Extreme Wind Shear Events in US Offshore Wind Energy Areas and the Role of Induced Stratification

As the offshore wind industry emerges on the U.S. East Coast, a comprehensive understanding of the wind resource — particularly extreme events — is vital to the industry’s success. Such understanding has been hindered by a lack of publicly available wind profile observations in offshore wind energy areas. However, the New York State Energy Research and Development Authority (NYSERDA) recently funded the deployment of two floating lidars within two current lease areas off the coast of New Jersey. These floating lidars provide publicly available wind speed data from 20 m to 200 m height with 20-m 5 vertical resolution. In this study, we leverage a year of these lidar data to quantify and characterize the frequent occurrence of high wind shear and low-level jet events, both of which will have considerable impact on turbine operation. We find that almost 100 independent events occur throughout the year with mean wind speed at 100 m height and power-law exponent of 16 m/s and 0.28, respectively. The events have strong seasonal variability, with the highest number of events in summer and the lowest in winter. A detailed analysis reveals that these events are enabled by an induced stable stratification when when warmer air 10 from the south flows over the colder mid-Atlantic waters, leading to a positive air-sea temperature difference. Copyright statement. This work was authored by the National Renewable Energy Laboratory, operated by Alliance for Sustainable Energy, LLC, for the U.S. Department of Energy (DOE). The views expressed in the article do not necessarily represent the views of the DOE or the U.S. Government. The U.S. Government retains and the publisher, by accepting the article for publication, acknowledges that the U.S. Government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this work, or allow 15 others to do so, for U.S. Government purposes.


Introduction
The offshore wind industry is rapidly developing on the US east coast and a comprehensive understanding of the wind resource in this area is critical for the industry's success. There are currently 15 active lease areas with over 21 Gigawatts (GW) of planned capacity spanning from Massachusetts to North Carolina ( Fig. 1) with an additional planned 86-GW capacity in all 20 U.S. waters by 2050 (BOEM, 2018). Proposed lease areas are located on the Atlantic Outer Continental Shelf (OCS) and span locations ranging from a minimum of 15 km to a maximum of over 100 km from the coastline. The proper planning, design, and operation of these wind farms require an in-depth understanding of the wind characteristics in the OCS, in particular the frequency and magnitude of extreme events that largely impact the power performance, safety, and operation of wind turbines (Musial and Ram, 2010;Rose et al., 2012;Archer et al., 2014). Extreme wind events relevant to wind turbine operation include rapid changes in flow direction and speed, or persistently high values of shear and veer (IEC, 2019). High vertical wind shear is of particular interest to wind energy as it has a direct effect on wind turbine power and reliability (Murphy et al., 2019;Gutierrez et al., 2014Gutierrez et al., , 2017Gutierrez et al., , 2019Colle and Novak, 2010).
In the last decade, a growing body of work has identified and characterized high-shear events in the U.S. Mid-Atlantic that are brought on by low-level jets (LLJs). These offshore LLJs, spanning from Maryland to New Jersey, have been investigated 30 with the Weather Research and Forecasting (WRF) model (Strobach et al., 2018;Colle et al., 2016;Nunalee and Basu, 2014), ship-borne lidar (Pichugina et al., 2017;Strobach et al., 2018), aircraft measurements , sodar (Helmis et al., 2013), radiosonde (Helmis et al., 2013;Colle and Novak, 2010;Nunalee and Basu, 2014), and radar wind profilers (Zhang et al., 2006;Nunalee and Basu, 2014). A consensus agreement among these studies is the frequent occurrence of persistent LLJs in this area during the warm season. While some studies were limited to heights above wind turbine operation (Nunalee 35 and Basu, 2014; Zhang et al., 2006), others found wind speed maxima at heights representative of a typical wind turbine rotor (Pichugina et al., 2017;Strobach et al., 2018;Colle and Novak, 2010).
These previous studies have found the development of LLJs in the Mid Atlantic to be strongly linked to static stability in the atmospheric boundary layer (ABL). Despite the importance of stratification to this and other wind phenomena, there is still a lack of consensus on what may be the prevailing conditions of stability and turbulence in this region. Analyses of near-shore 40 2 https://doi.org/10.5194/wes-2020-103 Preprint. Discussion started: 17 September 2020 c Author(s) 2020. CC BY 4.0 License. measurements have indicated both very low turbulence conditions (Bodini et al., 2019), indicative of stable stratification, and predominantly unstable conditions . Given the role of atmospheric stability in influencing wind profiles across the nominal rotor layer as well as turbine wake propagation, it is clear that more analysis and data are needed to better understand prevailing atmospheric conditions in the Atlantic OCS.
While the aforementioned studies were extremely valuable in providing an initial characterization of offshore wind condi-45 tions, limitations of the measurements used undermine their value to current US east coast wind energy lease areas. Many of the datasets were spatially disjunct (Pichugina et al., 2017;Strobach et al., 2018;Colle et al., 2016) or limited to coastal areas Helmis et al., 2013;Nunalee and Basu, 2014;Zhang et al., 2006). The only two experiments recorded in literature that were far enough from the coast to be representative of conditions that will be experienced by offshore wind plants were limited in duration to a maximum of one month (Helmis et al., 2013;Strobach et al., 2018;Pichugina et al., 2017).

