1. Introduction
Convective wind events, often defined as winds ≥ 35 kt (1 kt ≈ 0.51 m s−1; 40 mph) generated by a convective storm, occur over ocean and land. Those that occur over ocean can present a serious threat to maritime and aviation activities, especially if they occur without advanced warning. Given the above motivation, this study builds upon a prior analysis that used the Advanced Baseline Imager (ABI) and Geostationary Lightning Mapper (GLM) on board the GOES-16 satellite to analyze 13 ocean-based wind events detected by National Data Buoy Center (NDBC) buoys and Coastal-Marine Automated Network (C-MAN) stations as described in Thompson et al. (2021, hereafter T21). Here, Weather Research and Forecasting (WRF) Model–simulated characteristics of the corresponding ocean-based wind events are analyzed as a means of gaining a deeper understanding into the in-cloud processes associated with over-ocean wind events, beyond what was determined from ABI cloud-top brightness temperature trends and GLM lightning. The specific processes of interest include buoyancy, precipitation structures, and lightning.
T21 discussed how use of high temporal resolution data, specifically 1-min data, allows for increased observation of rapidly evolving and short lived weather events. Similar to the 1-min cloud-top brightness temperature (CTTB) and lightning characteristics of the ABI and GLM data, respectively, analyzed in T21, here, 1-min WRF cloud-top temperature (WCTT) and estimated lightning characteristics provided by the WRF Model simulations will be analyzed. As a means of defining a simulated wind event, each WRF wind event was required to have a negative vertical velocity (downdraft) of at least 10 m s−1. In addition to WCTT and lightning characteristics, as noted, the WRF Model output also provided in-cloud fields such as depth and cloud and precipitation mixing ratio values, as well as environmental characteristics associated with the wind events.
This paper examines the hypothesis introduced in T21, that prior to a wind event, there should be unique signatures in CTTB (here, WCTT) trends and lightning data that can be exploited to increase the situational awareness of pending oceanic wind events. Based on in-cloud processes and prior research, the presumption is that the coldest CTTB/WCTT (i.e., highest cloud tops) and peak lightning flash rates should occur nearly at the same time and prior to the occurrence of a wind event (Goodman et al. 1988). The science questions to accompany the hypothesis include:
-
Do the environmental characteristics of ocean-based wind events closely resemble those associated with land-based wet downbursts, or does a dry air layer influence their occurrence?
-
What are the main driving in-cloud mechanisms of ocean-based wind events?
-
If the coldest WCTT and peak estimated flash rate do not occur at the exact same minute, does one extreme consistently occur first?
-
Are trends in the WCTT and estimated lightning more useful than their specific magnitudes?
These questions, along with the hypothesis, will be discussed when explaining findings from the WRF Model–simulated ocean-based wind events. Unique signatures in the data can help operational weather forecasters identify in advance regions where ocean-based wind events are likely, thereby decreasing the threat to maritime and aviation activities.
The paper proceeds as follows: A brief background of convective wind events, specifically those analyzed with model simulated data, are provided in section 2. Then, a description of the WRF Model is provided in section 3 and the methodology is given in section 4. Finally, findings of the 48 WRF Model–simulated ocean-based wind events are discussed in section 5, followed by conclusions in section 6.
2. Background
As explained in T21, one specific type of convective wind event is the well-known and well-documented downburst, which is defined as a strong downdraft that causes an outburst of damaging winds, straight or curved and highly divergent, on or near the surface (Fujita and Byers 1977; Fujita and Wakimoto 1981; Fujita 1985). Downbursts can be further classified based on the amount of precipitation within them and their spatial extent (Fujita and Wakimoto 1983; Wakimoto 1985). Wet downbursts have at least 0.25 mm of rain near the time of occurrence, while dry downbursts have less than 0.25 mm of rain near the time of occurrence. Fujita and Wakimoto (1983) and Wakimoto (1985) considered “near the time of occurrence” to be “between the onset and end of the high winds, including intermediate calm periods, if any.” Melting, precipitation loading, and/or entrainment of dry midlevel air typically drive wet downbursts, while evaporative cooling and/or sublimation typically drive dry downbursts. A combination of wet and dry downburst mechanisms contribute to hybrid downbursts (Ellrod 1989; Pryor 2015). Based on spatial extent, a 4-km horizontal extent threshold of wind damage separates smaller downbursts (microbursts) from larger downbursts (macrobursts). Recall that most convective wind event studies have focused on land-based events, which is additional motivation for the present study.
Some downburst studies used models in their analyses. Knupp (1989, 1996) discussed model results of negative buoyancy in low levels being aided by cooling due to melting and evaporation. Barthe et al. (2010) applied adjustment factors to six storm parameters from WRF Model runs that reproduced the structure of a severe storm and an airmass thunderstorm. Maximum updraft velocity, which did not directly depend on storm volume, morphology, and microphysical parameterizations, and therefore was less dependent than ice mass flux, precipitation ice mass, cloud-top height, ice water path, or updraft volume on how well the model simulated storm characteristics, most reliably predicted the lightning rate (and better for the severe storm than the airmass thunderstorm); precipitation ice mass was the second most reliable parameter in predicting lightning rate. Srivastava (1985, 1987) used 1D time dependent models to observe driving mechanisms of dry and wet downbursts (a downdraft magnitude of at least 20 m s−1). With wet downbursts, rather stable lapse rates near moist adiabatic needed to be accompanied by high precipitation concentrations. Ice was favored over water to increase downdraft strength; small ice could completely melt as it descended a few kilometers through temperatures above 0°C, while similar size rain did not completely evaporate. With dry downbursts, unstable lapse rates near dry adiabatic required very little precipitation concentrations. Note that entrainment was not considered with wet downburst model simulations, but decreased the intensity of dry downburst model simulations.
Other studies have also used models in their downburst studies, with some related to downbursts observed during field campaigns (Proctor 1988, 1989; Tuttle et al. 1989; Hjelmfelt et al. 1989; Straka and Anderson 1993). Many studies noted the downburst driving mechanisms. The two microbursts simulated in the study by Proctor (1988) were primarily driven by evaporative cooling of rain and cooling as a result of melting hail. Tuttle et al. (1989) also investigated downburst driving mechanisms, and found that mixing ratio values supported mass loading and cooling due to melting. Hjelmfelt et al. (1989) noted contributions from mass loading, evaporation, and melting. Straka and Anderson (1993) noted a peak in the graupel/hail mixing ratio roughly 6–12 min prior to the strongest surface outflows and formation of low-level downdrafts. They also noted that with 250-, 500-, and 1000-m model horizontal grid resolution, the strongest maximum surface outflow differentials were found in the highest resolution, although both the 250- and 500-m model horizontal grid resolutions likely adequately simulated the general storm evolution characteristics. They showed that downward acceleration (up to 5 km) due to evaporation was greatest near the surface and decreased with height, hydrometeor loading was greatest near 5 km (also near the melting level) and decreased with height, and melting peaked near 3.5 km. Note that the driving mechanisms in the model studies above were similar to driving mechanisms explained in observation studies such Goodman et al. (1988) and Carey and Rutledge (1996) discussed in T21. With an overland microburst, Goodman et al. (1988) found that the event was likely driven by precipitation loading and melting that led to low-level cooling. Carey and Rutledge (1996) noted an association between melting of graupel and hail and microburst occurrence with use of dual-polarization radar data.
As mentioned in T21 and also relevant here, some studies focused on specific parameters or explained calculations used to predict wind events (Proctor 1989; Caplan et al. 1990; Wolfson 1990; Atkins and Wakimoto 1991; McCann 1994; Ellrod et al. 2000; Geerts 2001; Pryor and Ellrod 2004a,b; Kuchera and Parker 2006; Pryor 2015; Púčik et al. 2015; Grundstein et al. 2017). Those parameters included temperature, dewpoint, lapse rate, height of the melting level, mixing ratio, midtropospheric wind, equivalent potential temperature (θe), convective available potential energy (CAPE), and downdraft convective available potential energy (DCAPE). Recall that Lucas et al. (1994) found that on a sounding, the positive area between the level of free convection (LFC) and equilibrium level (EL), or the CAPE, tended to be “skinny” over the ocean and “fat” over land. They noted that with skinny CAPE, 4°–5°C of buoyancy was maintained through a large part of the troposphere and with fat CAPE, 8°–9°C or more of buoyancy could be achieved, but that compared to ocean soundings, the LFC was high and EL was low. Atkins and Wakimoto (1991) noted that the difference between surface or just above the surface values of θe and the minimum aloft values of θe exceeded 20 K in the afternoon on days associated with wet microbursts during a field campaign in Alabama. They noted that the difference was less than 13 K for days without a wet microburst. They also noted warm cloud bases, shallow subcloud layers, and low-level moisture capped by a midlevel dry layer associated with wet microbursts. Larger thermal instability for days with wet microbursts was seen in larger CAPE values, than the values on days without a wet microburst. Large DCAPE values, or the maximum energy available to a descending parcel, can indicate an increase in the potential for strong downdrafts and damaging outflow winds (Gilmore and Wicker 1998; Grundstein et al. 2017; NWS DCAPE 2023; NOAA SPC 2023). For additional explanations and references about differences between the vertical development of convective storms over land and over ocean, differences between lightning that occurs over land and over ocean, and correlations between cloud-top temperature and lightning characteristics, see T21.
