2.1 Platforms for measuring the lower atmosphere

Aerial measurements were made by research aircraft, uncrewed (i.e., remotely piloted) aerial systems (UASs), and from balloon- or parachute-borne soundings. These were mostly distributed along Tradewind Alley. Figure 2 shows the realization of the EUREC⁴A strategy: the EUREC⁴A-Circle (teal) and box L (orange) stand out, indicative of the number of times HALO and the ATR flew these patterns. The very large number of dropsondes deployed by HALO (black dots) gives further emphasis to the EUREC⁴A-Circle. Excursions by HALO and flights by the P-3 extended the area of measurements upwind toward the NTAS. The TO intensively sampled clouds in the area of ATR operations in the western half of the EUREC⁴A-Circle. UASs provided extensive measurements of the lower atmosphere, mostly in the area between the EUREC⁴A-Circle and Barbados. Due to their limited range many (Skywalkers, CU-RAAVEN, and quadcopter) only appear as dots on Fig. 2.

Different clusters of radiosonde soundings (evident as short traces, or whiskers, of gray dots) can also be discerned in Fig. 2. Those soundings originating from the BCO (342) and from the R/V Meteor (362) were launched from relatively fixed positions, with the R/V Meteor operating between 12.5 and 14.5^∘ N along the 57.25^∘ W meridian. East of the EUREC⁴A-Circle, sondes were launched by the R/V Ronald H. Brown (Ron Brown), which mostly measured air masses in coordination with the P-3 measurements between the NTAS and the EUREC⁴A-Circle. The R/V Maria Sibylla Merian (MS-Merian) and R/V L'Atalante (Atalante) combined to launch 424 sondes in total, as they worked water masses up and down the Boulevard. For most sondes, measurements were recorded for both the ascent and descent, with descending sondes falling by parachute for all platforms except the R/V Ron Brown. The synoptic environment encountered during EUREC⁴A, the radiosonde measurement strategy, and an analysis of the sonde data are described in more detail by Stephan et al. (2021).

Figure 3Flight time spent at different altitudes by different airborne platforms. Uncrewed aerial systems (UASs) shown in inset.

Download

HALO, the ATR, and most of the UASs emphasized statistical sampling. Hence flight plans did not target specific conditions, except to adjust the ATR flight levels relative to the height of the sub-cloud layer – but this varied relatively little. During planned excursions from its circling flight pattern, HALO also positioned its track for satellite overpasses – one by MISR (5 February 2021) and another by the core GPM satellite (11 February 2020). Measurements from the MPCK+ (a large CloudKite tethered to the R/V MS-Merian) emphasized the lower cloud layer, selecting conditions when clouds seemed favorable. The mini-MPCK was used more for profiling the boundary layer and the cloud-base region and was deployed when conditions allowed. The Twin Otter targeted cloud fields, often flying repeated samples through cloud clusters identified visibly, but also sampled the sub-cloud layer. The P-3 strategy was more mixed; some flights targeted specific conditions, and others were more statistically oriented (for example, to fill gaps in the HALO and ATR sampling strategy). The different sampling strategies are reflected in Fig. 3, where the measurements of HALO are concentrated near 10.2 km and those of the ATR at about 800 m, with relative uniform sampling of the trade wind moist layer by the Twin Otter. Figure 3 also shows the strong emphasis on sampling the lower atmosphere, with relatively uniform coverage of the lower 3 km. Except for the Twin Otter, which was limited to daytime operations, takeoff and landing times of the aircraft were staggered, with three night flights by the P-3, to better sample the diurnal cycle. Data papers for the individual platforms are being prepared and will describe their activities in greater detail.

Figure 4Map showing surface- and subsurface-platform trajectories, colored by (uncalibrated) near-surface water temperature for platforms with near-surface measurements.

2.2 Airborne platforms for measuring the upper ocean and air–sea interface

Four global-class research vessels – all equipped with surface meteorological measurements and underway temperature/salinity sampling devices – and scores of autonomous ocean-observing platforms (AOOPs) were deployed along Tradewind Alley and the Boulevard des Tourbillons. The tracks of the surface vessels are shown in Fig. 4. These tracks, colored by measurements of the near-surface water temperature, show slightly more variability in water temperatures along the Boulevard des Tourbillons, in contrast with more steady westward warming of surface temperatures following the trades along Tradewind Alley. The more dynamic situation along the Boulevard des Tourbillons, as compared to the situation on the Tradewind Alley, required a different measurement strategy. For the former, research vessels actively tracked and surveyed mesoscale features, and for the latter the sampling was more statistical so as to better support the airborne measurements and cloud characterization.

Figure 5Number of profiles sampling seawater properties at the indicated depth. Ship-based profiling is from CTD casts, underway CTDs, XBTs, and moving vessel profilers. AXBTs were dropped by the P-3.

