Instabilities, Dynamics, and Energetics accompanying Atmospheric Layering (IDEAL) Campaign: High-Resolution in situ Observations above the Nocturnal Boundary Layer

The Instabilities, Dynamics, and Energetics accompanying Atmospheric Layering (IDEAL) project was conceived to improve our understanding of the dynamics of sheet and layer (S&L) structures in the lower troposphere under strongly stable conditions. The approach employed a synergistic combination of targeted multi-point observations using small unmanned aircraft systems (sUAS) guiding direct numerical simulation (DNS) modeling to characterize the dynamics driving the S&L structures and associated flow features. The IDEAL research program consisted of two phases. The first was an observational 5 field campaign to systematically probe stable lower atmosphere conditions using multiple DataHawk-2 (DH2) sUAS. Coordinated, simultaneous multi-DH2 flights were guided by concurrent Integrated Sounding System (ISS) wind profiler radar and radiosonde soundings performed by NCAR Earth Observing Laboratory (EOL) participants. Additional sUAS flight guidance was obtained from real-time sUAS measurements. Following the field campaign, the second phase focused on high-resolution DNS modeling efforts guided by in-situ observations made during the first phase. This overview focuses on the details of the 10 observational phase that took place from 24 October to 15 November 2017 at Dugway Proving Ground (DPG), Utah. A total of 72 DH2 flights coordinated with 93 balloon-borne radiosondes were deployed in support of the IDEAL field campaign. Our discussion addresses the average atmospheric conditions, the observation strategy, and the objectives of the field campaign. Also presented are representative flight sorties and sUAS environmental and turbulence measurements. 1 15 https://doi.org/10.5194/amt-2021-173 Preprint. Discussion started: 29 June 2021 c © Author(s) 2021. CC BY 4.0 License.

Details have been characterized in terms of typical sheet thickness and stability, thickness of turbulent layers, Richardson Number, and turbulence Reynolds number through in-situ measurements from soundings, stationary observation towers and tethered lifting systems (TLS) (Balsley et al., , 2006Muschinski et al., 2001a), and more recently, using aircraft (Lawrence and Balsley, 2013;Muschinski and Wode, 1998). High-resolution multi-point measurements of temperature (Barat, 1982;30 Coulman, 1973;Frehlich et al., 2003;Hunt et al., 1985;Xing-Sheng et al., 1983) and VHF radar estimates (and comparison with theoretical models) of refractive index structure function (C 2 n ) (VanZandt et al., 1978;Woodman and Guillen, 1974) have established the intermittent nature of turbulence within deep layers. More recently, quantitative aircraft measurements of turbulence kinetic energy dissipation rate ( ) and the temperature structure function (C 2 T ) have characterized the small-scale turbulence features within shallow layers (Balsley et al., 2018;Eaton et al., 1998;Fernando et al., 2015; Muschinski et al., measurements over many hours. This also enables observations in marginal conditions (e.g., high winds) that would ground more expensive vehicles due to the risk of loss. Ten DH2 vehicles were brought to the 23 day IDEAL campaign.
-Ruggedness. The airframe is resilient foam, strengthened by a system of interior spars and flexures that absorb impacts, enabling the vehicle to "bounce" rather than break when landing on unprepared surfaces. It has a no-tail design, since 85 these extended members are easily broken, and unbreakable wing trailing edges and vertical fins. It also has a rear propeller with folding blades to prevent damage to the propulsion system during landing. In the IDEAL campaign, five DH2 aircraft were used extensively, of which two were retired due to accumulated wear. No aircraft were lost.
-Ease of operation. A custom autopilot provides automatic launch, landing, and vector field flight control (Lawrence et al., 2008), enabling a variety of measurement strategies to be set up with ease and flown under minimal operator su-90 pervision. Flight patterns can also be changed during flight to target specific volumes of interest, e.g., based on real-time measurements-an ability that was extensively used during IDEAL to identify and more thoroughly sample turbulence fields. A bungee cord is used for launch, guided by a simple two-rail launch ramp (see Figure 2).
-Gust-insensitive design. The unique aerodynamic design eliminates the roll moment due to sideslip, making the vehicle point into gusts rather than roll away from it, enabling well-behaved flight in high-wind and strong turbulence conditions.