50
Increasing investments in offshore wind energy along with continuous instrumentation developments have enabled a surge in deployments of offshore wind measurement systems. In particular, the emergence of buoy-mounted floating lidar has led to at least 10 and as many as 20 floating lidar deployments in the US east coast in recent years. At large, these data have been kept proprietary and any derived analyses have not been disseminated. In August and September 2019, however, the New York State Energy Research and Development Authority (NYSERDA) funded the deployment of two floating lidars (DNVGL, 55 2020) within two current lease areas in the New Jersey offshore wind area (Fig. 1). These floating lidars provide wind data at multiple heights across the rotor layer (Table 1). To our knowledge, these deployments provide the first publicly available comprehensive and relevant data set for the analysis of wind characteristics in US east coast active lease areas and, as such, are of immense value for wind energy research.
A cursory look at the NYSERDA data alone can reveal very important wind characteristics and phenomena. We show an 60 example of this in Fig. 2 where an intense high-shear event existing over a 2-day period is measured at the northeast (NE) buoy. Not only do we see frequent extreme shear across the nominal rotor area but also several very-low-level jet (VLLJ) events where the peak in the wind profiles is as low as 100 m. In the highlighted VLLJ and monotonic-shear periods, the time-averaged profiles reveal a power-law exponent of 0.59 and 0.32, respectively, when measured across a nominal rotor layer spanning between 40 m and 160 m. This corresponds to wind speed gradient, ∆U /∆z, values of 0.12 1/s and 0.08 1/s, respectively, across the rotor layer. The ability to accurately predict such events using numerical weather prediction (NWP) models is crucial for wind resource assessment, wind power forecasting, and for the timely implementation of operation and maintenance procedures to protect turbines from damage. A proper documentation of these extreme events will help to identify the shortcomings of the models needed for further improvement and also will guide the development of more accurate standard guidelines for offshore wind turbines. To our knowledge, the existence of these high-shear events, let alone their causes and 70 development, have not been previously studied in the US east coast offshore wind lease areas. Our goal is to characterize these events and understand the physical mechanisms governing their onset and dissipation. To do so, we leverage these novel floating lidar observations in the US offshore wind areas.  Time series of vertical profiles of wind speed at the two buoy sites are used to detect and characterize high-shear events that are 75 relevant for offshore wind development. The algorithm developed to detect these events discerns between two types of wind speed profiles: monotonic shear and VLLJ (Fig. 2). The algorithm is applied to each 10-minute-mean profile. When high shear is detected for a continuous period of one hour or longer, this period is defined as a high-shear event. To avoid double counting, separate events that are close in time and measured at the same site are merged into a single, longer event. This is done in two steps: first, events with lower shear that last one hour or less but are sandwiched in between two high-shear periods are 80 identified as integral part of the adjacent events and merged into them to form one, longer event; finally, two events that are within six hours of each other are merged into a single, long-lived event.
The monotonic-shear profiles refer to 10-minute averaged profiles in which the wind speed magnitude strictly increases with height ( Fig. 2, right-side profile). For the VLLJ cases, the wind speed magnitude increases up to a certain height and then decreases, revealing the presence of a LLJ with a nose below 200 m (Fig. 2, left-side profile). While the monotonic shear cases 85 could be the lower part of a LLJ with a nose above 200 m, the vertical extent of our measurements does not allow for that distinction to be made. For this reason, the algorithm was developed to distinguish between both.
The detection of both types of high-shear profiles is based on several conditions, as outlined below and shown by the schematic in Fig. 3. We define nominal hub height and rotor diameter values to be 100 m and 120 m, respectively (the rotor span being between 40 m and 160 m). These are assumed to be representative of an offshore wind turbine and are used here 90 to facilitate the interpretation of results in the context of offshore wind development. For the analysis performed here, only profiles with a hub-height wind speed greater than 3 m/s are considered. A profile is classified as "monotonic shear" if (i) the rotor-layer shear is greater than a pre-specified threshold value, ∆U ∆z rotor ≥ ∆U ∆z rotor_threshold .
A profile is classified as "VLLJ" if 95 (i) the height of maximum shear (as computed between z rotor_bottom and z) is between the second (40 m) and second-to-last (180 m) measurement height, (ii) the maximum shear across the rotor layer is greater than the same pre-specified threshold value used for the monotonic- where ∆U drop = U top − U nose and U top marks the top of the jet and is the first local minimum in wind speed identified above 105 the nose. The enforcement of both dimensional and nondimensional wind speed drop off criteria is based on previous work (Baas et al., 2009) but the threshold values are adjusted in magnitude here due to the limited vertical extent of the measurement data available. In the wind energy industry, the vertical wind shear is typically represented by the power-law exponent, α (IEC, 2019).
However, in this work, the variable used to quantify vertical wind shear is wind speed gradient between a reference height 110 (here taken as 40 m) and other heights above it. A relationship plot ( Fig. 4a) among wind speed at hub height, U 100m , wind speed gradient across the rotor, ∆U ∆z , and shear exponent, α, explains that the shear exponent can be very low even though a turbine faces high wind speed difference across its diameter. The shear exponent is non-dimensional and does not consider the magnitude of wind speed that a turbine actually faces. As a result, data points that would normally be considered as high shear by α often have relatively low wind speeds and would not pose a danger to wind turbines. To better capture events that do 115 pose that danger, we consider instead the ∆U ∆z metric -which does account for wind speed magnitude -as a threshold for detecting high wind shear events. The distribution of ∆U ∆z for the buoys are presented in Fig. 4b. The figure shows a long tail in the the distribution that captures a considerable number of high shear events. Setting a threshold at the 90 th percentile, as shown in the figure, is able to capture a large number of events while still ensuring that the shear values are extreme. Herein, for both types of profiles, the threshold shear value ∆U ∆z rotor_threshold is set to the 90 th percentile of the distribution of ∆U ∆z rotor 120 over the entire measurement period, which equals 0.035 s −1 (Fig. 4b) when averaged across the lidars. 3 Results