Download

Along Tradewind Alley, the R/V Meteor mostly worked along the line of longitude at 57.25^∘ W between 12.4 and 14.2^∘ N. The R/V Ron Brown, coordinating its measurements with the P-3, was stationed between the NTAS and the MOVE³ moorings in January and in the region upwind of the EUREC⁴A-Circle, near 55^∘ W in February. For both positions, SWIFT buoys were deployed and recovered in coordination with P-3 airborne expendable bathythermograph (AXBT) soundings. A Saildrone, two Wave Gliders, an AutoNaut (Caravela), four underwater gliders, and extensive conductivity–temperature–depth (CTD) casts from the two ships profiled the upper ocean Fig. 5.

Along the Boulevard des Tourbillons the R/V MS-Merian and the R/V Atalante studied the meso- and submesoscale dynamics. Both research vessels extensively profiled the ocean's upper kilometer using a wide assortment of instruments, including underway CTDs, moving vessel profilers, vertical microstructure profilers (VMP and MSS), expendable bathythermographs (XBTs), and expendable CTDs (XCTDs). Three ocean gliders (one SeaExplorer and two Slocum electric gliders) provided dense sampling (more than 1300 profiles, most to at least 700 m, Fig. 5) of subsurface structures associated with mesoscale eddies. Of the roughly 8000 upper-ocean profiles performed during EUREC⁴A, nearly three-fourths were performed in coordination with the eddy sampling along the Boulevard des Tourbillons. Four Saildrones, 22 drifters and four deployments of two air–sea fluxes observing prototypes, OCARINA and PICCOLO, substantially expanded the observations at the ocean–atmosphere interface. Five Argo floats equipped with a dissolved-oxygen sensor were deployed to allow a Lagrangian monitoring of the ocean surface and subsurface dynamics during and after the campaign.

To effectively survey features in the active waters of the Boulevard des Tourbillons, the sampling strategy and cruise plan were assessed daily, using information from the previous day's measurements, updates from satellite products, weather forecasts, and ocean predictions. Tailored satellite products and model predictions were provided by a variety of groups⁴ to help track and follow surface features in near real time.

2.3 Instrument clusters

EUREC⁴A set itself apart from past field studies both through new types of measurements, as performed by individual platforms, but also through the quantity or clustering of certain instruments. Instrument clustering means using similar instruments across a number of platforms so as to improve the statistical characterization of air masses and their evolution. The ability to make such measurements enables estimates of systematic and random measurement errors, giving rise to a different quality of measurement as compared to those made previously, especially in marine environments. Examples are described below and include the use of remote sensing, instruments for measuring stable water isotopologues, and drones. A platform-by-platform listing of the EUREC⁴A instrumentation is provided in Appendix B.

2.3.1 Remote sensing

EUREC⁴A included eight cloud-sensitive Doppler (W- and Ka-band) radars. Four zenith-staring instruments were installed at surface sites (BCO, R/V MS-Merian, R/V Meteor, and R/V Ron Brown) and three on aircraft (nadir, zenith on the ATR, HALO, and the P-3). The ATR flew a second, horizontally staring, Doppler system. Two scanning radars (a C-band system installed on Barbados and a P-3 X-band tail radar) and three profiling rain radars (one at the BCO, another at the Caribbean Institute for Meteorology and Hydrology (CIMH), and a third on the R/V MS-Merian) measured precipitation. The R/V MS-Merian additionally had an X-Band radar installed for wave characteristics and surface currents over a roughly 2 km footprint around the ship. Fourteen lidars were operated, four of which were advanced (high-spectral-resolution, multi-wavelength) Raman or DIAL (differential absorption lidar) systems for profiling water vapor and aerosol/cloud properties. The Raman systems (at the BCO, on the R/V MS-Merian, and on the R/V Meteor) were upward-staring surface-mounted systems, and the DIAL aboard HALO operated in a nadir-staring mode (Wirth et al., 2009). On the ATR a backscatter UV lidar operated alongside the horizontally staring radar, looking horizontally to provide an innovative planform view of cloudiness near cloud base. In total, six wind lidars and three ceilometers were operated from the BCO and all research vessels except for the R/V Atalante. As an example of the sensor synergy arising from the multitude of sensors, Fig. 6 shows water vapor flux profiles (Behrendt et al., 2020) estimated from co-located vertically staring Doppler wind lidar and Raman (water vapor) lidar measurements from the ARTHUS system (Lange et al., 2019) aboard the R/V MS-Merian. This type of measurement strategy, employing a dense network of remote sensors to both improve sampling and realize synergies, is increasingly emphasized for land–atmosphere interaction studies (e.g., Wulfmeyer et al., 2018), but it is more difficult to realize, and thus uncommon, over the ocean.

Figure 6Vertical latent-heat (vaporization enthalpy) flux as a function of time and height above the platform as measured from the combination of water vapor Raman lidar (ARTHUS) and Doppler wind lidar aboard the R/V MS-Merian. The mean value over the 3 d period is 100 W m⁻² at 200 m, and the fluxes are positive throughout the sub-cloud layer.