95
Normally, flights are not performed when surface winds exceed 10 ms −1 , or predicted winds aloft exceed 15 ms −1 . The vector field guidance uses a wind-aware algorithm to stabilize flight even when wind speed exceeds airspeed. During IDEAL, synoptic winds aloft often exceeded 20 ms −1 , limiting the flight ceiling to about 3 km.
-Flexible sensor interfacing. The custom DataHawk autopilot provides multiple serial interfaces (7 UART, 3 I2C, 4 SPI), enabling a variety of sensors to be supported, and their data stored on-board (micro SD card), and telemetered to the 100 ground station for real time display. Tables 1 and 2 provide details of the sensors employed for IDEAL. Sensors can be installed at various locations in the body or the wings of the airframe without altering the flight dynamics.
-Efficiency. Flight durations exceed 80 min, making altitudes of 5 km above a ground launch accessible with a typical 2 ms −1 ascent/descent rate, and a lateral range (out and back) of 30 km at a nominal airspeed of 15 ms −1 . For IDEAL, the DH2 was configured to make the following in-situ observations. 1. Measurement location and time. A UBlox M8N single-frequency GPS receiver provides horizontal position data and time at 5 Hz cadence. Altitude measurement is refined in post-flight analysis, to obtain high-vertical resolution, by calibrating the higher rate of response (100 Hz) barometric pressure altitude against the low-rate (∼ 5 Hz) GPS altitude.
Similarly, sensor measurement times are recorded at high-resolution by calibrating 10 microsecond microprocessor timer ticks to GPS time of week (TOW) data in post-flight time-alignment procedures. 3. High-resolution temperature. A custom (coldwire) thermometer uses a five micrometer diameter platinum wire to detect fine-scale temperature variations in the flow. With a time constant of 0.5 millisecond and a sampling rate of 115 800 Hz, temperature variations at wavenumbers of ∼25 m −1 can be detected at the nominal 15 ms −1 airspeed. The temperature is calibrated against the collocated (but slow) SHT temperature (to kelvin) in post flight analysis and used with high-resolution altitude to obtain high-vertical resolution potential temperature θ. Spectral analysis is also used to fit inertial sub-range power spectral density models to provide estimates of the turbulent temperature structure parameter 4. High-resolution airspeed. A custom pitot-static tube and a TE MS4515 differential pressure sensor provide 800 Hz airspeed data that is calibrated to ms −1 . Wavenumber resolutions simlilar to temperature fluctuations are obtained in velocity variations also, and spectral estimation methods are used to derive turbulent kinetic energy dissipation rate .
Filtered airspeed data are also used to estimate winds (described below). In addition, a custom (hotwire) anemometer uses a second 5 micrometer diameter platinum wire to detect fine-scale velocity variations, and these are also used to 125 estimate , but at higher confidence level due to the absence of motor vibration artifacts that typically appear in the pitot velocity spectra at high frequencies.
5. Horizontal Wind. Vehicle GPS velocity is combined with pitot airspeed and vehicle attitude to produce estimates of the horizontal wind at 1 Hz cadence in post flight analysis.
6. Atmospheric stability. The Brunt-Vaisala (buoyancy) frequency is evaluated using vertical gradient of high-resolution 130 potential temperature θ.
7. Forcing conditions. Destabilizing horizontal wind shear is assessed relative to the background layer stability via the gradient Richardson number, derived from the horizontal mean wind gradient with altitude, and the local buoyancy frequency.

135
An Integrated Sounding System (Parsons et al., 1994) was deployed to monitor the large-scale wind and thermodynamic environment, in close proximity to the UAV measurements. The ISS consisted of a Vaisala MW41 radiosonde sounding system, a LAP3000 915 MHz radar wind profiler, and Lufft WS700/WS800 surface meteorological sensors on a mast at 2 and 10 m.
Ninety-three balloon-borne RS41-SGP radiosondes were launched between 3:00 AM and 7:00 AM LT at 30 to 60 min intervals every night, providing five to nine soundings each measurement day.

140
The ISS automatically ingests surface observations from a set of reference sensors (T/RH, and wind using Lufft WS700 and pressure using Vaisala PTB210) at 1.8 and 3 m. To achieve frequent soundings (< 60 min apart), communications were terminated well before balloon burst (at 12 km) to enable launch preparations for subsequent soundings. The balloon Helium volume was adjusted to achieve a median ascent rate of ∼ 3.5 ms −1 .
The LAP3000 915 MHz radar wind profiler was operated in a low-height range mode to provide data at 60 m intervals 145 between 200 m to 4.5 km AGL. Due to the dry conditions, winds were measured only up to 2 km on most days. The radar employed five beam directions and raw Doppler spectra were recorded every 30 s. Zonal and meridional wind components were calculated from spectral moments averaged over 30 min.
Time-altitude data from radiosondes and the wind profiler were relayed hourly to the UAS flight deployment team to aid in-flight planning. Examples of ISS observations on 1 November 2017 are shown in Figure 3. Wind profiler data were used to 150 monitor relevant events like precipitation (descending features in signal to noise ratio (SNR)), low-level jets (midnight at ∼2 km), KHI (overturning features in the plots -enlarged in the insets of SNR and vertical velocity, e.g. see Figure 3).