3. Data source: WRF Model
The WRF Model is a numerical weather prediction (NWP) system with two dynamical solvers: the Advanced Research WRF (ARW) core and the Nonhydrostatic Mesoscale Model (NMM) core (Skamarock et al. 2008; NCAR 2017). This study used Version 3.8.1 of the WRF-ARW Model, which solves a set of fully compressible, nonhydrostatic (with a run-time hydrostatic option) governing equations, with 0.5° domain Global Forecast System (GFS) analysis data from the National Centers for Environmental Information (NCEI) website every 6 h as input, to simulate what the likely atmospheric conditions were on select days.
Each model simulation produced four domains, which moving from the outermost to innermost domain had temporal and horizontal resolutions of 15 min and 13.5 km, 10 min and 4.5 km, 5 min and 1.5 km, and 1 min and 0.5 km, respectively. With each model simulation, the two outermost domains were initiated 12 h prior to the two innermost domains, in order to provide enough spin up time. The two outermost domains provided 30–60 h of output data, while the two innermost domains provided 18–48 h of output data. Data from the innermost domain, with the 1-min and 0.5-km temporal and horizontal resolutions and explicitly resolved convection, were the focus of this study. The 1-min temporal resolution was chosen to match that of the ABI data used in T21. Each model simulation used a model top of 50 hPa (∼20 km), with 47 staggered dimension vertical levels (e.g., geopotential height) and 46 unstaggered dimension vertical levels (e.g., pressure). The WRF Model physics options used in each model simulation are provided in Table 1. Note that when Bolgiani et al. (2020) analyzed microbursts using various combinations of physics schemes, they felt that each combination produced the general characteristics found from observations. This study used variables provided directly in the WRF output, as well as variables that were calculated from the direct output. Such 3D variables include velocity (u, υ, and w), temperature, dewpoint, equivalent potential temperature, mixing ratios (rainwater, cloud water, graupel, snow, ice, and water vapor), pressure, and geopotential height. A WRF-estimated flash rate was also used (McCaul et al. 2009) as a means of estimating lightning flash rates associated with the wind events, as discussed below.
Model physics used with each WRF Model simulation.
4. Methodology
a. WRF Model settings
The first step with WRF Model–simulated wind events was to establish the WRF Model domains centered on the 13 observed wind event cases that were discussed in T21. Each WRF simulation was based on the date, time, and location of the buoy and C-MAN station observed wind events. Since the 13 buoy and C-MAN station observed wind events occurred over eight separate days, this resulted in eight separate WRF Model simulations, with the domain settings as described above. Each model simulation used the same model physics options (Table 1). The actual WRF Model UTC start time was based on the wind event time as observed by the buoy or C-MAN station. Noting that the time of the WRF Model–simulated wind event likely will be offset in time and space from the observed event, the (two innermost domain) model simulations were started at least 6 h prior to the time of the observed event, and ended at least 6 h after the time of the observed event. Furthermore, the 12 h of additional simulation time for the two outermost domains allowed enough forecast spin-up time prior to providing output for the higher resolution innermost domains. Since an offset in location was also possible, the domains extended beyond the location of the buoy or C-MAN station. Extending the time and location also allowed for the detection of multiple wind events associated with each observed wind event. Unlike the fixed location and sparse density of buoys and C-MAN stations, the WRF Model was able to observe the entire ocean area in each domain.
b. Model-based cell identification and tracking
Once the model simulations completed, domain 4 grid boxes were searched to locate negative vertical velocity values of at least 10 m s−1 in the lowest 2 km over the ocean. Vertical velocity was one of the 3D variables provided directly with the model simulation output. The temporal resolution of 1 min with domain 4 matched that of the ABI data used in T21, although the 0.5-km convection resolving horizontal resolution with the WRF data were higher than the 2-km resolution with band 13 ABI data. The largest negative vertical velocity values, associated with separate cells and separated by time and/or distance, were then taken as possible wind events given they represented possible evolving microbursts. This often led to multiple WRF Model–simulated events being associated with a single buoy or C-MAN station observed event.
To determine WCTT values, the location of cloud tops first needed to be identified. This study used the highest vertical location with 15 dBZ as the cloud top. With their WRF Model output, Monte et al. (2019) used thresholds of 0.2 for cloud fraction and 0.002 for optical depth to find the highest vertical level with a “cloud,” which was then considered the cloud top. Chae and Sherwood (2010) defined “cloud top” to be the location in a column where the cumulative water content from above reached more than 0.2 g kg−1, which they noted was roughly unit optical depth. In this study, identifying the cloud top as the highest altitude with a specified small cloud water mixing ratio value likely would have missed large amounts of cloud regions. Some columns with an identified cloud top based on the highest altitude of 15 dBZ would have been considered cloud-free if the cloud water mixing ratio was used to identify the cloud top because the column had cloud water mixing ratios of zero at each level even though reflectivity values as high as about 50 dBZ existed at some altitude throughout the column, positive rainwater mixing ratios existed in the lower levels, and positive graupel, snow, and ice mixing ratios existed in the middle levels. More grid boxes that were classified as cloud regions even though the cloud water mixing ratio was zero at all levels, existed later in the tracked cell’s life. Some grid boxes early in a tracked cell had positive cloud water mixing ratios, which then turned to values of zero later in the cell’s life. In such cases, cloud water mixing ratios might have been converted into other mixing ratios like rainwater as the cell matured and decayed, kind of like a “balancing act” with all of the mixing ratios (Bao et al. 2019). It is noted that using 15 dBZ as cloud top likely underestimates cloud-top altitude (and therefore also underestimates—provides warmer—WCTT values). This study considered using the highest vertical location with an ice mixing ratio of 0.05 g kg−1 or the highest vertical location with an ice mixing ratio plus cloud water mixing ratio of 0.05 g kg−1 to identify the location of the cloud top and determine the WCTT. With two events, it was found that using the values based on the ice mixing ratio plus the cloud water mixing ratio expanded the extent of the warmest cloud area (e.g., seen in contour plots, not shown), but that the time series plots (not shown) generated from the coldest WCTT temperatures did not change when the ice mixing ratio plus cloud water mixing ratio was used instead of just the ice mixing ratio. The largest differences between WCTT based on the highest vertical location of 15 dBZ or WCTT based on the highest vertical location with an ice mixing ratio of 0.05 g kg−1 occurred more than 15 min after the time of the wind event, and after the time most useful to provide advance notice of an impending wind event. With the two events, for minutes with data from T − 5 through T + 5, the largest temperature difference was 1.2 K and the largest percent difference was 0.6% when looking at the WCTT values when using the highest vertical location of 15 dBZ or the highest vertical location with an ice mixing ratio of 0.05 g kg−1. It is also noted that for both events, the coldest temperature occurred at the time of the wind event, time T, when using both the highest vertical location of 15 dBZ and the highest vertical location with an ice mixing ratio of 0.05 g kg−1 to determine WCTT. At time T, the temperature difference was 0.0 K and the percent difference was 0.0% with both events.
Here, the 3D reflectivity values were directly computed in the model simulations (NCAR 2017). Note that direct 3D output from the model simulations included perturbation potential temperature, perturbation and base pressure, and perturbation and base geopotential. From those direct output values, potential temperature (θ) was calculated as the perturbation potential temperature plus 300 K, p was calculated as the sum of the perturbation pressure and base pressure, and geopotential height (Zg) was calculated as the sum of the perturbation geopotential and base geopotential, divided by 9.81 (NCAR 2017). Also note that the directly outputted u and υ components of velocity, reflectivity, and mixing ratios, along with pressure had 46 vertical levels, while the w component of velocity, along with geopotential height had 47 vertical levels. When geopotential height needed to match with variables with 46 levels (i.e., staggered dimensions versus unstaggered dimensions), the geopotential height was taken as the sum of the geopotential height below and geopotential height above, divided by 2. Temperature (T) values were calculated by rearranging Poisson’s equation. To find temperature at the cloud top, first the highest altitude with 15 dBZ had to be identified. When this occurred between two levels, it was assumed a linear trend existed and 101 steps were applied. The step closest to 15 dBZ was taken as the cloud-top location, and Zg, p, and θ at the step were also calculated based on a linear trend. The temperature at 15 dBZ was then considered the WCTT, and those WCTT values were contoured so that cell identification and tracking could then be done.
WRF-based cells were then identified at the time of the event, T, and tracked from T − 45 through T + 30, or for their duration if less than those 76 min of interest, as explained with ABI data for the 13 buoy and C-MAN station observed events in T21. Unlike with ABI data, a cell boundary was determined each minute with the WRF data. The cell boundary enabled grid boxes within the cell to be identified, which then allowed for in-cloud characteristics to be analyzed and a WRF-based estimated flash rate to be calculated (explained next). A maximum WCTT value was used as the cell boundary. Based on cell differences among the events, no single maximum WCTT value fit all convective cells associated with the 48 events. The most common maximum WCTT value used as a boundary was 220 K, followed by 240 K (e.g., colder cell within a larger cloud region versus more isolated, warmer cell), although 230 and 260 K were also used. If the cell of interest was isolated from other cells, a cell boundary could be found by providing the tracking software: (i) the maximum allowed temperature, and (ii) one grid box location within the cell each minute, which then automatically searched and found the cluster of adjacent grid boxes that formed the cell. When the cell of interest was not well isolated, additional manual assessment was needed to identify the separation between cells.