Detected Events
We first summarize the results of the high-shear detection algorithm in Fig. 5. A large number of events are detected at both lidars, most of which are less than 10 hours but some which extend for more than two days. All the events identified based on 125 the detection criteria are marked as "high shear" events and presented in this section and include both VLLJ and monotonicshear cases. The total number of detected events are 104 and 92 for Northeast (NE) and Southwest (SW) buoy, respectively.
To explain why there are more events at the SW buoy, we must first better understand the atmospheric conditions in which these events are able to occur. We begin this investigation in the next section by looking at seasonal and diurnal trends in event frequency.

Seasonal and Diurnal Dependence
We explore seasonal and diurnal trends in the high-shear events in Fig. 6. In Fig. 6a & Fig. 6b, we consider the number of 10-minute average data points as they depend on hours of diurnal cycles and months, respectively. In Fig. 6c, we consider actual event counts by month. We see in Fig. 6a a clear diurnal trend in the high-shear events, with event frequency increasing after noon and dropping after 22:00. Indeed, events are twice as likely to happen during the night than during the morning. We 135 see in Fig. 6b and Fig. 6c that there is also a strong seasonal trend in event frequency. Events are largely concentrated in the spring months (i.e., March through June) and are much less frequent in the rest of the year. In particular, the month of June has the highest number of events (16 events, on average) and November has the lowest number of events (on average 1 event).
The presence of strong diurnal and seasonal trends in the number of high-shear events suggest the influence of local meteorological conditions, particularly, the atmospheric stability. Indeed, we expect this to be the case which follows the well-140 established relationships between high wind shear, LLJs and thermodynamic atmospheric stability established by previous works (Monin and Obukhov, 2009;Stull, 1988;Poulos et al., 2002;Wharton and Lundquist, 2012). In the next section, we explore this possible relationship between high shear and atmospheric stability in more detail.