Download

More standard, but still unprecedented by virtue of its space–time–frequency coverage, was the contribution of airborne, surface, and space-based passive remote sensing to EUREC⁴A. Three 14-channel microwave radiometers operated from surface platforms, and a 25 channel nadir-staring system operated from HALO (Mech et al., 2014; Schnitt et al., 2017). Handheld sun-photometer measurements were made on all four research vessels, and an automated system operated from Ragged Point, near the BCO, provided additional constraints on estimates of aerosol loading (from lidars) and column water vapor (from radiometers). Infrared radiometers for measuring the surface skin temperature were operated on the ATR, HALO, the R/V Ron Brown, the BOREAL, and CU-RAAVEN UASs, as well as on the five Saildrones. For estimating fluxes of radiant energy, broadband longwave and shortwave radiometers were installed on three of the airborne (zenith and nadir) and surface (zenith) platforms. In addition, HALO and the R/V Meteor hosted high-spectral-resolution systems measuring shortwave and near-infrared down- and upwelling radiances (Wendisch et al., 2001). Near-real-time geostationary GOES-East satellite imagery and cloud product retrievals between 19^∘ N–5^∘ S and 49–66^∘ W were collected, with finer temporal resolution every minute (between 14 January and 14 February, with a few data gaps from diversions to support hazardous weather forecasting in other domains) archived over most of this domain. ASTER's high-resolution (15 m visible and near-infrared, and 90 m thermal) imager on board TERRA was activated between 7–17^∘ N and 41–61^∘ W. It recorded 412 images of 60 km × 60 km in 25 overpasses between 11 January and 15 February. These images are complemented by Sentinel-2 data with images at 10 m resolution in some visible–near-infrared bands and 20 m resolution in shortwave-infrared bands relevant for cloud microphysical retrievals.

The intensity of remote sensing instrumentation in the vicinity of the EUREC⁴A-Circle will support efforts to, for the first time, observationally close the column energy budget over the ocean, as well as efforts to test hypotheses that link precipitation to processes across very different time and space scales.

Figure 7Water stable isotopologues mass fractions (a) binned by altitude over all EUREC⁴A measurements, including samples from near-surface waters. Percent of measurements at each altitude (b) associated with winds from the east (45 to 135^∘ from north).

Download

2.3.2 Stable water isotopologues

EUREC⁴A benefited from an unusually complete and spatially extensive network of stable water isotopologue measurements (H${}_{\mathrm{2}}^{\mathrm{18}}$O, H${}_{\mathrm{2}}^{\mathrm{16}}$O, and HDO) distributed across multiple platforms. Seven laser spectrometers and five precipitation sampling systems especially designed to avoid post-sampling re-evaporation were deployed. At the BCO, two laser spectrometers provided robust high-frequency measurements of isotopologues in water vapor and 46 event-based precipitation samples were collected. Three ships – the R/V Atalante, the R/V Meteor, and the R/V Ron Brown – were similarly equipped and in addition collected ocean water samples (340 in total) from the underway water line and the CTDs. These samples have been analyzed in the laboratory together with 50 shipboard rainfall samples. Two of the high-frequency laser spectrometers were mounted on the ATR and P-3 to measure the vertical distribution of water isotopologues. The airborne measurements also added continuity, sampling air masses between the BCO and R/V Meteor stations and between the R/V Meteor and the upwind R/V Ron Brown. The measurements provided very good coverage through the lower (3 km) atmosphere. Air-parcel backward trajectories based on three-dimensional wind fields from the operational ECMWF analyses indicate that boundary layer air came almost exclusively from the east, with a more heterogeneous origin of air masses sampled above 2500 m (Fig. 7; see also Aemisegger et al., 2021). Large-scale context for the in situ measurements will be provided by retrievals of atmospheric HDO and H${}_{\mathrm{2}}^{\mathrm{16}}$O from spaceborne instruments.

The size of the network of isotopologue measurements and the degree of coordination among the different measurement sites will enable investigations of the variability of the stable water isotopologues – in space and time, in ocean water, atmospheric vapor, and precipitation following the trades – that were previously not possible.

Figure 8Ocean and atmospheric boundary layer upwind of the BCO. Dots show CU-RAAVEN measurements of the density potential temperature vs. altitude, as well as underwater glider measurements of the temperature below the surface. Values are normalized to compensate for differences associated with either synoptic variations or from variations in the depth of the sampled planetary boundary layers. Blue dots show profile of cloud fraction from all MPCK profiles. The dashed (black) line marks the potential temperature of near-surface air isentropically lifted from the surface; the slope discontinuity at the lifting condensation level (690 m) marks the shift from an unsaturated to a saturated isentrope. The temperature difference between the sea surface and the lower atmosphere is taken from Saildrone data.

Download

3.1 Testing hypothesized cloud-feedback mechanisms

As described by Bony et al. (2017), EUREC⁴A was conceived as a way to test the hypothesis that enhanced mixing of the lower troposphere desiccates clouds at their base, in ways that warming would enhance (Rieck et al., 2012; Sherwood et al., 2014; Brient et al., 2016; Vial et al., 2016) but the signal of which has not been possible to identify in past measurements (Nuijens et al., 2014). In addition, recent research suggests that clouds in the trades tend to organize in mesoscale patterns (Stevens et al., 2019b) selected by environmental conditions (Bony et al., 2020). These findings raise the additional question as to whether changes in the mesoscale cloud organization with evolving environmental conditions might play a role in low-cloud feedbacks. To address these questions, EUREC⁴A developed techniques to measure the strength of convective-scale and large-scale vertical motions in the lower troposphere, together with the co-incident cloud-base cloud fractions, in addition to other possible drivers of changes in mesoscale cloud patterns, such as coherent structures within the sub-cloud layer, radiative cooling, and air mass trajectories.

