribution produce shape parameter values ranging from 2.5 to 3, which are larger than those previously used, and scale parameter values approximately equal to the mean friction velocity. Preliminary estimates indicate that increasingly complex and realistic initial winds exhibit friction velocity distributions that d epart significantly from the Weibull distribution, with shape parameter values that range up to ~7. Describing the PDF of LES friction velocities with a Weibull distribution has known inadequacies, represents to date the best available simple analytic description of wind distributions for the modeling community. Therefore, these fit parameters are best used with caution, especially in high wind events, and applied only to conditions similar to those modeled.

&amp;lt;p&amp;gt;The ability to estimate flux exchanges between the sea-surface and the atmosphere has tremendous importance on weather prediction and climate simulations. These exchanges are influenced by wave processes - growth and decay, and turbulent interactions at the air-sea interface. For momentum, the ensemble of these exchanges is presented as the sea-surface drag (Cd), which increases with (10-m high) wind intensity till about 20-30 m/s, and decreases thereafter. The reason for this decrease remains less understood, mainly due to (i) our inability to explicitly measure the individual wind-wave exchanges, and (ii) the inability of existing semi-empirical parameterizations to explain the Cd behavior. To this end, we developed a physically based stress parameterization for a coupled wind-wave model, capable of reproducing both wave growth and wave breaking stresses at the air-sea interface. The advantage of such a numerical approach, over field experiments, is that it allows us to investigate the different process, under different constraining environments, in-order to disentangle the factors in play on Cd. Our coupled model enables a two-way interaction between the ocean-waves and turbulent flow. and can simulate (i) the main turbulent eddies of the air-flow, and (ii) the wind-wave interactions. After evaluating the model against published field experiments we use it to explore the impact of wave growth and wave-breaking on the Cd under strong winds. Our results demonstrate that under strong winds the air-flow gets separated from the sea-surface, a process associated with wave-breaking, resulting in the turbulent flow sensing a smoother surface as against an actually rough sea surface, thereby decreasing Cd. Finally, our model allows us to investigate the sensitivity of Cd to different influencing factors under strong winds.&amp;lt;/p&amp;gt;

This paper evaluates the extratropical cyclones within three operational global ensembles [the 20-member Canadian Meteorological Centre (CMC), 20-member National Centers for Environmental Prediction (NCEP), and 50-member European Centre for Medium-Range Weather Forecasts (ECMWF)]. The day-0–6 forecasts were evaluated over the eastern United States and western Atlantic for the 2007–15 cool seasons (October–March) using the ECMWF’s ERA-Interim dataset as the verifying analysis. The Hodges cyclone-tracking scheme was used to track cyclones using 6-h mean sea level pressure (MSLP) data. For lead times less than 72 h, the NCEP and ECMWF ensembles have comparable mean absolute errors in cyclone intensity and track, while the CMC errors are larger. For days 4–6 ECMWF has 12–18 and 24–30 h more accuracy for cyclone intensity than NCEP and CMC, respectively. All ensembles underpredict relatively deep cyclones in the medium range, with one area near the Gulf Stream. CMC, NCEP, and ECMWF all have a slow along-track bias that is significant from 24 to 90 h, and they have a left-of-track bias from 120 to 144 h. ECMWF has greater probabilistic skill for intensity and track than CMC and NCEP, while the 90-member multimodel ensemble (NCEP + CMC + ECMWF) has more probabilistic skill than any single ensemble. During the medium range, the ECMWF + NCEP + CMC multimodel ensemble has the fewest cases (1.9%, 1.8%, and 1.0%) outside the envelope compared to ECMWF (5.6%, 5.2%, and 4.1%) and NCEP (13.7%, 10.6%, and 11.0%) for cyclone intensity and along- and cross-track positions.

The weather and climate conditions of the Mediterranean region is greatly influenced by mid-latitude cyclones. The region is considered as one of the most vulnerable areas to climate change, and Mediterranean cyclones and their characteristics might change in the future. We selected the western part of the Mediterranean region as our study area. In this area we aim to identify cyclones and get an understanding of their characteristics on the basis of ECMWF reanalysis data and the historical simulations of global climate models from the CMIP6 project. In the study we identify the potential areas of low pressure systems based on mean sea level pressure, with temporal resolution of 6 hours, and then determine their trajectories. Mediterranean cyclones affect other parts of Europe, including Hungary and its vicinity. We aim to analyze what portion of the annual precipitation is connected to Mediterranean cyclones and their frontal systems, and how it will change in the future. For this purpose, we use precipitation data from the above-mentioned databases. This study is focused on the present and historical conditions. Our findings are validated by comparing the results of reanalysis and CMIP6-simulation data. The study continues with the evaluation of future trends based on different climate scenarios (RCP and/or SSP scenarios upon availability), which represent different anthropogenic impacts and mitigation efforts. The results can serve as essential input for further impact studies, and then for decision makers and stakeholders when building the long-term strategies of their corresponding sectors. Acknowledgements: The study contributes to the COST CA19109 action (MEDCYCLONES). Research leading to this study has been supported by the following sources: the Hungarian National Research, Development and Innovation Fund (under grants K-120605 and K-129162), and the National Multidisciplinary Laboratory for Climate Change (RRF-2.3.1-21-2022-00014).