Atmospheric Stability and Turbulence
In this section we explore the role of atmospheric stability and turbulence in driving these high-shear events. To measure 145 atmospheric stability, we are limited to quantifying the difference in 2-m air temperature, T a , and the sea surface temperature, SST , given the lack of temperature measurements aloft. We denote this air-sea temperature difference as ∆T from herein. To measure turbulence, we use the turbulence intensity (TI) measurements at 100 m as measured by the floating lidars, denoted T I 100m . In Fig. 7a & Fig. 7b, we plot distributions of ∆T and T I 100m , where the full data set are shown in blue with the high-shear events shown in orange.

150
It is clear from Fig. 7a that high-shear events are strongly associated with a positive air-sea temperature difference (∆T > 0). The distribution of T I 100m is shown for both high-shear events and the full data set in Fig. 7b. The high shear events have turbulence intensity mostly within the bin of 4% to 6% (mean T I 100m , 5.1%) whereas mean turbulence intensity of all the data-set is 8.3%. Focusing only on the high-shear events (i.e., the orange distributions), we plot ∆T and T I distributions by wind direction in Fig. 7c and Fig. 7d. We see that these high-shear events are almost exclusively associated with southwesterly 155 flow with a mean wind direction of 217 o . Referring to Fig. 1, we see that southwesterly flow is about parallel to the coastline and features an area of very large ocean fetch.
The observations in Section 3.3 suggest the role of induced stable stratification in causing these high-shear events. It is possible that warmer air coming from the southwest encounters the colder waters of the Mid-Atlantic, causing a positive air-sea temperature difference. This temperature difference would then induce stable stratification where vertical turbulent 160 exchange from surface winds to those aloft would be reduced and a degree of "decoupling" of winds aloft from the surface would occur. Combined with the long ocean fetch where surface roughness is low, this is likely leading to very low turbulence in the winds aloft at the floating lidars, sufficient to cause high wind shear and allow for the formation of low-level jets.
We provide evidence of this induced stratification in Fig. 8 for two high-shear events. As shown for both case studies, the onset of high shear aligns with the switch from a negative ∆T to a positive ∆T value. Notably, the end of the second high 165 shear event aligns with the switch back to a negative ∆T value. Furthermore, we see that the change in sign in ∆T is driven by changes in the air temperature, T a , while the SST remains relatively constant before, during, and after the high shear events. So indeed, the arrival of warm air from the southwest and the resulting induction of stable stratification appears to be a dominant contributor to these high-shear events.
We further examine the role of the air-sea temperature difference in influencing wind conditions in Fig. 9. Here we consider is shown with red color in the background. We see that wind speed at hub height is almost constant when the temperature difference is negative, but increases sharply when temperature difference is positive. The linear increase of wind speed with 175 increase of positive temperature difference (∆T > 0) suggests that the strength of extreme events is highly dependent on the magnitude of positive temperature difference. On the other hand, turbulence intensity at hub height drops as the temperature difference approaches zero, showing a strong dependency on static stability (Fig. 9b). There is an upward trend in the turbulence intensity after ∆T =2 o C. This could be due to a low density of the data within the bin. Similar to wind speed, both the shear 10 https://doi.org/10.5194/wes-2020-103 Preprint. Discussion started: 17 September 2020 c Author(s) 2020. CC BY 4.0 License.

Spatial Variability
In this section we briefly explore potential reasons for 13% more events being observed at the SW buoy. In Table 2 we show a comparison of mean atmospheric variables between the two buoys, both for the high-shear cases and for the full data set. To perform a proper inter-comparison between the buoys, timestamps that are common for both buoys are only considered. We see in Table 2 that the local air temperature at the NE buoy is lower than the SW buoy. Furthermore, the change of air temperature between the buoys, T a,SW -T a,N E , is higher than the change of SST between the buoys, SST SW -SST N E .
Therefore, the lower air temperature at the NE buoy is largely responsible for its lower air-sea temperature difference relative to the SW buoy. This higher air-sea temperature difference at the SW buoy corresponds to notably lower TI and a slightly higher wind speed gradient across the rotor relative to the NE buoy, although the latter difference is small and could be within 190 the measurement uncertainty.