Once the grid boxes within the cell boundary were known, cell-specific values were calculated. The cold convective cell-top WCTT each minute was defined as the average value of the coldest 5% (rounded to the nearest number) of the WRF grid boxes within the cell. If a cell covered less than 20 grid boxes, the WCTT was the value of the coldest grid box. Using a WCTT based on the coldest 5% instead of a single value each minute decreased the impact from any outliers and calculated a value based on a coarser resolution (an average of multiple individual 0.5-km horizontal resolution grid box values), more similar to that of the 2-km ABI band 13 CTTB values (essentially an average of 16 0.5-km WRF Model grid boxes) in T21.
c. WRF Model parameterized lightning
It is noted that the WRF Model setup used in this study did not exactly match that used in McCaul et al. (2009). McCaul et al. (2020) looked at 12 different WRF Model configurations of four microphysics schemes and three planetary boundary layer (PBL) schemes to determine how different configurations impacted the generated lightning. They ran the model on a 4 km × 4 km nested mesh and had 42 vertical terrain-following levels. Although one of the PBL schemes was the Mellor–Yamada–Janjic scheme, they did not include the WDM6 microphysics scheme. McCaul et al. (2020) did include the WRF single moment 6-class (WSM6) microphysics scheme, and they found that most configurations did not produce “statistically significant” differences compared to the reference scheme; the two experiments that had statistically significant differences used Mellor–Yamada–Nakanishi–Niino (MYNN) PBL physics.
d. Environmental variables
Environmental characteristics prior to the wind events were examined with WRF-based sounding data. The soundings were taken in the vicinity of the wind event, about 1–3 h prior to occurrence. The location of the sounding was required to be cloud-free at the time of and leading up to the time when the data were collected. A cloud-free region was identified as grid boxes with no WRF-detected reflectivity at any level. The variables used as input for the soundings (pressure, geopotential height, temperature, dewpoint, wind speed, and wind direction) were taken as the average of the values in a cluster of nine adjacent grid boxes (3 grid boxes × 3 grid boxes). The WRF Model provided p and Zg information at each grid box, and the T and Td at each grid box was calculated as described above. Horizontal wind speed was calculated from the u and υ components of the wind, which were directly provided in the WRF output. The wind direction was calculated as arctan(−u, −υ). By providing Zg, p, T, Td, wind speed, and wind direction to the SHARPpy program, a skew T diagram figure and values for lapse rate [surface to 3 km above ground level (AGL), 3–6 km AGL, 850–500 hPa, and 700–500 hPa], precipitable water, mixed layer (ML) CAPE, the height of the LFC and EL, and DCAPE were provided through their built-in calculations (Blumberg et al. 2017).
5. Results
The WRF Model–simulated 48 oceanic wind events that corresponded to the 13 buoy and C-MAN station observed oceanic wind events that were discussed in T21. This section will discuss the findings from the WRF Model–simulated events, starting with the environmental characteristics outside the convective storms prior to the wind events, followed by the variables inside the cloud including the WCTT and estimated lightning trends for the combined 48 oceanic wind events. Next, the environmental characteristics outside the convective storms, followed by the analysis of the in-cloud variables, will be discussed for two representative wind events. The discussions for both the combined 48 events and the two representative events will also address how the findings relate to the guiding study hypothesis, along with the science questions that are listed in the introduction. Additional discussion related to the hypothesis and science questions is located in the conclusions section.
a. Combined WRF Model–simulated oceanic wind events
The 48 WRF Model–simulated wind events that corresponded to the 13 buoy and C-MAN station observed wind events from 2018 (discussed in T21) were also located in the eastern Gulf of Mexico and in the Atlantic Ocean from Florida northward through South Carolina (Fig. 1 and Table 2). The modeled negative vertical velocity values associated with all wind events ranged from 10.1 to 20.7 m s−1 in the lowest 2 km, fitting the definition of an over-ocean wind event.
(a) Location (denoted by the blue asterisk), date, time, and wind gust of the 13 buoy and C-MAN station-detected wind events explained in T21. Location names of five numbers represent buoys, and location names of four letters and one number represent C-MAN stations. The green numbers (1–48) represent the 48 WRF Model–simulated wind events and correspond to the event numbers in Table 2. The two orange numbers, 19 and 27, are the individual WRF Model–simulated wind events described in detail. The four domains for the model simulations associated with (b) WRF event 19 and (c) WRF event 27 show the areas with 13.5-km horizontal resolution in black, 4.5-km horizontal resolution in red, 1.5-km horizontal resolution in green, and 0.5-km horizontal resolution in blue. The locations of event 19 and event 27 are denoted by a purple asterisk in the corresponding figures.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-22-0020.1
The date, time, and peak negative vertical velocity of the 48 WRF Model–simulated wind events from 2018. The 48 events spanned different seasons, which relates to different convection characteristics. Note that the locations of the events are marked in Fig. 1. WCTT and WCTT T± indicate the value and time of occurrence of the coldest WCTT value associated with each wind event. The WRF-estimated flash rate and estimated flash rate T± indicate the value and time of occurrence of the peak WRF-estimated flash rate associated with each wind event.
1) Environment outside the cloud
For the 48 wind events, 34 separate WRF-based soundings were generated prior to the wind events. Table 3 lists values from each sounding, as well as from three composite soundings. Note that wind events that shared a sounding are marked with matching symbols next to the event number in the table. The composite sounding based on all events is shown in Fig. 2a. Note that this composite sounding includes events from different seasons, which can relate to different convection characteristics. Figure 2b is a composite sounding for the events that occurred in March and April, while Fig. 2c is a composite sounding for the events that occurred in June, July, August, and September.
The date and time of each WRF-based sounding. All of the ocean-based wind events occurred in 2018. Note that the events that shared a sounding have a matching symbol next to the event number. Values from the soundings include precipitable water (PW); lapse rates (LRs) for 850–500 hPa, 700–500 hPa, surface–3 km AGL, and 3–6 km AGL; the ML CAPE; DCAPE; and the largest θe difference. The composite soundings are based on events from all months (All), March and April (MA), and June–July–August–September (JJAS).
Composite sounding for the WRF Model–simulated wind events that occurred in (a) all months; (b) March and April; and (c) June, July, August, and September. Note that the composite sounding for all months includes events from different seasons, which can relate to different convection characteristics.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-22-0020.1
The composite sounding based on all months revealed more moisture near the surface with somewhat drier conditions near 300 hPa, exhibiting traits similar to a typical overland wet downburst sounding. Dewpoint depressions of 2.5°C existed at 968 and 953 hPa, while Td depressions of 7.9° and 13.9°C existed at 458 and 302 hPa, respectively (Table 4). The composite sounding had 1.81 in. (4.6 cm) of precipitable water. The range of precipitable water values for the individual soundings was 1.29–2.25 in. (3.3–5.7 cm). The smallest values were associated with events that occurred in March and April. The ML CAPE for the composite sounding was 1852 J kg−1, which revealed instability, and the LFC and EL were at 1119 and 13 036 m, respectively. Individual ML CAPE values ranged from 469 to 3585 J kg−1. The typically low altitude for the LFC and high altitude for the EL led to skinny CAPE, which Lucas et al. (1994) found with over-ocean soundings. The composite sounding DCAPE was only 648 J kg−1. That was not an extremely large value, especially when compared to overland downburst cases, suggesting there was limited negative buoyancy production potential from evaporation. Individual sounding DCAPE values ranged from 312 to 965 J kg−1. Although a few individual soundings did have a DCAPE value that exceeded 800 J kg−1, none of the individual soundings had a DCAPE value that exceeded 1000 J kg−1. For overland storms, a DCAPE value that exceeds 800 J kg−1 is typically considered “decent” or “nominal to high,” while a value that exceeds 1000 J kg−1 is typically considered “significant” and associated with low-level lapse rates that help transport high momentum air to the surface and results in possible strong winds (Gilmore and Wicker 1998; Grundstein et al. 2017; NWS DCAPE 2023; NOAA SPC 2023). The composite sounding lapse rates for 850–500 hPa, 700–500 hPa, the surface–3 km AGL, and 3–6 km AGL were all near the moist adiabatic lapse rate, at 6.3°, 6.2°, 6.7°, and 6.2°C km−1, respectively. Recall that a typical overland wet microburst had lapse rates near moist adiabatic. With the composite sounding, a maximum near surface θe of 348.2 K at 1012 hPa and a minimum aloft θe of 328.0 K at 619 hPa (and 586 hPa) resulted in a θe difference of 20.2 K. This difference slightly exceeded the 20-K threshold noted by Atkins and Wakimoto (1991) for days with wet microbursts. Atkins and Wakimoto (1991) noted that days without wet microbursts had θe differences less than 13 K. From the individual soundings, 25 of the 34 soundings had θe differences that exceeded 20.0 K. The remaining nine soundings had θe differences that were greater than 13.0 but less than 20.0 K. The range of values was 13.9–29.1 K. With the composite sounding, the strongest winds existed near 200 hPa, with values near 50 kt. Near the surface, winds were 15–19 kt (7.7–9.8 m s−1).