Figure 9Divergence of the horizontal wind versus height (a), and vertical pressure velocity versus height (b). Divergence estimated from dropsonde measurements and vertical pressure velocity derived from these for the two sets of circles flown on 5 February. The black dashed line on the rightmost panel denotes vertical pressure velocity averaged over all EUREC⁴A-Circle dropsonde measurements.

Download

To make the desired measurements required HALO and the ATR to fly closely coordinated flight patterns, ideally sampling different phases of the diurnal cycle (Vial et al., 2019). This was realized by HALO circling (at an altitude of 10.2 km) 3.5 times over 210 min. Within this period three full sounding circles were defined by a set of 12 dropsonde launches, one for each 30^∘ change in heading. The start time of successive sounding circles was offset by 15 min so as to distribute the sondes through the period of circling. During this time HALO also provided continuous active and passive remote sensing of the cloud field below. Flying 50 min “box” patterns just above the estimated cloud base (usually near a height of about 800 m, Fig. 3), the ATR provided additional remote sensing, as well as in situ turbulence and cloud microphysical measurements. After two to three box patterns, the ATR flew two to four L-shaped wind-aligned and wind-perpendicular patterns (the “L” in Fig. 1) – at the top, middle, and bottom of the sub-cloud layer – before returning to Barbados to refuel for a second mission. While the ATR was refueling, HALO made an excursion, usually in the direction of the R/V Ron Brown and the NTAS buoy. On all but two occasions the ATR returned to the measurement zone after refueling (about 90 min later) to execute a second round of sampling, accompanied by HALO returning for another 210 min tour of the EUREC⁴A-Circle. All told this resulted in 18 coordinated (4 h) flight segments, one of which involved the P-3 substituting for HALO on one of its nighttime flights.

Figure 10Illustration for January 28 lidar, February 5 lidar and radar, and February 11 radar cloud field observed at cloud base by the ATR with horizontal lidar (detected cloud boundaries denoted by red dots) and radar measurements.

Download

A first target of the flight strategy was the measurement, for each sounding circle, of the vertical profile of mass divergence using dropsondes (following Bony and Stevens, 2019). In Fig. 9 the vertical pressure velocity, ω, estimated from this divergence is averaged over a set of three circles for the two 5 February circling periods. Also shown is the average over all circles over all days. The continuity of the divergence within a circle and across two circling periods – although on some flights vertical motion can change more markedly across sets of circles – gives confidence that the measurements are capturing a physical signal. It also shows, for the first time from measurements on this scale, how the mean ω reduces to the expected climatological profile, with a magnitude (of about 1 h Pa h⁻¹) similar to what is expected if subsidence warming is to balance radiative cooling.

The second target of the flight strategy was the measurement of the cloud fraction at cloud base through horizontal lidar–radar measurements by the ATR. In fields of optically thin shallow cumuli (such as those associated with the cloud patterns observed on 28 January), cloud droplets were too small to be detected by the radar, but the lidar could detect the presence of many successive clouds along a roughly 10 km line of sight, i.e., half of its box-pattern width (Fig. 10; Chazette et al., 2021). In the presence of larger cloud droplets, normally associated with larger or more-water-laden clouds, such as on 11 February, the radar detected larger droplets and rain drops over a range of 10 km (Fig. 10). The lidar–radar synergy will provide, for each ATR box, the cloud fraction and the distribution of cloud geometric and optical properties at cloud base. The second, vertically pointing ATR cloud radar allows a characterization of the aspect ratio of clouds, which may help infer the mesoscale circulations within the cloud field. These measurements, associated with new methods developed to estimate the cloud-base mass flux (Vogel et al., 2020), and to characterize the mesoscale cloud patterns from GOES-16, MODIS, or ASTER satellite observations (Stevens et al., 2019b; Mieslinger et al., 2019; Bony et al., 2020; Denby, 2020; Rasp et al., 2021), will make it possible to test cloud-feedback mechanisms and advance understanding of the processes underlying the formation of the mesoscale cloud patterns, as well as whether they influence the hypothesized feedback mechanisms.

3.2 Quantifying processes influencing warm-rain formation

As highlighted by Bodenschatz et al. (2010), the range of scales, from micro- to megameters, that clouds encompass has long been one of their fascinating aspects. Measurements made during EUREC⁴A quantified, for the first time, the main processes that influence trade wind clouds across this full range of scales. By doing so, long-standing questions in cloud physics were addressed, including (i) whether microphysical processes substantially influence the net amount of rain that forms in warm clouds and (ii) how important is the interplay between warm-rain development and the mesoscale organization of cloud fields. These questions identify precipitation development as the link among processes acting on different scales and hence guided EUREC⁴A's measurement strategy.

