Assessing Downburst Kinematics Using Video Footage Analysis

Abstract

Measurements of downburst outflows using standard meteorological instruments (e.g., anemometers) are rare due to their transient and localized nature. However, video recordings of such events are becoming more frequent. This short communication (Technical Note) study presents a new approach to estimating the kinematics of a downburst event using video footage recordings of the event. The main geometric dimensions of the event, such as downdraft diameter, cloud base height, outflow depth, and the radius of the outflow at a given moment in time, are estimated by sizing them against reference structures of known dimensions that are present in the video footage. From this analysis, and knowing the frame rate of the video recording, one can estimate the characteristic velocities in the downburst event, such as the mean downdraft velocity and the mean velocity of the radial outflow propagation. The proposed method is tested on an August 2015 downburst event that hit Tucson, Arizona, United States. The diameter of the downburst outflow increased with the time from approximately 1.10 km to 3.35 km. This range of values indicates that the event was a microburst. The mean descending velocity of downburst downdraft was 8.9 m s⁻¹ and the horizontal velocity of outflow propagation was 17.7 m s⁻¹. The latter velocity is similar to the measured wind gust at the nearby weather station and Doppler radar. The outflow depth is estimated at 160 m, and the cloud base height was approximately 1.24 km. Estimating the kinematics of downbursts using video footage, while subject to certain limitations, does yield a useful estimation of the main downburst kinematics that contribute to a better quantification of these localized windstorms.

1. Introduction

Downbursts are strong downdrafts that emerge from a thunderstorm and spread out radially after hitting the surface. The near-surface winds in the outflow can exceed 60 m s⁻¹, which is equivalent to wind gusts in an EF3-rated tornado [1]. However, such a downburst would be considered an extremely strong windstorm because most outflow winds range from approximately 10 to 40 m s⁻¹ [2]. The duration of the event can be anywhere between approximately 5 to 20 min [3]. The near-surface wind speed and direction during downburst wind events can change in a matter of minutes, even seconds, which, along with their localized nature, often makes them difficult to analyze and quantify using past and current research methods [4].

Downbursts were systematically analyzed by Prof. Theodore Fujita from the University of Chicago in Illinois, United States, in response to a crash landing of Eastern Airlines Flight 66 at John F. Kennedy Airport (New York, NY, USA) on 24 June 1975. A total of 112 people lost their lives in this tragedy (see [5] for Fujita’s contribution to downburst research). It was first noted from the reports at the airport that some aircraft near the event noticed little to no adverse weather conditions while others, including Flight 66, experienced hazardous winds. This observation gave the researchers a clue that the meteorological phenomenon associated with the hazard had to be highly transient and spatially localized. This event, among other reports and past research, led Prof. Fujita to define the cause of the crash as a downburst (a phenomenon earlier suggested by Horace Byers) since downbursts resemble a downdraft that spreads outward upon hitting the ground [5]. Downbursts are not only a danger to aircraft during take-off and landing, but they can also threaten human safety on the ground and are hazardous to building infrastructure, in particular low-rise buildings. For example, convective storms were the most devastating wind type in the provinces of Ontario and Quebec, Canada, from 2008 to 2021 [6]. These windstorms, which include downbursts, caused about 67% of the insured losses in these two provinces.

Downbursts are subdivided into different categories depending on factors such as precipitation present during the event, diameter of downdraft, and diameter of radially outflowing wind. There are both wet and dry downbursts; wet downbursts are accompanied by precipitation at the surface, and dry downbursts have little to no precipitation present. Microbursts are categorized as small-scale downbursts with outflow diameters below 4.0 km and generally, although not always, consist of more intense winds than macrobursts [7]. Macrobursts are downbursts with a diameter of outflow larger than 4 km [1].