Additional environmental characteristics of event 19, event 27, and the composite soundings based on events from all months (All), March and April (MA), and June–July–August–September (JJAS). Note that with the dewpoint depression aloft, two values are listed for each sounding. Those dewpoint depression values, and the corresponding pressure level for each value, are separated by a slash (/).
2) Variables inside the cloud
Based on all 48 wind events, the coldest WCTT occurred at an average of T − 4.2 and a median of T − 3, with a range from T − 23 to T + 7 (Table 2). The peak WRF-estimated flash rate occurred at an average of T − 1.1 and a median of T − 1, with a range from T − 13 to T + 16. With 36 of the events, the peak estimated flash rate occurred within ±5 min of the coldest WCTT. Related to a science question listed in the introduction, out of the combined 48 events, 42 events, or 87.5%, had the coldest WCTT occur before or at the same time as the peak estimated flash rate, while six events, or 12.5%, had the coldest WCTT occur after the peak estimated flash rate (Table 2). Due to the method used to calculate the WRF-estimated flash rate (recall that it looked at the likelihood of lightning occurring at each grid box), even wind events with low estimated peak flash rates had rather “smooth” time series plots. Recall that some GLM-based flash rate time series from T21 often had “jagged” lines due to flash rates changing from 0 to 1 or a few flashes from minute to minute. With the WRF-estimated flash rate, many minutes had non-integer values, caused by the calculation [Eq. (2)] used to determine lightning amounts, which led to smoother plots. The time series plots support the hypothesis that the coldest WCTT and peak flash rate should occur at the same time, with decreasing WCTT trends corresponding to increasing estimated flash rate trends and increasing WCTT trends corresponding to decreasing estimated flash rate trends. Although many time series plots also support the hypothesis that the coldest WCTT and peak flash rate should occur prior to a wind event, some coldest WCTT and peak estimated flash rate values occurred at the time of the wind event or even after the wind event.
Box-and-whisker plots of WCTT, changes in WCTT, and WRF-estimated flash rate each minute for the 48 wind events are shown in Fig. 3, where the blue line reveals the mean value each minute. Note that with the WRF Model–simulated events, T − 27 was the earliest a cell was tracked. Many WRF-based cells were shorter lived, with colder minimum WCTT values than the cells tracked in ABI data from T21. This is likely related to the higher (convection resolving) resolution of 0.5 km with WRF-data, compared to the coarser 2-km resolution of band 13 ABI data. Like with ABI and GLM data, some changes in WCTT are subtle. The WCTT does show a decrease in temperature leading up to the wind event time, followed by an increase in temperature, while the estimated flash rate values exhibit an increase leading up to the wind event time, followed by a decrease.
Box-and-whisker plots of (top) WCTT, (middle) change in WCTT, and (bottom) WRF-estimated flash rate for the 48 wind events. The lower quartile and upper quartile are marked by the ends of the gray boxes plotted each minute. The median is marked by the black line inside each gray box. The minimum and maximum values are marked by the lines that extend from the ends of the gray boxes. The mean value is marked by the blue line. The x axis is from T − 30 through T + 30. The time T − 0, which is highlighted by the yellow box, is when the wind events occurred.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-22-0020.1
Time series plots of cloud depth, based on the LCL as cloud base and the highest altitude with 15-dBZ reflectivity as cloud top, revealed that cloud depth typically increased as the WCTT decreased. The maximum depth occurred at an average of T − 2.9 and median of T − 2.5, with a range from T − 18 to T + 9. Recall that the coldest WCTT occurred at an average of T − 4.2. A comparison of the box-and-whisker plot of cloud depth in Fig. 4 and the box-and-whisker plot of WCTT in Fig. 3, shows the general trend of a decrease in WCTT corresponding to an increase in cloud depth, and an increase in WCTT corresponding to a decrease in cloud depth. The cloud base for each event was low, with only five events having a cloud base that exceeded 1 km during its lifetime. This implied a moist environment, indicative of the humid subtropical to tropical environments the oceanic wind events occurred in, and that most of the available CAPE was being used by convective clouds. The low cloud bases in model simulated wind events resemble the low cloud bases often associated with overland wet downbursts. The cloud-top trends were very similar to the cloud depth, with maximum cloud tops occurring at an average of T − 3.1 and median of T − 2.5, with a range from T − 18 to T + 9. WCTT values decreasing as the cloud-top altitude increased shows that as an updraft moved the cloud to higher altitudes within the troposphere, it reached colder temperatures. The box-and-whisker plot of maximum depth from 0° to −15°C shows that the depth tended to decrease leading up to a wind event (Fig. 4). As the depth decreased, it is likely that less charge separation existed, which would relate to a decreasing flash rate. A decrease in the depth from 0° to −15°C prior to a decrease in overall cloud depth could relate to the downdraft forming within the cloud as an updraft was still present and continuing to move the cloud top to higher and colder altitudes, such as during the mature stage of a thunderstorm.
Box-and-whisker plots of (top) the maximum cloud depth based on the LCL as cloud base and uppermost altitude with 15-dBZ reflectivity as cloud-top and (bottom) the maximum depth of the layer from 0° to −15°C for the 48 wind events. The lower quartile and upper quartile are marked by the ends of the gray boxes plotted each minute. The median is marked by the black line inside each gray box. The minimum and maximum values are marked by the lines that extend from the ends of the gray boxes. The mean value is marked by the blue line. The x axis is from T − 30 through T + 30. The time T − 0, which is highlighted by the yellow box, is when the wind events occurred.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-22-0020.1
Box-and-whisker plots of the maximum (a) qT, (b) qr, (c) qc, (d) qg, (e) qs, and (f) qi from the surface to cloud top for the 48 wind events. The values are normalized based on the number of levels from the surface to cloud top. The uppermost altitude of 15-dBZ reflectivity was used as cloud top. The lower quartile and upper quartile are marked by the ends of the gray boxes plotted each minute. The median is marked by the black line inside each gray box. The minimum and maximum values are marked by the lines that extend from the ends of the gray boxes. The mean value is marked by the blue line. The x axis is from T − 30 through T + 30. The time T − 0, which is highlighted by the yellow box, is when the wind events occurred.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-22-0020.1
With the three frozen hydrometeors in the region from 0° to −15°C, the largest contribution came from graupel. With the box-and-whisker plots, the sum of the frozen hydrometeors and graupel alone showed very similar trends (Fig. 6). A slight increase occurred first, with decreasing values beginning at about T − 6. Snow revealed a slight increasing trend leading up to and through the time of the wind event, while ice remained more constant with time. As these frozen hydrometeors melted, they contributed to the negative buoyancy needed for the downdraft. The presence of frozen hydrometeors also supports cloud charging and lightning occurrence, consistent with the lightning analysis above.
Box-and-whisker plots of the maximum (a) qF, (b) qg, (c) qs, and (d) qi within the layer from 0° to −15°C for the 48 wind events. The values are normalized based on the number of levels within the layer from 0° to −15°C. The lower quartile and upper quartile are marked by the ends of the gray boxes plotted each minute. The median is marked by the black line inside each gray box. The minimum and maximum values are marked by the lines that extend from the ends of the gray boxes. The mean value is marked by the blue line. The x axis is from T − 30 through T + 30. The time T − 0, which is highlighted by the yellow box, is when the wind events occurred.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-22-0020.1
b. Individual WRF Model–simulated oceanic wind events
Two representative wind events will be highlighted in the WRF Model analyses: one that occurred at 0126 UTC 11 April 2018 over the Gulf of Mexico and a second that occurred at 1643 UTC 3 June 2018 over the Atlantic Ocean. Both wind events exhibited representative WCTT and WRF-simulation estimated lightning trend signatures relative to an oceanic wind event.
1) 0126 UTC 11 April 2018
Wind event 19 occurred at 0126 UTC 11 April 2018 over the Gulf of Mexico, with a vertical velocity of −19.3 m s−1 (Fig. 1 and Table 2). This was the third strongest wind event and was associated with the wind event that occurred at buoy 42003 at 2150 UTC 10 April 2018 (discussed in T21). The cell of interest was tracked from T − 10 through T + 25.
A WRF Model–based sounding from 0000 UTC 11 April 2018, 1 h 26 min before the time of the wind event, is shown in Fig. 7. Specific values are also provided in Tables 3 and 4. Similar to the composite soundings, this event also exhibited characteristics often found with overland wet downbursts, including moisture near the surface and a somewhat drier region aloft, along with near moist adiabatic lapse rates.
WRF-based sounding at 0000 UTC 11 Apr 2018. This sounding is associated with event 19, which occurred at 0126 UTC 11 Apr 2018.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-22-0020.1
The cell of interest was first tracked at T − 10 when the WCTT was 211.8 K (Fig. 8). The WCTT decreased as the cell moved toward the wind event location, and reached 203.7 K at T − 5, followed by its coldest value of 198.2 K at T − 2 (Fig. 9, Table 2). As the WCTT decreased, the WRF-estimated flash rate increased (Fig. 8). The estimated flash rate increased from 0.2 flashes per minute at T − 10 to the peak value of 27.6 flashes per minute at time T, 2 min after the coldest WCTT. After the wind event at time T, the WCTT continued to warm and the estimated flash rate continued to decrease.