On the particle scale, measurements were performed to characterize aerosols and to quantify how small-scale turbulence mixing processes influence droplet kinematic interactions and activation. Aerosol properties and turbulence both imprint themselves on the cloud microstructure and thereby affect the formation of precipitation (Broadwell and Breidenthal, 1982; Cooper et al., 2013; Li et al., 2018; Pöhlker et al., 2018; Wyszogrodzki et al., 2013). In most cases, not only the magnitude, but also the sign of the hypothesized effects can be ambiguous, if not controversial. For example, by acting as an additional source of cloud condensation nuclei (CCN), Saharan dust may retard the formation of precipitation (Levin et al., 1996; Gibson et al., 2007; Bailey et al., 2013), but if present as giant CCN, it may have the opposite effect (Jensen and Nugent, 2017).

On the cloud scale, the intensity of rain and the evaporation of raindrops can lead to downdrafts, cold pools, and mesoscale circulations which can lift air parcels, producing secondary and more sustained convection (e.g., Snodgrass et al., 2009). These cloud-scale circulations, which the EUREC⁴A-Circle measurements quantified, may also change the vigor and mixing characteristics of cloud. This could in turn influence precipitation formation, a process that Seifert and Heus (2013) suggest may be self-reinforcing, consistent with an apparent link between precipitation and mesoscale cloud patterns such as “fish” or “flowers” (Stevens et al., 2019b).

On larger (20 to 200 km) scales, horizontal transport, which determines whether or not Saharan dust reaches the clouds, as well as factors such as the tropospheric stability, or patterns of mesoscale convergence and divergence, which influence cloud vertical development, may affect the efficiency of warm-rain production. In addition to the characterization of the environment from the dropsondes, the positioning of surface measurements (R/V Meteor, R/V Ron Brown, and BCO) helped characterize the Lagrangian evolution of the flow, also in terms of aerosol and cloud properties.

Figure 11Measurements within the Tradewind Alley test section (5 February) define a multi-scale cloud chamber. The figure highlights clustering on different scales. Scanning C-band radar (POLDIRAD) 0.6^∘ scan (11:25:25 UTC) is overlain on the 11:33:41 UTC brightness temperature (at 10.6 µm) measured by GOES-16, with coincident segments of the HALO and ATR flight tracks. Radar images from the ATR (horizontal and zenith) and HALO (nadir) are shown (all radar imagery shares the same color scale), as well as cloud water and updraft velocity from a penetration of cloud by the Twin Otter (later in the day, at 18:32 UTC, near 13.55^∘ N, 58.26^∘ W at 1910 m). Visual image from the specMACS instrument, with POLDIRAD reflectivity contours superimposed, shows the cloud visualization along a segment of the HALO flight track. MPCK+ hologram measurements (made in the southern portion of the circle – 12.25^∘ N, 57.70^∘ W at 1084 m – on 17 February) demonstrate the capability to measure the three-dimensional distribution of individual cloud droplets colored by size.

Figure 11 shows an example of the cascade of measurements, spanning scales covering 10 orders of magnitude. On the smallest O(10⁻⁵ m) scale, a sample holographic image from an instrument mounted on the MPCK+ shows the spatial and size distribution of individual cloud drops. In situ measurements and airborne remote sensing document the cloud microphysical structure and its relationship to the properties of the turbulent wind field. On scales of hundreds of meters to a few kilometers, vertically and horizontally pointing cloud radars and lidars characterize the geometry and the macrophysical properties of clouds. On yet larger O(10⁵ m) scales, the spatial organization and clustering of clouds and precipitation features are captured by satellite, by high-resolution radiometry from high-altitude aircraft, and by the C-band scanning radar, POLDIRAD (Schroth et al., 1988).

Figure 12Aerosol characteristics measured in the Tradewind Alley highlight two periods (31 January to 6 February and 9–12 February) of CCN-laden air. Dust mass density from the R/V Ron Brown (a), which was mostly east of 55^∘ W. Normalized histogram showing the relative frequency of occurrence of different CCN concentration levels (c). Note that the periods of observation at the two locations are only partly overlapping.

Download

An example of how the measurements upwind and downwind of the EUREC⁴A-Circle helped constrain its aerosol environment is shown in Fig. 12. Two periods with larger CCN number concentration (near 450 cm⁻³), both associated with periods of elevated mineral dust, can be identified in measurements made aboard the R/V Ron Brown (east of 55^∘ W) and from the ground station at Ragged Point (Pöhlker et al., 2018). The slight lag of the Ragged Point measurements relative to those on the R/V Ron Brown is consistent with the positioning of the two stations and the westward dust transport by the mean flow. The episodes of elevated dust are believed to be from Saharan dust outbreaks, which are unusual in the (boreal or northern) winter months (Prospero et al., 2020) and can greatly increase CCN number concentrations (Wex et al., 2016). In between these events, CCN number concentration are 3-fold smaller (150 cm⁻³), which we take as representative of the clean maritime environment.