While high-frequency anemometer measurements of downbursts provide the most precise information in terms of mean wind and turbulence properties of these transient events [8], they are point measurements or, at best, line measurements if there are several anemometers installed on a tower, which leads to the low probability of downburst passing over the instrument. Furthermore, very intense downbursts can damage anemometers. Also, anemometer measurements alone are not sufficient to provide information such as the size and intensity of downdraft, downburst location relative to the anemometer, or the depth of downburst outflow, to name a few. On the other hand, the spatiotemporal characteristics of downburst-producing thunderstorms, as well as upper regions of downburst outflow and downdraft, can be resolved using one or preferably multiple Doppler radars [9]. However, the near-surface flow field cannot be resolved using Doppler radars due to ground clutter, obstacles blocking radar beams (e.g., mountains and tall buildings), and Earth’s curvature effects. While Doppler lidars can measure velocities close to the surface due to their shorter wavelengths compared to Doppler radars, they are short range, and, furthermore, heavy precipitation attenuates shortwave lidar beams. Given the lack of reliable downburst measurements and even less reliable, rough estimates of downburst sizes and their governing velocities, there is a need to propose new, simplistic methods to determine the basic properties of downburst outflows.

This technical note aims to determine to which extent the main kinematics of a downburst—namely, the governing geometric and velocity parameters—can be estimated through a method of analyzing high-resolution video footage. To achieve this objective, our study is analyzing a downburst event that occurred on 8 August 2015 in Tucson, Arizona, United States. This methodology could simplify future downburst research using simple video camera recordings of downburst events. Section 2 describes data and methodology that were used to conduct this research. Section 3 provides the main findings in terms of the characteristic velocities and geometric scales of the 2015 Tucson downburst. Section 4 discusses the main uncertainties of the proposed methodology, while Section 5 summarizes the main results of this research.

2. Data and Methods

This research was conducted by analyzing video footage of a downburst event that occurred in Tucson, Arizona on 8 August 2015, at 15:24 Mountain Standard Time (MST; Figure 1). The event was filmed using a high-resolution Nikon D610 camera 35.9 mm × 24 mm (Full frame DSLR) with a recording frame rate of 60 Frames Per Second (FPS). A 20 mm Nikon lens was used for recording. The camera was mounted on a tripod at a height of approximately 1.5 m above the ground at a location indicated in Figure 1. The elevation angle of the camera was approximately 0°. An edited video of the original recording is available here: https://youtu.be/jIhesf1_WTo?si=yq7iSvP8RPGeH9qN (accessed on 29 September 2024) [10]. The photographer (Bryan Snider) kindly provided the authors with the unedited version of the video recording, which is 17 min long.

The downburst was also detected at the Davis-Monthan Air Force Base weather station (station ID: KDMA). The meteorological measurements include air temperature, pressure, and wind speed recordings. The downburst signature was also recorded by the nearby Doppler radar located some 40 km southeast of the anemometer location (Figure 1). This radar (KEMX; 31°53′37″ N, 110°37′49″ W) is a part of the NEXRAD (Next Generation Weather Radar) network of S-band (frequency 2700 to 3000 MHz) Doppler radars across the United States. The radar is situated at an altitude of approximately 1.62 km above mean sea level. The Doppler velocity range of this radar is about 230 km.

To determine the main kinematics of a downburst event, the radius of outflow ($r$), the diameter of downdraft ($d$), the height of outflow ($H$), and downdraft decent height ($h$) must all be estimated. In many cases, the bottom cloud base is considered as a proxy for $h$. The location at which the downburst downdraft first touched the surface is estimated by matching landmarks and scenery located in the video to locations on Google Earth Pro^® maps. Once the center of the event is estimated, certain landmarks surrounding it, such as the one shown in Figure 2, are scaled for height or length and stacked vertically or horizontally to estimate the height and horizontal dimensions of the downburst, as observed in the video footage. The utility pole was used due to its simple geometry and vertical orientation. The height of the unitily pole (31 m) was measured using the 3D Path feature in Google Earth Pro^®. This tool was designed to measure the height and width of three-dimensional (3D) buildings and the distance from points on the building to the ground.