Very Low-Level Jets
Up to this point, the analysis considered high-shear events irrespective of the profile characteristics across a nominal rotor span.
Here, we focus on a subset of 10-minute periods that are interspersed within these high-shear events: those with a VLLJ. These events are of particular interest to wind energy applications as they subject the rotor not only to high shear, but also to negative 195 shear when the jet nose is within the rotor span.
Out of the 104 (92) high-shear events detected for the SW (NE) buoy, 30% (26%) feature VLLJs and 9% (7%) are made up entirely of VLLJ profiles. These profiles were not detected at any specific point of the high-shear events. Instead, they occurred at the beginning, end, and throughout the longer-lived events. A simple statistical analysis of these VLLJ profiles confirms that they are highly relevant for wind turbine operation: the most common nose wind speeds are between 9 m/s and 12 m/s, and the 200 most common nose heights 80 m and 100 m. As expected, the predominant wind direction during these VLLJ occurrences is consistent with that for the long-lived, high-shear events: primarily from the SW sector. These VLLJs exhibit a clear seasonal signature, being most frequently in spring and not occurring at all in winter (Fig. 11a). No clear diurnal signature can be identified (Fig. 11b), as is expected for the offshore environment where diurnal fluctuations are less pronounced than in land.
The highest shear values seen throughout this year of measurements correspond to VLLJ profiles, as evidenced by the 205 pronounced tail of the VLLJ maximum-shear distributions in Fig. 12a. When the nose of the jet is within the rotor swept area, a portion of the rotor will experience negative shear. Here, we quantify how much of the rotor experiences negative vs. positive shear for each VLLJ profile using the turbine-jet relative distance parameter [ξ, Gutierrez et al. (2017Gutierrez et al. ( , 2019. These values are shown in Fig. 12b: -1 indicates entirely positive shear across the rotor, 0 half negative and half positive, and 1 entirely negative. This analysis reveals that the nominal rotor defined here experiences at least some negative shear during most of the VLLJ 210 profiles identified: less than 1% of VLLJs have ξ = −1. More than 50% of the VLLJ profiles identified have more negative than positive shear across the rotor (1 > ξ > 0). While the mean negative shear is not too high (i.e., ∆U /∆z=-0.024 s −1 for both buoys), the distribution reveals a noticeable tail where ∆U /∆z < -0.035 s −1 (Fig. 12c). While previous work (Gutierrez  et al., 2017) has found that negative shear can decrease loads on the wind turbine system (primarily at the nacelle and tower), the positive shear in these profiles has been directly linked to an increase in static and dynamic loads relative to a well-mixed 215 profile (Gutierrez et al., 2016). A recent study (Gutierrez et al. (2019)) investigated the symmetry in wind turbine loads when the rotor experiences half positive, half negative shear and found complex interplay between the tower, blades, and gravitational loads. The complexity of this aero-structural problem and the nature of these boundary layer profiles off the U.S. east coast highlight that more studies are needed to support the successful deployment of offshore wind turbines in the U.S. distribution of turbine-jet relative distance parameter for all VLLJ profiles (b); distribution of shear above VLLJ nose (between nose and local wind speed minimum measured above it), shown only for 10-minute periods with VLLJ profile (c).
The high-shear periods measured at the two sites had substantially lower turbulence levels than the remainder of the data.