The degree of aerosol variability should aid efforts to untangle the relative role of different factors influencing warm-rain formation. Helping in this regard is that variations in CCN concentrations are not too rapid to call into question the idea of associating a 3 h period of measurements on the EUREC⁴A-Circle with a particular concentration of CCN: 50 % of the Ragged Point measurements change by less than 10 % over a 3 h period, and only 20 % of the time are changes larger than 30 % measured.

3.3 Sub-cloud mass, matter, energy, and momentum budgets

Early field studies extensively and compellingly documented the basic structure of the lower atmosphere in the trades (Riehl et al., 1951; Malkus, 1958; Augstein et al., 1974; Brummer et al., 1974; Garstang and Betts, 1974). What remains poorly understood is the relative role of specific processes, particularly those acting at the mesoscale, in influencing this structure. A specific question that EUREC⁴A aims to answer is the importance of downdrafts, and associated cold pools (Rauber et al., 2007; Zuidema et al., 2012), in influencing boundary layer thermodynamic structure and momentum transport to the surface. A related question is whether the links between the cloud and sub-cloud layer depend on the patterns of convective organization, for instance as a result of differences in the circulation systems that may accompany such patterns.

Figure 13Lidar profiling of the lower atmosphere using the CORAL lidar at the Barbados Cloud Observatory. The upper panel (a) shows the relative humidity in the lower 5 km over the entirety of the campaign. The lower panel (b) shows (from left to right) show the specific humidity over a 4 h period marked by a large intrusion of cloud layer air on 2 February and the associated aerosol/cloud backscatter. Also shown is the Lagrangian evolution of humidity, or backscatter features, with dashed arrows following the descent with time of features (white arrow in the upper panel: following RH feature; black arrow in the lower panel: following backscatter feature) being indicative of the magnitude of vertical velocity variations on different temporal scales. Gray bars on lower plots (at 21 h) are missing data.

Download

For quantifying the sub-cloud layer budgets, as for many other questions, a limiting factor has been an inability to measure mesoscale variability in the vertical motion field. EUREC⁴A's measurements not only address this past short coming, but the ship-based sounding network additionally quantifies the mean vertical motion at different scales. The arrangement of measurements, particularly flight segments, was designed to quantify the Lagrangian evolution of air masses, with legs repeated on every mission at levels attuned to the known structure of the lower troposphere, i.e., near the surface, in the middle, near the top, and just above the sub-cloud layer, as well as in and just above the cloud layer. Past studies using a single aircraft, albeit in a more homogeneous environment, demonstrate that such a strategy can close boundary layer moisture and energy budgets (Stevens et al., 2003). Doing so also aids quantification of the vertical profile of turbulent transport and contributions associated with horizontal heterogeneity and sets the stage for estimating mass and energy budgets through the entire atmospheric column.

To address the measurement challenge posed by an environment rich in mesoscale variability, EUREC⁴A made use of additional aircraft and a larger array of surface measurements (also from uncrewed platforms) as well as extensive ship and airborne active remote sensing, and a network of water stable isotopologues (as presented in Sect. 2.3.2). At the BCO, aboard the R/V Meteor and on the R/V MS-Merian, advanced Raman lidars provided continuous profiling of water vapor, clouds, temperature, and aerosols. The nadir-staring WALES lidar on HALO likewise profiled water vapor, clouds, and aerosols. As an example of this capability, Fig. 13 presents relative humidity data (deduced from temperature and absolute humidity retrievals) from the BCO lidar. These measurements document the time–height evolution of water vapor in the boundary layer, something impossible to assess from in situ measurements, which measure at only a few levels, or soundings, which are sparse in time.

Figure 14Sub-cloud layer lidar wind versus height above the R/V Meteor. Panel (a) shows the value of the wind speed in the sub-cloud layer, above 200 m. Fluctuations of the near-surface wind speed from a 3-hourly running mean value (b) are shown with an expanded vertical scale. The lower time series (c) shows the ratio of the wind speed at 40 m (the lowest remotely sensed level) to its value at 200 m.

Download

The BCO lidar measurements quantify the structure of moist or dry layers in the free atmosphere, as well as variations in the cloud and sub-cloud layers, illustrating days of more nocturnal activity (centered on 1 February), and also features presumed to be the signature of mesoscale circulations. Analyses of Meteor data show a signature of the diel cycle (0.54 K), but it is more pronounced (1.27 K) over the BCO – both at the surface and as sensed by the BCO lidar at 400 m. Both a slight slackening of the winds and an upwind adjustment in response to diurnal heating of the island could be responsible for the amplification of the diel cycle over the BCO.

Possible mesoscale circulations are the focus of the magnification in the lower panels (of Fig. 13). Shown are measurements in the lower 3 km for a 5 h period late on 2 February 2020. During this period aerosol-poor air appears to descend adiabatically into the cloud layer (near 2 km), coincident with a large-scale fold of cloud layer air into the sub-cloud layer. This results in a sharp contact discontinuity (aerosol front) near 21:00 UTC, which extends to the surface and is also evident in the water vapor field. Typically the marine boundary (sub-cloud) layer is viewed as a turbulent layer that primarily interacts with the much-larger-scale evolution of the free atmosphere through small-scale entrainment at its top. Events such as the one shown in Fig. 13 suggest that in addition to downdrafts and the cold pools they feed, circulations on scales commensurate with and larger than the depth of the sub-cloud layer may be important for boundary layer budgets.