Time series plot of WCTT in black and WRF-estimated flash rate in gray for the WRF Model–simulated wind event at 0126 UTC 11 Apr 2018. The wind event occurred at T − 0, which is highlighted by the yellow box.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-22-0020.1
WCTT at (a) T − 5, (b) T − 2, (c) T, and (d) T + 5 for the WRF Model–simulated wind event that occurred at 0126 UTC 11 Apr 2018. The location of the wind event is marked by the large purple asterisk in the center of each image. The (smoothed) cell of interest is enclosed by the gray outline. The coldest WCTT of 198.2 K occurred at T − 2.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-22-0020.1
As the WCTT decreased and therefore the cloud-top height increased, the cloud depth based on the distance between the LCL and uppermost altitude with 15-dBZ reflectivity, increased (Fig. 10). The (maximum) cloud depth began at 12.4 km at T − 10, with cloud bases based on the LCL from 0.173 to 0.203 km and cloud tops based on the uppermost altitude with 15 dBZ from 12.5 to 12.6 km. The cloud depth increased to 14.2 km at time T, 2 min after reaching the coldest WCTT and at the same time as the peak estimated flash rate. At time T, cloud-base altitude ranged from 0.085 to 0.251 km and cloud-top altitude ranged from 11.7 to 14.3 km. The depth from 0° to −15°C followed a similar pattern, increasing as the WCTT decreased and estimated flash rate increased, then decreasing as the WCTT increased and estimated flash rate decreased (Fig. 10). At T − 10 the (maximum) depth from 0° to −15°C was 2.5 km, with 0°C altitudes ranging from 4.5 to 4.6 km and −15°C altitudes ranging from 6.8 to 7.0 km. The depth increased to 3.5 km at T − 4, 2 min before the coldest WCTT and 4 min before the peak estimated flash rate, with 0°C altitudes ranging from 3.7 to 4.4 km and −15°C altitudes ranging from 6.6 to 7.3 km.
Time series plot of the maximum cloud depth based on the LCL as cloud base and the uppermost altitude with 15-dBZ reflectivity as cloud top in black and the maximum depth of the layer from 0° to −15°C in gray for the WRF Model–simulated wind event at 0126 UTC 11 Apr 2018. The wind event occurred at T − 0, which is highlighted by the yellow box.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-22-0020.1
With mixing ratios, qT had the most contribution from rainwater (liquid form) and graupel (frozen form) (Fig. 11). The contribution from graupel exceeded that from rainwater. The time series plot of qT, or the condensate loading term from Eq. (1), revealed an increasing trend that corresponded to times of decreasing WCTT values and increasing estimated flash rate values, followed by a decreasing trend that corresponded to the times of increasing WCTT values and decreasing estimated flash rate values. The peak qT value occurred at T − 4, 2 min before the coldest WCTT and 4 min before the peak flash rate. Related to satellite observations, a GOES-16 derived field like cloud optical thickness (in addition to lightning and cloud-top height) could be used to identify oceanic wind events. The value remained roughly the same for about 5 min before beginning to decrease. Recall that the qT values in the time series plot are normalized by the number of WRF-based altitude levels from the surface to cloud top. Cross sections of graupel and rainwater, the largest frozen and liquid hydrometeor contributors to qT, through the wind event location near time T revealed that the largest graupel mixing ratios were associated with the strongest positive vertical velocity value at altitudes above about 5 km (Fig. 12). Near time T, the 0°C levels ranged from about 3.7–4.3 km. Another region of enhanced graupel mixing ratio values was located near the uppermost levels of a region of negative vertical velocity, about 3–5 km, which was the location of the downdraft nearing the surface with temperatures below freezing or slightly warmer than freezing. A region of negative vertical velocity near 10 km had graupel mixing ratio values similar to those associated with the region of positive vertical velocity. Figure 12 shows that at T − 2, the largest rainwater mixing ratios were associated with the downdraft nearing the surface, at about 1–2 km. That enhanced area of rainwater mixing ratio values continued to move closer to the surface as the region of negative vertical velocities moved closer the surface. The mixing ratio values support that melting and precipitation loading helped drive the wind event.
Time series plot of the maximum qT, qr, qc, qg, qs, and qi from the surface to cloud top, with the uppermost altitude with 15-dBZ reflectivity as cloud top, for the WRF Model–simulated wind event at 0126 UTC 11 Apr 2018. The wind event occurred at T − 0, which is highlighted by the yellow box. Note that values are normalized by the number of levels from the surface to cloud top.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-22-0020.1
Vertical cross section of (a)–(d) qg and (e)–(h) qr through the wind event location from T − 2 through T + 1 for the wind event at 0126 UTC 11 Apr 2018. The solid gray lines represent positive vertical velocities and the dashed gray lines represent negative vertical velocities. The vertical velocity values are contoured every 5 m s−1.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-22-0020.1
With the sum of the three frozen hydrometeor mixing ratios, graupel, snow, and ice, within the region from 0° to −15°C the value increased from T − 10 until reaching a maximum at T − 6 (Fig. 13). This maximum occurred 4 min before the coldest WCTT and 6 min before the peak flash rate. After T − 6, the sum of the three frozen hydrometeors began to decrease. Graupel contributed most of the sum, followed by smaller amounts from snow and ice, respectively. The increase in the frozen hydrometeor mixing ratios occurred as the estimated flash rate increased and WCTT decreased, likely correlating to the charging and separation needed for lightning to occur. Positive rainwater mixing ratio values above freezing, as shown in Fig. 12, likely meant supercooled water, which is important for lightning formation, was also present.
Time series plot of the maximum qF, qg, qs, and qi within the layer from 0° to −15°C for the WRF Model–simulated wind event at 0126 UTC 11 Apr 2018. The wind event occurred at T − 0, which is highlighted by the yellow box. Note that values are normalized by the number of levels from 0° to −15°C.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-22-0020.1
Figure 14 is a time series plot of the terms in the vertical equation of motion [Eq. (1)] each minute for the cell that was tracked from T − 10 through T + 25. Note that qT is not normalized like it was in Fig. 11. The top plot shows the acceleration value associated with each term based on the levels from the surface to cloud top, while the bottom plot shows the acceleration value associated with each term based on the levels from the surface to 5 km. From the surface to cloud top, as the cell was beginning to be tracked, condensate loading contributed the most negative buoyancy. Thermal buoyancy initially contributed to positive buoyancy, but began to contribute negative buoyancy just before the time of the wind event and as the negative buoyancy from condensate loading was lessening. As the negative buoyancy from thermal buoyancy continued to increase, the contribution from the condensate loading decreased. Although the vertical gradient of pressure perturbation term initially contributed to the negative buoyancy and the perturbation pressure buoyancy term later contributed to the negative buoyancy, both contributions were smaller than those from the thermal buoyancy and condensate loading terms. From the surface to 5 km, each of the four terms continuously contributed negative buoyancy. Thermal buoyancy and condensate loading contributed the most negative buoyancy. Thermal buoyancy initially provided an increasing amount of negative buoyancy and condensate loading initially provided a decreasing amount of negative buoyancy, before negative buoyancy contributions from both terms became more level and dominated by thermal buoyancy.
Time series plot of the acceleration values of the terms in the vertical equation of motion each minute (top) from the surface to cloud top and (bottom) from the surface to 5 km for the WRF Model–simulated wind event at 0126 UTC 11 Apr 2018. The wind event occurred at T − 0, which is highlighted by the yellow box.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-22-0020.1
2) 1643 UTC 3 June 2018
Wind event 27 occurred at 1643 UTC 3 June 2018 over the Atlantic Ocean with a vertical velocity of −16.4 m s−1 (Fig. 1 and Table 2). This was the 17th strongest wind event. The cell of interest was tracked from T − 16 through T + 30.
A WRF Model–based sounding from 1530 UTC 3 June 2018, 1 h 13 min before the time of the wind event, is shown in Fig. 15. Specific values are also provided in Tables 3 and 4. Similar to the previous event and the composite soundings, this event exhibited characteristics often found with overland wet downbursts, including moisture near the surface and a somewhat drier region aloft, along with near moist adiabatic lapse rates.
WRF-based sounding at 1530 UTC 3 Jun 2018. This sounding is associated with event 27, which occurred at 1643 UTC 3 Jun 2018.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-22-0020.1
The cell of interest was first tracked at T − 16 when the WCTT was 238.9 K (Fig. 16). One minute later, at T − 15, the WCTT was 235.2 K (Fig. 17). The WCTT continued to decrease as the cell moved toward the wind event location and reached its coldest value of 204.9 K at T − 6 (Fig. 16, Table 2). The WCTT then began to increase, warming to 208.8 K at time T and 211.0 K at T + 5. As the WCTT decreased, the WRF-estimated flash rate increased (Fig. 16). The estimated flash rate increased from 1.4 flashes per minute at T − 16 to the peak value of 10.9 flashes per minute at T − 1, 5 min after the coldest WCTT. After the wind event at time T, the WCTT continued to warm and the estimated flash rate continued to decrease.