Campaign Meteorological Conditions
Weather Research and Forecasting (WRF) model to predict the weather conditions at the US Army Test and Evaluation Command (ATEC) Ranges (Knievel et al., 2017;Liu et al., 2008). The system is a product of collaboration between ATEC and NCAR.
The local surface conditions were obtained using a network of towers that includes 31 SAMS and 50 mini-SAMS. Each SAMS reports 5 min averaged wind speed and direction at 2 m and 10 m, temperature, and relative humidity (T/RH) at 2 m, 160 and precipitation. The mini-SAMS towers provide additional 10 m T/RH measurements with average values reported every minute. Doppler radar wind profilers provided real-time wind profiles from 120 m up to 5 km. The forecasts included expected synoptic-scale patterns, for example, expected times of frontal passage, development of surface inversions, and cloud cover.
During the campaign, DH2s were flown between 2 and 8 AM LT to sample the evolution of nocturnal atmospheric conditions.
Weather briefings were provided to the team each day at 0:30 AM, so that launch sites and deployment strategies could be 165 specified based on the most recent information. An example dataset showing 4DWX model predictions of surface temperature and horizontal winds at 0900 UT (2 AM LT) on 6 November 2017 is shown in Figure 4. Figure 5 shows the observed horizontal wind vector from the 449 MHz radar wind profiler on November 6 between 0200 Z and 0800 Z (7:00 PM to 1:00 AM LT).
Observed T/RH, and winds from all the soundings throughout the campaign are shown in Figure 6. Conditions were mostly dry with occasional evening precipitation. Surface winds during the first week of the campaign (24 October to 1 November 170 2017) were consistently strong from the South (see top right panel in Figure 6). Thereafter, surface winds were consistently from the North. The 4DWX model predicted Northerly surface winds in the valley for the last two weeks of the campaign, and the wind forecasts agreed closely with the DPG MET 449 MHz wind profilers. Predicted surface temperatures were between 0 and -5 • C for most nights.     A total of 14 flight sorties from FS1 and 13 sorties from FS2 were performed. Most of these involved three simultaneous DH2 flights. Each sortie had one vertical "sounding" aircraft A1. Trajectories of the other "lateral" aircraft varied significantly, 210 depending on the conditions relayed by recent radiosonde data and on the conditions observed by A1. Some of these lateral trajectories concentrated on a particular turbulent layer evolution, moving laterally while slowly ascending and descending through a narrow altitude range to observe spatial and temporal variability in the layer. Other sorties, e.g., as shown in Figure   10, sought to investigate temporal evolution of multiple layers with measurements that were displaced evenly in time.   of mountain waves. Temperature gradients as steep as 0.18 Km −1 or ∼ 18Γ (with tropospheric dry adiabatic lapse rate Γ ∼ 9.8 × 10 −3 Km −1 ) were typically observed across most sheets (see enlarged inset on Figure 12).

225
One particular case is presented through Figures 11 to 14 Figures 12 (from S1 at 03 LT) and 14 (from S2 at 05 LT) exhibit signs of diminishing turbulence likely leading to relaminarization enabling formation of a steep temperature gradient that is characteristic of a highly stable sheet: the extinction of turbulence is apparent in the abrupt reduction of just below 1400 m from the first ascent to the first decent in Figure 14.