220
This is exemplified in Fig. 13 where TI is given as a function of wind speed for all 10-minute periods without a high-shear profile (black) and those with a VLLJ profile (colors). Note that the monotonic shear profiles are not included here, but their turbulence distribution is similar than that of the VLLJ profiles. As expected, most of the data (the profiles not flagged as having high shear) follows a decreasing trend with wind speed up to a certain point, and then sees a slight increase as wind speeds go up again and generate mechanical turbulence. For example, the SW buoy goes from 5.9% TI at 8 m/s to 7.8% TI at 225 20 m/s. The same is not seen for the VLLJ-exclusive data: a TI value of 4.9% at 8 m/s decreases even further as the wind speed increases, to about 3.7% at 20 m/s. This is likely due to the surface-layer high shear: the wind speed increase at hub height does not necessarily translate to an equally large wind speed increase near the ocean surface. These low values of turbulence also suggest stable atmospheric stratification, which has been found to support LLJ formation not only on land but near the shore in the U.S. eastern coast Colle and Novak (2010). buoys. Distributions are shown separately for all 10-min periods without a high-shear profile (black) and those with a VLLJ profile (colored).
Only wind speed bins with at least ten VLLJ profiles are shown. Monotonic shear periods are excluded here for clarity.

Synoptic Overview
Our analysis to this point has demonstrated the frequency of extreme high-shear events that are caused by stable stratification induced by warmer air from the southwest flowing over colder mid-Atlantic waters. In this section, we examine the synoptic conditions that lead to the arrival of warmer southwest air.
Synoptic conditions during these high shear events generally consist of a surface low pressure system centered west of the 235 floating lidar locations and a region of high pressure to the east as depicted in Fig. 14a. The exact location of these pressure systems deviates from case to case but the general pattern holds resulting in a large southerly component to the near surface winds. The directional component of the wind speeds is an important feature as winds coming from the south typically results in warmer air being advected into the area. Additionally, winds with a southwesterly component may be coming from onshore and can contain much higher air temperatures due to stronger heating over land during the day. Further, the long fetch over the 240 ocean results in low turbulent conditions.
Of the 86 days that registered an event, nearly 75% were observed to have this general synoptic set up. This synoptic setup has been seen in previous studies pertaining to offshore low-level jets in the mid-Atlantic region such as Zhang et al. (2006), Colle and Novak (2010), Helmis et al. (2013), and Strobach et al. (2018). While these studies each provide different mechanisms for the low-level jet formation, the synoptic setups are generally consistent with each other. In most cases, the 245 cyclone to the west advances towards the east or northeast denoted by the blue arrow in Fig. 14a.
Many of the stronger events coincide with the western low pressure system strengthening and moving eastward as the pressure gradient ahead of the cold front tightens and increases the wind speeds over the floating lidars (see Fig. 14b). Of the 10 longest events (averaging 30 hours in duration), 7 exhibited a tightening of the gradient and increase in wind speed as the event progressed. Helmis et al. (2013) and Strobach et al. (2018) found a similar tightening of the pressure gradient during cases 250 of offshore low-level jets in the mid-Atlantic resulting in a strengthening of the wind speeds and shifting of the winds to contain a stronger westerly component. Interestingly, the western low pressure systems in the two longest events were associated with named winter storms (Isaiah and Ruth, respectively). In fact, 12 out of 16 named winter storms that impacted the East Coast were also associated with high shear events giving credence to the idea that strong low pressure systems over the CONUS may produce the synoptic setup required for these offshore high shear events. Expanding to consider the 25 longest events 255 (averaging 19 hours in duration) shows that only 12 exhibit this synoptic structure. This implies that while it is common in the longest events in this area, it may not be a good characterization of all events including those with a much shorter duration.
Lastly, many of the events end around the time of frontal passages as depicted in Fig. 14c. This can be seen in Fig. 8 ( however, five of these events are within the ten longest duration events. While this is clearly not applicable to the majority 265 of events, many events, especially those that are around six hours or less in duration, are difficult to determine how the event ends as the synoptic charts are output at six hour intervals. Other noticeable features that were seen in the synoptic charts around the time an event ended were stationary fronts or shortwave troughs (which are commonly associated with changes in wind direction but no, or slight, changes in temperature). Additionally, some events are considered to have "begun" or "ended" erroneously due to missing data either before or after the event, respectively. In these cases, it is not possible to determine the 270 physical process that produced or destroyed the high shear event.
There are no clear synoptic differences between the VLLJ events and monotonic shear events. This may be due to the limited observational height where jet noses above 180 m cannot be determined. It is possible that some events that are not considered VLLJs are, in fact, LLJs with noses above 180 m. Additionally, it is possible that only subtle differences in the air temperature, wind speed, and/or wind direction are able to augment the wind profile such that an LLJ nose develops, or doesn't develop, 275 below 180 m.
For the event days that did not display the setup illustrated in Fig. 14 (roughly one quarter of event days), 13% displayed synoptic conditions with a surface high pressure system over the mid-Atlantic region. A similar synoptic environment is found in a case study within (Nunalee and Basu, 2014) where daily low-level jets formed in coastal New Jersey under an area of high pressure centered over the mid-Atlantic states. Additionally, one event occurred as Tropical Storm Arthur approached 280 the lidars from the south off the coast of South Carolina and moved north-northeast. Wind directions, in this case, were from almost directly east, however, air temperatures became warmer than the sea surface temperature as the high shear event began.
From this, it becomes apparent that warm air advection over relatively colder water is an essential ingredient to the formation of these high shear events that is typically caused by flow with a large southerly component. Figure 14. A simplified schematic of the synoptic conditions for high shear events at the beginning (a), during (b), and as the event ends (c). Grey lines represent theoretical isobars, arrows represent typical wind directions, speed, and relative air temperature to the floating lidars (green star), L and H represent low and high pressure centers, respectively.