Similar considerations also apply to the momentum budget of the trades. In Dixit et al. (2020) idealized large-eddy simulations are shown to underestimate the flux of momentum in the sub-cloud layer, something they hypothesize to arise from an absence of mesoscale circulations in the simulations. As an example of efforts to quantify such processes Fig. 14 shows the total wind speed measured in the sub-cloud layer by the long-range wind lidar aboard the R/V Meteor. The lower panel documents kilometer-scale wind speed variations on the order of 2 ms⁻¹ that extend into the surface layer (derived from the short-range wind lidar, defined with respect to 3-hourly running means). One question asked is whether, for a given surface friction, convectively driven flows can sustain a relatively large near-surface wind, and weaker surface layer wind shear, than expected from shear-driven turbulence alone. The third panel shows that the ratio of wind speeds at 40 m to wind speeds at 200 m, as a measure of surface layer wind shear, is close (ca. 0.95) to unity. Combined with surface heat and momentum fluxes measured by other platforms, the lidars provide a unique opportunity to identify the influence of (moist) convection on wind stress at the surface.

3.4 Ocean mesoscale eddies and sub-mesoscale fronts and filaments

Mesoscale eddies, fronts, and filaments – not unlike the mesoscale circulations that are the subject of increasing attention in the atmosphere – are coherent structures that may be important for linking surface mixed layer to the interior ocean dynamics (Carton, 2010; Mahadevan, 2016; McWilliams, 2016). By virtue of a sharp contrast with their surroundings, these structures can efficiently transport enthalpy, salt, and carbon through the ocean. Though satellite observations have enhanced knowledge of their occurrence and surface imprint (Chelton et al., 2001), the sparsity of direct observations limits our ability to test understanding of such structures, in particular subsurface eddies. Understanding of the role of these types of structures is further limited by their short lifespans (hours to days) and small spatial scales (0.1 to 10 km), which make them difficult to observe. These facts motivated ocean observations during EUREC⁴A, as did recent work suggesting that such coherent structures, in particular localized upwelling, downwelling, straining, stratification variability, wave breaking, and vertical mixing, may couple with and influence atmospheric processes, including cloud formation (Lambaerts et al., 2013; Renault et al., 2016; Foussard et al., 2019).

Figure 15Surface density gradients at different horizontal length scales. Bivariant histogram shows counts versus (1 to 200 km) and strength of gradient as measured by Saildrones (note the log-scale color bar). Inset is power spectral density of the surface density gradients, calculated by averaging periodograms constructed for each vehicle after de-trending the data and smoothing the data with a 2 km Gaussian filter. The red line shows the linear regression best fit slope of −2.3.

Download

To address these questions, measurements during EUREC⁴A attempted to quantify how near-surface currents, density, and waves varied across and within different dynamical regimes, e.g., for mesoscale eddies, fronts, and filaments. Such measurements aimed to answer specific questions not unlike those posed for the atmospheric boundary layer, namely to quantify the contribution of such structures to the spatial and temporal variability of the upper ocean. EUREC⁴A distinguished itself from past campaigns that have attempted similar measurements – LatMix (Shcherbina et al., 2013), OSMOSIS (Buckingham et al., 2016), and CARTHE (D'Asaro et al., 2018) – by virtue of the number and diversity of observing platforms deployed (Saildrones, underwater gliders, instrumentally enhanced surface and subsurface drifters, Wave Gliders, an AutoNaut, and biogeochemical Argo floats). These mapped the ocean down to 1000 m or more, simultaneously across both the Tradewind Alley and the Boulevard des Tourbillons (Fig. 2). These measurements have resulted in an unprecedented view of a large spectrum of ocean temporal and spatial scales across different oceanic environments.

Figure 16Eddies in the Boulevard des Tourbillons (map) with vertical cross section (A–B transect near 10^∘ N, 58^∘ W on the map) showing ship acoustic Doppler current profiler (SADCP) currents from the R/V Atalante (bottom left) and salinity from CTD casts (right). Surface-eddy field derived from satellite altimetry (Pujol et al., 2016). Eddy contours are detected automatically by the TOEddies algorithm (Laxenaire et al., 2018). The position of subsurface eddies (200 to 600 m deep) as identified from the eddy detection method (Nencioli et al., 2010) applied to vector currents measured by SADCPs are shown by red circles. A subsurface eddy freshwater anomaly is indicative of South Atlantic origins.