Time series plot of WCTT in black and WRF-estimated flash rate in gray for the WRF Model–simulated wind event at 1643 UTC 3 Jun 2018. The wind event occurred at T − 0, which is highlighted by the yellow box.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-22-0020.1
WCTT at (a) T − 15, (b) T − 6, (c) T, and (d) T + 5 for the WRF Model–simulated wind event that occurred at 1643 UTC 3 Jun 2018. The location of the wind event is marked by the large purple asterisk in the center of each image. The (smoothed) cell of interest is enclosed by the gray outline. The coldest WCTT of 204.9 K occurred at T − 6.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-22-0020.1
Cloud (maximum) depth, based on the LCL as cloud base and the uppermost altitude with 15-dBZ reflectivity as cloud top, started at 9.5 km at T − 16 then increased to 13.8 km at T − 6 (Fig. 18). The increase in cloud depth corresponded to the decrease in WCTT, which reached the coldest value at the same time the cloud was the deepest, T − 6, along with the increase in estimated flash rate. At T − 16, the cloud-base altitude ranged from 0.139 to 0.236 km and the cloud-top altitude ranged from 9.6 to 9.7 km. At T − 6, the cloud-base altitude ranged from 0.127 to 0.319 km and the cloud-top altitude ranged from 9.7 to 14.0 km. After reaching its tallest value at T − 6, cloud depth then decreased as the WCTT increased and the estimated flash rate decreased. The depth from 0° to −15°C increased slightly from 3.2 km at T − 16 to 3.4 km at T − 14 as the estimated flash rate was increasing and the WCTT was decreasing, before it began to decrease (Fig. 18). At T − 16, the 0°C altitude ranged from 4.1 to 4.8 km and the −15°C altitude ranged from 7.2 to 7.7 km. At T − 14, the 0°C altitude ranged from 4.1 to 4.6 km and the −15°C altitude ranged from 7.1 to 7.5 km. At time T, those altitudes ranged from 4.1 to 4.7 km and from 6.9 to 7.2 km, respectively.
Time series plot of the maximum cloud depth based on the LCL as cloud base and the uppermost altitude with 15-dBZ reflectivity as cloud top in black and the maximum depth of the layer from 0° to −15°C in gray for the WRF Model–simulated wind event at 1643 UTC 3 Jun 2018. The wind event occurred at T − 0, which is highlighted by the yellow box.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-22-0020.1
The qT had the most contribution from rainwater (liquid form) and graupel (frozen form) (Fig. 19). The time series plot revealed that qT increased from T − 16 until reaching its highest value at T − 11, 5 min before the time of the coldest WCTT value. The WCTT experienced a decreasing trend and the estimated flash rate experienced an increasing trend from T − 16 through T − 11. The qT then began to decrease, with a slight increase a few minutes before time T. Again, recall that the qT values in the time series plot are normalized by the number of WRF-based altitude levels from the surface to cloud top. Cross sections of graupel and rainwater, the largest frozen and liquid hydrometeor contributors to qT, through the wind event location near time T are shown in Fig. 20. At time T, a region of enhanced graupel mixing ratios existed at the uppermost portion and above a region of low level negative vertical velocities, from about 3.5 to 9 km. At time T, the 0°C altitudes ranged from 4.1 to 4.7 km. With the rainwater mixing ratio, an enhanced region existed with the negative vertical velocity region near the surface. The mixing ratio values support that melting and precipitation loading helped drive the wind event.
Time series plot of the maximum qT, qr, qc, qg, qs, and qi from the surface to cloud top, with the uppermost altitude with 15-dBZ reflectivity as cloud top, for the WRF Model–simulated wind event at 1643 UTC 3 Jun 2018. The wind event occurred at T − 0, which is highlighted by the yellow box. Note that values are normalized by the number of levels from the surface to cloud top.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-22-0020.1
Vertical cross section of (a)–(d) qg and (e)–(h) qr through the wind event location from T − 2 through T + 1 for the wind event at 1643 UTC 3 Jun 2018. The solid gray lines represent positive vertical velocities and the dashed gray lines represent negative vertical velocities. The vertical velocity values are contoured every 5 m s−1.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-22-0020.1
With the sum of the three frozen hydrometeor mixing ratios within the region from 0° to −15°C, the value slightly decreased from T − 16 to T − 14 before increasing until T − 10, 4 min before the coldest WCTT (Fig. 21); the value then decreased. The graupel mixing ratio slightly decreased then increased to its maximum value at T − 10 before decreasing again. Lower contributions came from snow and ice. The snow mixing ratio peaked at T + 17 and T + 18 and the ice mixing ratio peaked at T − 3. Increasing frozen hydrometeor mixing ratio values as the WCTT decreased likely aided the charging and separation needed for lightning.
Time series plot of the maximum qF, qg, qs, and qi within the layer from 0° to −15°C for the WRF Model–simulated wind event at 1643 UTC 3 Jun 2018. The wind event occurred at T − 0, which is highlighted by the yellow box. Note that values are normalized by the number of levels from 0° to −15°C.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-22-0020.1
Figure 22 is a time series plot of the terms in the vertical equation of motion [Eq. (1)] each minute for the cell that was tracked from T − 16 through T + 30. Note that with the condensate loading term, qT is not normalized like it was in Fig. 19. The top plot shows the acceleration value associated with each term based on the levels from the surface to cloud top, while the bottom plot shows the acceleration value associated with each term based on the levels from the surface to 5 km. As the cell was beginning to be tracked, from both the surface to cloud top and from the surface to 5 km, condensate loading contributed the most negative buoyancy. As the cell progressed, less negative buoyancy was provided by condensate loading. Initially, the thermal buoyancy term provided negative buoyancy near the surface and positive buoyancy at higher altitudes. Negative buoyancy was continuously provided by thermal cooling from the surface to 5 km. When the upper altitudes were included, the contribution from positive buoyancy decreased and eventually gave way to slightly negative buoyancy after the time of the wind event. The contribution from the perturbation pressure buoyancy term remained fairly consistent and positive with time from the surface to cloud top, and changed from slightly negative before the wind event to slightly positive after the wind event from the surface to 5 km. The contribution from the vertical gradient of pressure perturbation term remained negative with time from the surface to 5 km, and changed from slightly positive to negative from the surface to cloud top.
Time series plot of the acceleration values of the terms in the vertical equation of motion each minute (top) from the surface to cloud top and (bottom) from the surface to 5 km for the WRF Model–simulated wind event at 1643 UTC 3 Jun 2018. The wind event occurred at T − 0, which is highlighted by the yellow box.
Citation: Weather and Forecasting 38, 11; 10.1175/WAF-D-22-0020.1
6. Conclusions
For this study, it was hypothesized that prior to a wind event, there should be unique signatures in WCTT trends and lightning data that can be exploited to increase the situational awareness of pending oceanic wind events. As stated in T21, based on in-cloud processes and prior research, the presumption was that the coldest WCTT and peak lightning flash rate should occur at about the same time and prior to the wind event, which is reflective of a cumulonimbus cloud growing to a maximum altitude with an overshooting top that manifested itself as a WCTT signature. Also, as a main cumulonimbus cloud updraft ascends, an increase in lightning occurs due to a lofting of frozen hydrometeors and subsequent in-cloud charge production.
The overall findings from the combined 48 WRF Model–simulated ocean-based wind events supported the hypothesis that unique signatures of ocean-based wind events should exist in WCTT and lightning trends. Analyses of both the observations and simulations supported the main hypothesis that the coldest cloud top and peak lightning should occur near the same time and prior to the wind event, which therefore defined the unique signatures associated with over ocean wind events. However, like with the 13 observed events, it is clear from the WRF Model data that some storms produce stronger updrafts (i.e., colder WCTT values), occur in environments with larger instability, reach greater altitudes, and produce more lightning than others.
The analyses of the 48 WRF Model–simulated ocean-based events also addressed science questions, including the following: Do the environmental characteristics of ocean-based wind events closely resemble those associated with land-based wet downbursts, or does a dry air layer influence their occurrence? The environmental characteristics of the 48 ocean-based wind events did resemble those associated with land-based wet downbursts (Fig. 2 and Tables 3 and 4). While overland dry downbursts are typically associated with drier air near the surface and more moisture aloft, lapse rates near dry adiabatic, and high cloud bases, overland wet microbursts are typically associated with moist air near the surface and drier regions aloft, lapse rates near moist adiabatic, and low cloud bases, which were also characteristics often associated with the WRF Model–simulated ocean-based wind events.
Another science question addressed with the WRF Model–simulated wind events was as follows: What are the main driving in-cloud mechanisms of ocean-based wind events? Like with an overland wet downburst, the ocean-based wind events simulated by the WRF Model were driven by melting, precipitation loading, and entrainment of dry-mid level air. From the box-and-whisker plots, qT peaked prior to the wind event, then experienced a general decreasing trend beginning at about T − 5. With the time series plots of the individual terms in the vertical equation of motion for the two individual wind events, the thermal buoyancy and condensate loading terms were shown to provide negative acceleration, which is supported by the high precipitable water values associated with the wind events. As the frozen hydrometeors (graupel, snow, and ice) melted, they too contributed to the negative buoyancy. Low cloud bases and moisture near the surface limited evaporation’s contribution to negative buoyancy.