After the distances are measured using structures of known height as proxies (Figure 2), a characteristic downburst velocity can be estimated as

$$v={\displaystyle \frac{∆s}{∆t}}$$

(1)

where the distance interval, $∆s$, is the spatial distance of interest (e.g., $d$, $h$, and $r$) and the time interval, $∆t$, is the time in the video footage that elapsed during the outflow passage over the distance interval $∆s$. All times are measured relative to $t$ = 0, which is the moment the downburst downdraft touched the ground (Figure 3), i.e., just before the outflow starts to spread out radially.

Afterward, velocities in the radially spreading outflow are calculated at different time intervals and subsequently averaged to determine the mean outflow propagation velocity ($u$). This velocity is also known as the gust front propagation velocity. Similarly, the mean velocity of downdraft descent ($v$), as it descends from the cloud base to the ground, is calculated by averaging velocities obtained over different time intervals. For the descending velocity ($v$), which was found to be less variable than the outflow velocity, approximately 50 s intervals at 60 FPS were used with the respective estimated distances. Similarly, for outflow propagation ($u$), approximately 40-s intervals at 60 FPS were used with respective estimated distances traveled over this time.

The estimates of $u$ and $H$ are subsequently used to calculate the air density in the downburst outflow (${\rho }_2$), by solving the gust front propagation velocity equation [9,11]:

$$u=k\sqrt{{\displaystyle \frac{gH\left({\rho }_2-{\rho }_1\right)}{{\rho }_1}}}$$

(2)

for ${\rho }_2$, namely:

$${\rho }_2={\rho }_1\left(1-{\displaystyle \frac{u^2}{k^{2}gH}}\right)$$

(3)

Here, $k=0.77$ is a constant related to the ratio of internal to gravitational forces, and $g$ is the acceleration due to gravity (9.81 m s⁻¹). In Equation (3), the ambient air density, ${\rho }_1$, is calculated using the equation of state for ideal gas:

$${\rho }_1={\displaystyle \frac{P_a}{R{T}_a}}$$

(4)

where the surface ambient air pressure ($P_a$) and temperature ($T_a$) were obtained from the Davis–Monthan Air Force Base weather station, and $R$ = 287 J kg⁻¹ K⁻¹ is the gas constant for dry air. The effect of relative humidity on air density is neglected.

3. Results

The diameter of the downdraft ($d$) from $t$ = −139 s to $t$ = 0 was estimated to be approximately 1.1 km. The diameter of the outflow ($2r$) from $t$ = 0 to $t$ = 187 s was estimated to reach a maximum value of 3.35 km (Figure 4). While this assessment confirms that the event is a downburst, it also classifies it as a microburst since the radial outflow spread is over a distance of less than 4 km.

The height from the cloud base to the ground ($h$) was estimated using the structure in Figure 4. Our analysis found that the cloud base was at approximately 1.24 km above ground level (a.g.l.). Similarly, the height of outflow ($H$) is estimated to be 160 m a.g.l. (Figure 4). From the literature, this value of $H$ is lower than expected for outflow height. The outflow height is more commonly measured to be approximately 400 m a.g.l. or more [11]. The small value of H (160 m a.g.l.) may, in part, be due to the uncertainty in the depth perception in the video recording as well as the inability to accurately observe the true height of moving wind through video footage. Another possible source of uncertainty is the large variability in the measurements of $H$ extracted at different time frames. The values varied between approximately 260 m to 110 m a.g.l. However, it should also be noted that microbursts, on average, are characterized by lower outflow depths than the larger-scale downbursts.

The velocity of decent ($v$) was estimated to be 8.9 m s⁻¹, and the velocity of outflow propagation ($u$) was estimated to be 17.7 m s⁻¹. This value of $v$ falls within the range of maximum downdraft speed of 6 to 22 m s⁻¹ associated with the downdraft diameters in the range between 1.5 and 3 km [3,9,11]. We further observe that the gust wind speeds observed at the nearby Davis–Monthan Air Force Base weather station (Figure 5) are within the range of uncertainties of the estimated gust front propagation speed. Moreover, the typical signature of a diverging downburst outflow is also observed on the Doppler radar imagery in Figure 6. Notice that the maximum radial velocity towards the radar reached 15 m s⁻¹, which is once again comparable to the estimated value from the video recording. Figure 7 shows that the near-surface environment around the time of the observed downburst was characterized by low relative humidity, which, in combination with moist mid-troposphere and dry high troposphere, is conducive to downbursts [12,13].