285
This study has revealed the frequent occurrence of extreme high shear events in US mid-Atlantic offshore wind lease areas.
These events were characterized based on data from two floating lidars recently deployed by NYSERDA. We identified approximately 100 high-shear events over a year, with some events lasting up to three days. The magnitude of these events was striking, with maximum and mean hub-height wind speeds of 33 m/s and 16 m/s, respectively, and maximum and mean of power-law wind shear exponent across the rotor of 0.82 and 0.28, respectively. These values are substantially higher than 0.2, 290 the number proposed in the design standards to identify extreme shear conditions relevant to turbine operation IEC (2019). It is clear that once wind farms are built in these areas, these extreme events will have substantial effects on wind turbine power generation and structural response.
Fortunately, these extreme events seem to be fairly predictable. We found that their occurrences were strongly associated with a positive air-sea temperature difference, which occurs when warmer air from the southwest flows over the colder waters 295 of the mid-Atlantic, thereby inducing a stable stratification. These events largely occurred in spring and early summer when the air-sea temperature difference was greatest, and very seldom in fall and winter when the air-sea temperature difference is the lowest. The atmospheric conditions leading to these high-shear events is consistent with previous work (Colle and Novak, 2010;Zhang et al., 2006), which had attributed offshore LLJs closer to the coast. The measurements analyzed herein reveal that the high shear and jets persist further from the coast, at offshore distances where wind development is planned.

300
The high-shear events were characterized by low turbulence: ∼ 4.7% TI on average, in contrast to 8.1% when all the data are considered. We note, however, that the accuracy of TI measurements from the floating lidars was not assessed in this study.
Future work examining such accuracy would be valuable, provided of course that high frequency wind speed measurement by the floating lidar is made available.
The VLLJ events were especially notable, given their dominant nose heights of 80 m and 100 m and the impact such profiles 305 will have on turbine power generation. Although these events were fairly infrequent, this fact likely has more to do with the upper limit of 200 m from the lidar measurements. Had measurements been available above this height, it is likely that many of the identified monotonic shear events may actually be LLJs with noses above 200 m. Given increasing wind turbine hub heights and rotor diameters (e.g., the IEC 15-MW reference turbine with blade tips extending up to 300 m), further analysis of LLJs above 200 m is warranted.

310
In identifying these events, we relied on the wind speed gradient, ∆U /∆z, rather than the industry standard power law exponent, α (IEC, 2019). The α parameter is non-dimensional and does not consider the magnitude of wind speeds. Consequently, we found that extreme wind shear events could have low values of α while, conversely, low magnitude wind speed events could have high values of α. These results suggest revisiting the standard use of α in turbine design standards and the consideration of alternative parameters such as ∆U /∆z.

315
The public availability of floating lidar data was crucial for this analysis. Although many floating lidars are currently deployed in U.S. offshore wind areas, most data are kept confidential and not available for these types of analyses. Moving forward, future availability of additional floating lidars will be valuable in further characterizing the regional differences in extreme wind shear events and how they depend on factors such as proximity to the coastline, latitude, and seasonal changes in SST. Furthermore, these floating lidars will become vital in validating NWP models in offshore wind areas, especially their 320 ability to accurately predict these high shear events.