The analyses also addressed the science question: If the coldest WCTT and peak flash rate do not occur at the exact same minute, does one extreme consistently occur first? Supporting the main hypothesis, the peak estimated flash rate with the WRF Model–simulated events occurred within 5 min of the coldest WCTT with 36 of the 48 events. Based on all 48 events, 42 had the coldest WCTT occur before or at the same time as the peak estimated flash rate. The 13 observed ocean-based wind events from T21 had similar results, where, with 11 of the 12 events that had lightning, the peak flash rate occurred within 5 min of the coldest CTTB. With 8 of the 12 events, the coldest CTTB occurred before or at the same time as the peak flash rate.
The final science question addressed in this study was the following: Are trends in the WCTT and lightning more useful than their specific magnitudes? It is noted that there was a range in the overall coldest WCTT and estimated peak flash values with the 48 WRF simulated events. Recall that with both the model simulated and observed ocean-based wind events, most events exhibited the trend of a decreasing WCTT/CTTB and corresponding increasing flash rate leading to the time of the wind event, followed by an increasing WCTT/CTTB and corresponding decreasing estimated flash rate after the event. While those trends and correlations would have identified most of the events, if instead, specific magnitudes of a cold WCTT/CTTB or large peak flash rate alone were used to determine whether or not a wind event would occur, many events might be missed.
Findings from this study, and the similarities to those discussed in T21, can further help operational weather forecasters identify storms capable of producing significant ocean-based wind events. With both the observed and model simulated ocean-based wind events, the recurring trend of the corresponding decrease in CTTB/WCTT and increase in flash rate leading up to the event, followed by the corresponding increase in CTTB/WCTT and decrease in flash rate was seen with most events, and together the trends comprised the event signal. With the WRF Model–simulated events, the additional in-cloud processes and environmental characteristics associated with the storm provided a more robust understanding of the events. Using the CTTB/WCTT and lightning trends, in conjunction with variables such as DCAPE, CAPE, LCL or cloud-base altitude, cloud depth, θe, θυ, mixing ratios, lapse rates, dewpoint depressions, and LFC and EL altitudes, future work could produce an algorithm or thresholds that once an automated tracking procedure for cumulus and cumulonimbus clouds was implemented, operational weather forecasters could use to help identify regions of possible ocean-based wind events, thereby providing advanced warning for maritime activities to avoid the dangerous regions.
Acknowledgments.
This research was funded by Marshall Space Flight Center sponsored Award NNM11AA01A. Thanks to Dr. Kevin Knupp, Dr. Michael Folmer, and Dr. Larry Carey for their suggestions that improved the quality of this research. The authors thank two anonymous reviewers for their comments.
Data availability statement.
The WRF Model datasets were simulated using version 3.8.1 of the WRF-ARW Model (https://www2.mmm.ucar.edu/wrf/users/; NCAR 2017). The GFS analysis data used as input were downloaded from the NCEI website.
REFERENCES
Atkins, N. T., and R. M. Wakimoto, 1991: Wet microburst activity over the southeastern United States: Implications for forecasting. Wea. Forecasting, 6, 470–482, https://doi.org/10.1175/1520-0434(1991)006<0470:WMAOTS>2.0.CO;2.
Bao, J.-W., S. A. Michelson, and E. D. Grell, 2019: Microphysical process comparison of three microphysics parameterization schemes in the WRF Model for an idealized squall-line case study. Mon. Wea. Rev., 147, 3093–3120, https://doi.org/10.1175/MWR-D-18-0249.1.
Barthe, C., W. Deierling, and M. C. Barth, 2010: Estimation of total lightning from various storm parameters: A cloud-resolving model study. J. Geophys. Res., 115, D24202, https://doi.org/10.1029/2010JD014405.
Blumberg, W. G., K. T. Halbert, T. A. Supinie, P. T. Marsh, R. L. Thompson, and J. A. Hart, 2017: SHARPpy: An open-source sounding analysis toolkit for the atmospheric sciences. Bull. Amer. Meteor. Soc., 98, 1625–1636, https://doi.org/10.1175/BAMS-D-15-00309.1.
Bolgiani, P., S. Fernández-González, F. Valero, A. Merino, E. García-Ortega, J. L. Sánchez, and M. L. Martín, 2020: Simulation of atmospheric microbursts using a numerical mesoscale model at high spatiotemporal resolution. J. Geophys. Res. Atmos., 125, e2019JD031791, https://doi.org/10.1029/2019JD031791.
Bolton, D., 1980: The computation of equivalent potential temperature. Mon. Wea. Rev., 108, 1046–1053, https://doi.org/10.1175/1520-0493(1980)108<1046:TCOEPT>2.0.CO;2.
Caplan, S. J., A. J. Bedard Jr., and M. T. Decker, 1990: The 700-500 mb lapse rate as an index of microburst probability: An application for thermodynamic profilers. J. Appl. Meteor., 29, 680–687, https://doi.org/10.1175/1520-0450(1990)029<0680:TMLRAA>2.0.CO;2.
Carey, L. D., and S. A. Rutledge, 1996: A multiparameter radar case study of the microphysical and kinematic evolution of a lightning producing storm. Meteor. Atmos. Phys., 59, 33–64, https://doi.org/10.1007/BF01032000.
Chae, J. H., and S. C. Sherwood, 2010: Insights into cloud-top height and dynamics from the seasonal cycle of cloud-top heights observed by MISR in the west Pacific region. J. Atmos. Sci., 67, 248–261, https://doi.org/10.1175/2009JAS3099.1.
Chen, F., and J. Dudhia, 2001a: Coupling an advanced land surface-hydrology model with the Penn State–NCAR MM5 modeling system. Part I: Model implementation and sensitivity. Mon. Wea. Rev., 129, 569–585, https://doi.org/10.1175/1520-0493(2001)129<0569:CAALSH>2.0.CO;2.
Chen, F., and J. Dudhia, 2001b: Coupling an advanced land surface-hydrology model with the Penn State–NCAR MM5 modeling system. Part II: Preliminary model validation. Mon. Wea. Rev., 129, 587–604, https://doi.org/10.1175/1520-0493(2001)129<0587:CAALSH>2.0.CO;2.
Chen, F., Z. Janjić, and K. Mitchell, 1997: Impact of atmospheric surface-layer parameterization in the new land-surface scheme of the NCEP mesoscale Eta model. Bound.-Layer Meteor., 85, 391–421, https://doi.org/10.1023/A:1000531001463.
Cotton, W. R., G. H. Bryan, and S. C. van den Heever, 2011: Storm and Cloud Dynamics. 2nd ed. Academic Press, 809 pp.
Dudhia, J., 1989: Numerical study of convection observed during the Winter Monsoon Experiment using a mesoscale two-dimensional model. J. Atmos. Sci., 46, 3077–3107, https://doi.org/10.1175/1520-0469(1989)046<3077:NSOCOD>2.0.CO;2.
Ek, M. B., K. E. Mitchell, Y. Lin, E. Rogers, P. Grunmann, V. Koren, G. Gayno, and J. D. Tarpley, 2003: Implementation of Noah land surface model advances in the National Centers for Environmental Prediction operational mesoscale Eta model. J. Geophys. Res., 108, 8851, https://doi.org/10.1029/2002JD003296.
Ellrod, G. P., 1989: Environmental conditions associated with the Dallas microburst storm determined from satellite soundings. Wea. Forecasting, 4, 469–484, https://doi.org/10.1175/1520-0434(1989)004<0469:ECAWTD>2.0.CO;2.
Ellrod, G. P., J. P. Nelson III, M. R. Witiw, L. Bottos, and W. P. Roeder, 2000: Experimental GOES sounder products for the assessment of downburst potential. Wea. Forecasting, 15, 527–542, https://doi.org/10.1175/1520-0434(2000)015<0527:EGSPFT>2.0.CO;2.
Fujita, T. T., 1985: The downburst: Microburst and macroburst. SMRP Research Paper 210, 122 pp., http://hdl.handle.net/10605/262010.
Fujita, T. T., and H. R. Byers, 1977: Spearhead echo and downburst in the crash of an airliner. Mon. Wea. Rev., 105, 129–146, https://doi.org/10.1175/1520-0493(1977)105<0129:SEADIT>2.0.CO;2.
Fujita, T. T., and R. M. Wakimoto, 1981: Five scales of airflow associated with a series of downbursts on 16 July 1980. Mon. Wea. Rev., 109, 1438–1456, https://doi.org/10.1175/1520-0493(1981)109<1438:FSOAAW>2.0.CO;2.
Fujita, T. T., and R. M. Wakimoto, 1983: Microbursts in JAWS depicted by Doppler radars, PAM, and aerial photographs. Preprints, 21st Conf. on Radar Meteorology, Edmonton, AB, Canada, Amer. Meteor. Soc., 638–645.
Geerts, B., 2001: Estimating downburst-related maximum surface wind speeds by means of proximity soundings in New South Wales, Australia. Wea. Forecasting, 16, 261–269, https://doi.org/10.1175/1520-0434(2001)016<0261:EDRMSW>2.0.CO;2.
Gilmore, M. S., and L. J. Wicker, 1998: The influence of midtropospheric dryness on supercell morphology and evolution. Mon. Wea. Rev., 126, 943–958, https://doi.org/10.1175/1520-0493(1998)126<0943:TIOMDO>2.0.CO;2.