Using Equation (4) with the values of $P_a$ = 101,280 Pa, $T_a$ = 307.15 K, ${\rho }_1$ is calculated to be 1.15 kg m⁻³. Then, knowing that H = 160 m, u = 17.7 m s⁻¹, the calculated value of air density in the downburst outflow is ${\rho }_2$ = 1.53 kg m⁻³. This value is rather high for air density close to the surface, and the uncertainty is likely due to the issues associated with the estimates of $H$, as discussed above. If, for example, $H$ = 400 m., which is closer to the expected value of the outflow height, the air density would be approximately 1.30 kg m⁻³. While still high, this value is much closer to the expected range of values for air density in downburst outflows [11].

4. Uncertainties

Photographs and video recordings taken by cameras are two-dimensional projections of a 3D world. As a result, the imagery is influenced by a lack of depth perception. In our research, the distance between the downdraft touchdown and the camera position (Figure 1) was determined by carefully analyzing which objects in the video were visible before but not after the moment of downburst impact on the surface ($t$ = 0). The disappearance of objects in the outflow can be attributed partly to precipitation and reduced visibility within the downdraft. In other words, if a building was visible in the footage for $t$ < 0 but no longer visible at $t$ = 0, the distance between the camera and that building was used to estimate the distance between the camera and the near-side edge of the downdraft. Assuming the downdraft is circular and has a diameter equal to that measured from the video analysis (Figure 4), one can then estimate the location of the downburst’s centerline touchdown on the surface (Figure 1). However, not all downdrafts are perfectly circular, and image distortion often occurs near the edges of the frame. For instance, while the distance between two objects near the center of the image and those close to the edges may appear similar in the footage, these distances may differ in reality due to camera lens distortion. Most of the dimensions and information extracted from the recording come from the central part of the image, where distortion is minimized. The distortion effect likely had the greatest impact on the estimation of the downburst descent height.

The raw video was recorded at a resolution of 3840 × 2160 pixels. Although higher video quality could offer more detail on the fine-scale features of the outflow, this resolution is considered sufficient to capture the primary kinematic properties of the event. While further postprocessing, such as additionally adjusting contrast, colors, and lighting to account for various environmental factors, could likely improve the analysis, that was not done in this study. The objective here was to demonstrate that a raw video recording of standard resolution, without the use of advanced editing techniques, can still be used to obtain quantitative information about downbursts.

Another limitation of this research is that the proposed method has been validated using only a single downburst event. However, given the purpose of this article as a technical note, the primary aim is to introduce a new technique for downburst research and promptly share it with the broader scientific community. Future research articles will apply this methodology to a larger number of thunderstorm wind cases to further validate its effectiveness.

5. Conclusions

This technical note proposes a simple methodology to estimate the main kinematic properties of a downburst outflow. The analysis is based on the downburst that hit Tucson, Arizona, United States, in the afternoon of 8 August 2015. During this event, a weather station near Tucson recorded wind gusts of approximately 18 m s⁻¹. Characteristic velocities and dimensions of the downburst are assessed by comparing their size in video footage against nearby structures of the known size. Namely, a structure, such as a transmission pole, is used as a “measuring stick” to horizontal and vertical distances in the outflow. Using the estimates of the distances and knowing the frame rate of the video recording, we estimate characteristic velocities in the outflow, such as the mean velocity of descending downdraft and the mean velocity of the radial propagation of downburst outflow.

By comparing the values of diameter, height, and velocity estimated through the analysis of video footage to the expected values based on Doppler radar measurements from a weather radar near Tucson and those in literature, it can be concluded that video footage analysis can yield useful estimations of basic downburst kinematics. While the results are not as reliable as those obtained from the anemometer, Doppler radar, or lidar measurements, this method can be used to give a first-order estimate of the governing kinematic properties of downbursts. For example, such assessment is useful in wind engineering analysis of structural responses to downburst winds [14,15] and validation of numerical models [16].