235
The confined elevation of C 2 T immediately below 1400 m in Figure 14 further supports this conclusion.     GWs and dynamically stable mean shears, referred to as multi-scale dynamics (MSD), employed idealized high resolution DNS (Balsley et al., 2018;Fritts et al., 2009b;Fritts and Wang, 2013;Fritts et al., 2013). These DNS simulations revealed 255 a diversity of secondary instability dynamics exhibiting small-scale versions of larger-scale dynamics accompanying more idealized GW breaking and KHI shear instability events. Importantly, multiple examples of these induced dynamics closely resembled small-scale instabilities in high-resolution atmospheric imaging at higher altitudes (Miller et al., 2015;Fritts et al., 2017Fritts et al., , 2019Geach et al., 2020;Hecht et al., 2021), confirming the ability of these DNS to describe both idealized and MSD instability dynamics. While highly idealized, the MSD also predicted the emergence of these dynamics within induced S&L structures that emerged from the idealized initial conditions. These initial DNS results, and the emerging DH2 measurement capabilities, provided the motivations for the IDEAL program, and expanded such MSD studies are contributing to exploration of the implications of IDEAL measurements.
The DNS of MSD featured a single initial monochromatic GW having an amplitude of a = (dθ/dz) min /(dθ/dz) = 0.5 and an intrinsic frequency ω = N/10. A constant mean stability N, and Re = 50000 was assumed to enable instabilities and turbulence structures accompanying GW-FS dynamics that extend to very small-scales (Balsley et al., 2018;Fritts and Wang, 2013;Fritts et al., 2013). Example fields from the MSD are shown with (x, z) cross sections at the spanwise domain center at two times separated by 1.5T b , for buoyancy period T b , in Figure 17, left and right. The upper and lower panels show log 10 and N 2 , respectively. Red arrows identify KHI progressing from right to left along the most highly-stratified vortex sheet initiated by a GW propagating from right to left and elevating the vortex sheet accompanying its most upward displacement.

270
Counter-intuitively, the strongest KHI arise at the most stable N 2 because emerging large N 2 arise due to vertical convergence of the flow, and of the local shear, dU/dz, causing the local Richardson number, Ri = N 2 /(dU/dz) 2 , to decrease below 0.25 because N 2 increases linearly, but (dU/dz) 2 increases quadratically with vertical convergence, thus yielding the minimum Ri at the maximum N 2 . Importantly, the log 10 fields reveal that MSD KHI make comparable, or larger, contributions to total at these times.

275
An additional DNS designed to explore KHI MSD in environments allowing mis-aligned initial KHI along their axes in the absence of GW modulations of the initial shear layer was carried out. This DNS was carried out at Re = 5000, minimum Ri = 0.1 and employed an initial noise spectrum that led to several cases of "tube and knot" (T&K) dynamics that were examined in detail by Fritts et al. (2021a) and Fritts et al. (2021b). Figure 18 shows (x,y) and (x,z) planes of log 10 spanning 1T b . Two interesting sites within the DNS model domain were examined in detail: one at the spanwise location denoted "a" at 280 top and bottom in Figure 18 containing two vortex tubes linking to a common KH billow core, and a second at the spanwise location denoted "b" exhibiting a single vortex tube linking two adjacent billow cores.
Both of these T&K dynamics lead to significantly accelerated transitions to, and enhanced turbulence relative to regions exhibiting turbulence transitions in the absence of T&K dynamics. See, for example, the rapid expansion of intensifying log 10 in the central region, and to the upper left and lower right of the central billow core, around location "a" at successive times at 285 top in Figure 18. Corresponding vertical cross sections along "a" in the center row at bottom in Figure 18 reveal that the vortex tubes either side of the central KH billow drive its more rapid breakdown to turbulence relative to the other locations shown.
Related, but less aggressive, T&K dynamics also accompany the single vortex tube linking adjacent KH billow cores at left and center at location "b" at top in Figure 18. These T&K dynamics are likewise more aggressive than in their absence; see the advanced transitions to turbulence at this location (site "b" in the bottom row at bottom in Figure 18) relative to that in the top 290 row exhibiting no T&K dynamics.
The expected wide-spread presence of such dynamics, given the highly-modulated local sheared environments in which they arise, such as seen to arise in Figure 17 from initially two-dimensional initial conditions, suggest that KHI T&K dynamics in the atmosphere are more the rule than the exception, and were likely major contributors to the small-scale dynamics and S&L structures measured during the IDEAL field program.
295 Figure 17. log10 (x, z) (top frames) and corresponding N 2 (x, z) (bottom frames) evolution at y = 0. The two set of frames presented here are 1.5T b apart in time.
In this overview, we have presented the goals and the observational strategy of IDEAL program. The sensors and platforms used have been described in detail. An example has been provided using data from Sorties 1 and 2 on 6 November 2017. The example clearly shows the evolution of S&L structures in the lower atmosphere.
So far, only the data from the vertical sounding aircraft (A1) in each IDEAL sortie have been analyzed. Much remains to