Goodman, S. J., D. E. Buechler, P. D. Wright, and W. D. Rust, 1988: Lightning and precipitation history of a microburst-producing storm. Geophys. Res. Lett., 15, 1185–1188, https://doi.org/10.1029/GL015i011p01185.
Grundstein, A., M. Shepherd, P. Miller, and S. E. Sarnat, 2017: The role of mesoscale-convective processes in explaining the 21 November 2016 epidemic thunderstorm asthma event in Melbourne, Australia. J. Appl. Meteor. Climatol., 56, 1337–1343, https://doi.org/10.1175/JAMC-D-17-0027.1.
Hjelmfelt, M. R., H. D. Orville, R. D. Roberts, J. P. Chen, and F. J. Kopp, 1989: Observational and numerical study of a microburst line-producing storm. J. Atmos. Sci., 46, 2731–2744, https://doi.org/10.1175/1520-0469(1989)046<2731:OANSOA>2.0.CO;2.
Janjić, Z. I., 1994: The step-mountain Eta coordinate model: Further developments of the convection, viscous sublayer, and turbulence closure scheme. Mon. Wea. Rev., 122, 927–945, https://doi.org/10.1175/1520-0493(1994)122<0927:TSMECM>2.0.CO;2.
Kain, J. S., 2004: The Kain–Fritsch convective parameterization: An update. J. Appl. Meteor., 43, 170–181, https://doi.org/10.1175/1520-0450(2004)043<0170:TKCPAU>2.0.CO;2.
Knupp, K. R., 1989: Numerical simulation of low-level downdraft initiation within precipitating cumulonimbi: Some preliminary results. Mon. Wea. Rev., 117, 1517–1529, https://doi.org/10.1175/1520-0493(1989)117<1517:NSOLLD>2.0.CO;2.
Knupp, K. R., 1996: Structure and evolution of a long-lived, microburst-producing storm. Mon. Wea. Rev., 124, 2785–2806, https://doi.org/10.1175/1520-0493(1996)124<2785:SAEOAL>2.0.CO;2.
Knupp, K. R., and W. R. Cotton, 1985: Convective cloud downdraft structure: An interpretive survey. Rev. Geophys., 23, 183–215, https://doi.org/10.1029/RG023i002p00183.
Kuchera, E. L., and M. D. Parker, 2006: Severe convective wind environments. Wea. Forecasting, 21, 595–612, https://doi.org/10.1175/WAF931.1.
Lim, K.-S. S., and S.-Y. Hong, 2010: Development of an effective double-moment cloud microphysics scheme with prognostic cloud condensation nuclei (CCN) for weather and climate models. Mon. Wea. Rev., 138, 1587–1612, https://doi.org/10.1175/2009MWR2968.1.
Lucas, C., E. J. Zipser, and M. A. Lemone, 1994: Vertical velocity in oceanic convection off tropical Australia. J. Atmos. Sci., 51, 3183–3193, https://doi.org/10.1175/1520-0469(1994)051<3183:VVIOCO>2.0.CO;2.
Markowski, P., and Y. Richardson, 2010: Mesoscale Meteorology in Midlatitudes. John Wiley and Sons, 430 pp.
McCann, D. W., 1994: WINDEX—A new index for forecasting microburst potential. Wea. Forecasting, 9, 532–541, https://doi.org/10.1175/1520-0434(1994)009<0532:WNIFFM>2.0.CO;2.
McCaul, E. W., Jr., S. J. Goodman, K. M. LaCasse, and D. J. Cecil, 2009: Forecasting lightning threat using cloud-resolving model simulations. Wea. Forecasting, 24, 709–729, https://doi.org/10.1175/2008WAF2222152.1.
McCaul, E. W., Jr., G. Priftis, J. L. Case, T. Chronis, P. N. Gatlin, S. J. Goodman, and F. Kong, 2020: Sensitivities in the WRF lightning forecasting algorithm to parameterized microphysics and boundary layer schemes. Wea. Forecasting, 35, 1545–1560, https://doi.org/10.1175/WAF-D-19-0101.1.
Mlawer, E. J., S. J. Taubman, P. D. Brown, M. J. Iacono, and S. A. Clough, 1997: Radiative transfer for inhomogeneous atmospheres: RRTM, a validated correlated-k model for the longwave. J. Geophys. Res., 102, 16 663–16 682, https://doi.org/10.1029/97JD00237.
Monte, S., C. Vigorito, M. Bertaina, S. Ferrarese, K. Shinozaki, and S. Briz, 2019: WRF (Weather Research and Forecasting) model and radiative methods for cloud top height retrieval along the EUSO-SPB1 trajectory. Proc. 36th Int. Cosmic Ray Conf., Madison, WI, https://arxiv.org/pdf/1909.09782.pdf.
NCAR, 2017: ARW version 3 modeling system user’s guide. MMM/NCAR, 434 pp., https://www2.mmm.ucar.edu/wrf/users/docs/user_guide_V3/user_guide_V3.8/ARWUsersGuideV3.8.pdf.
NOAA SPC, 2023: SPC experimental sounding analysis system. NOAA, accessed 1 February 2022, https://www.spc.noaa.gov/exper/soundings/help/.
NWS DCAPE, 2023: Overview of downward convective available potential energy (DCAPE). 10 pp., https://www.weather.gov/media/lmk/soo/DCAPE_Web.pdf.
Proctor, F. H., 1988: Numerical simulations of an isolated microburst. Part 1: Dynamics and structure. J. Atmos. Sci., 45, 3137–3160, https://doi.org/10.1175/1520-0469(1988)045<3137:NSOAIM>2.0.CO;2.
Proctor, F. H., 1989: Numerical simulations of an isolated microburst. Part II: Sensitivity experiments. J. Atmos. Sci., 46, 2143–2165, https://doi.org/10.1175/1520-0469(1989)046<2143:NSOAIM>2.0.CO;2.
Pryor, K. L., 2015: Progress and developments of downburst prediction applications of GOES. Wea. Forecasting, 30, 1182–1200, https://doi.org/10.1175/WAF-D-14-00106.1.
Pryor, K. L., and G. P. Ellrod, 2004a: Recent improvements to the GOES microburst products. Wea. Forecasting, 19, 582–594, https://doi.org/10.1175/1520-0434(2004)019<0582:RITTGM>2.0.CO;2.
Pryor, K. L., and G. P. Ellrod, 2004b: WMSI—A new index for forecasting wet microburst severity. Electron. J. Oper. Meteor., 5, http://nwafiles.nwas.org/ej/pdf/2004-EJ3.pdf.
Púčik, T., P. Groenemeijer, D. Rýva, and M. Kolář, 2015: Proximity soundings of severe and nonsevere thunderstorm in central Europe. Mon. Wea. Rev., 143, 4805–4821, https://doi.org/10.1175/MWR-D-15-0104.1.
Romps, D. M., 2017: Exact expression for the lifting condensation level. J. Atmos. Sci., 74, 3891–3900, https://doi.org/10.1175/JAS-D-17-0102.1.
Skamarock, W. C., and Coauthors, 2008: A description of the Advanced Research WRF version 3. NCAR Tech. Note NCAR/TN-475+STR, 113 pp., https://doi.org/10.5065/D68S4MVH.
Srivastava, R. C., 1985: A simple model of evaporatively driven downdraft: Application to microburst downdraft. J. Atmos. Sci., 42, 1004–1023, https://doi.org/10.1175/1520-0469(1985)042<1004:ASMOED>2.0.CO;2.
Srivastava, R. C., 1987: A model of intense downdrafts driven by the melting and evaporation of precipitation. J. Atmos. Sci., 44, 1752–1774, https://doi.org/10.1175/1520-0469(1987)044<1752:AMOIDD>2.0.CO;2.
Straka, J. M., and J. R. Anderson, 1993: Numerical simulations of microburst-producing storms: Some results from storms observed during COHMEX. J. Atmos. Sci., 50, 1329–1348, https://doi.org/10.1175/1520-0469(1993)050<1329:NSOMPS>2.0.CO;2.
Thompson, K. B., M. G. Bateman, and J. R. Mecikalski, 2021: Signatures of oceanic wind events in geostationary cloud top temperature and lightning data. Wea. Forecasting, 36, 407–423, https://doi.org/10.1175/WAF-D-19-0214.1.
Tuttle, J. D., V. N. Bringi, H. D. Orville, and F. J. Kopp, 1989: Multiparameter radar study of a microburst: Comparison with model results. J. Atmos. Sci., 46, 601–620, https://doi.org/10.1175/1520-0469(1989)046<0601:MRSOAM>2.0.CO;2.
Wakimoto, R. M., 1985: Forecasting dry microburst activity over the high plains. Mon. Wea. Rev., 113, 1131–1143, https://doi.org/10.1175/1520-0493(1985)113<1131:FDMAOT>2.0.CO;2.
Wakimoto, R. M., 2015: Mesoscale meteorology: Microbursts. Encyclopedia of Atmospheric Sciences, 2nd ed. G. R. North, J. A. Pyle, and F. Zhang, Eds., Academic Press, 355–360.
Wolfson, M. M., 1990: Understanding and predicting microburst. Proc. 16th Conf. on Severe Local Storms, Kananaskis Park, AB, Canada, Amer. Meteor. Soc., 340–351.