Download

The richness of structure observed in the upper ocean during EUREC⁴A can be quantified by the distribution of surface temperature fronts. All seagoing platforms contributed to observing the upper-ocean temperature structure, surveying a wide region and a large spectrum of ocean scales, and thus can contribute to this measure of upper-ocean variability. An example from one such platform, a Saildrone, is shown in Fig. 15. The sensitivity of frontal density gradients to spatial resolution was explored by subsampling data from 0.08 to 100 km (Fig. 15). For each length scale, the percentage frequency of each density gradient was calculated. This analysis demonstrates that smaller length scales yield larger density gradients. The largest gradients were found at spatial scales of only 1 km and were associated with strong, local freshening. These are believed to be associated with small-scale, but intense, rain showers, a potentially far-reaching idea given the importance of rain for linking processes at different scales in the atmosphere (e.g., Sect. 3.2). The analysis further documents self-similar (power law) scaling between 19 and 1900 km with a slope of −2.3. There is evidence of a scale break at around 25 km. Surface quasi-geostrophic turbulence generally predicts a slope of $-\mathrm{5}/\mathrm{3}$ or steeper (Callies and Ferrari, 2013; Rocha et al., 2016; Lapeyre, 2017).

A wide array of instruments deployed from all four ships (CTDs, underway CTDs, mounted vessel profilers, microstructure profilers, XBTs, XCTDs, Doppler current meter profilers, five BGC Argo floats) and the seven underwater gliders (e.g., Fig. 5) profiled water properties and ocean currents. This array of measurements, guided by near-real-time satellite data and real-time ship profiling, revealed a surprisingly dense and diverse distribution of mesoscale eddies. All of the measured eddies captured by satellite data (Fig. 16) were shallow, extending to a depth of about 150 m (Fig. 16) and transporting warm and salty North Atlantic tropical water swiftly northward. Below but not aligned with the surface structures and separated by strong stratification, large subsurface anticyclonic eddies (and on some occasions cyclonic eddies) extended from 150 to 800 m and carried large quantities of water from the South Atlantic northward. An example sampled by the R/V Atalante along a southwest- to northeast-aligned transect near 50^∘ N and 58^∘ W is illustrated in Fig. 16. Here a ca. 200 km eddy characterized by a 0.2 PSU freshwater anomaly was measured carrying water, which was likely subducted in the south Atlantic, northward. The anomaly was associated with a circulation of ∼1 ms⁻¹ with maximum velocities near 300 m extending downward to a depth of about 800 m. EUREC⁴A observations such as these will be essential for understanding the complex dynamics of the upper ocean and the extent to which they can be captured by a new generation of kilometer-scale coupled climate models.

Figure 17Near-surface temperature (T_sea) from drifters and gliders in the three EUREC⁴A study regions. Panel (a) shows the tracks of the instruments colored by T_sea. The magnification (upper left) expands the domain of the Caravela (and underwater glider) measurements in Region A (near 57^∘ W). January and February SWIFT buoy (T_sea at −0.3 m) deployments in Region B. Saildrone (T_sea at −0.5 m) measurements across an eddy near 11^∘ N, with anti-cyclonic currents (at −5 m) shown by vectors, in Region C. Panel (b) shows time series of T_sea measurements by the different instruments. Probability density, p, of air–sea temperature differences measured by two SWIFT buoys between 04:00 UTC 4 February and 14:00 UTC 6 February (c).

Download

Figure 18CO₂ fugacity (fCO₂) measurements from different surface vessels (upper right) and versus longitude (lower). The presentation contrasts strong variability in fCO₂ in association with eddies and salinity variations along Boulevard des Tourbillons (orange in lower panel, track in upper right) versus in the Tradewind Alley (blues). Microscopic/epifluorescence image of several filaments of Trichodesmium, an N₂-fixing cyanobacterium that is found in the region (upper left), and mats of seaweed (Sargassum, photo by Wiebke Mohr, upper center) which were frequently observed and difficult to navigate from some of the uncrewed surface vehicles.

3.5 Air–sea interaction

What distinguished EUREC⁴A from the many previous campaigns focused on air–sea interaction was its interest in assessing how circulation systems, in both the ocean and the atmosphere, influence surface exchange processes. These interests extended to interactions with ocean biology and their impact on both CO₂ exchange and profligate amounts of seaweed (Sargassum) that have, in past years, developed into a regional hazard. To study these processes EUREC⁴A made use of a flotilla of uncrewed devices and a wealth of nadir-staring airborne remote sensing, specifically designed to characterize the air–sea interface on a range of scales.

Ocean eddies, fronts, and filaments influence the atmosphere by perturbing air–sea surface fluxes (Chelton and Xie, 2010; O'Neill et al., 2012) – a process that may also feed back on the ocean by causing a damping of the (sub)mesoscale activity (Renault et al., 2018). As an example, Sullivan et al. (2020) use large-eddy simulation to show how small-scale ocean fronts perturb the boundary layer through its depth, giving rise to circulations on scales much larger than that of the boundary layer, or of the front itself (their Fig. 12). These lead to large perturbations in vertical mixing and, one can speculate, on patterns of cloudiness. Similarly, clouds influence the downward longwave and shortwave irradiance, which influences both the sea surface temperature and atmospheric temperatures directly, something that Naumann et al. (2019) have shown to commensurately power (2 to 200 km) circulations.

In the area near and within the EUREC⁴A-Circle (Region A), measurements sought to quantify how surface exchange processes vary with circulation (cloud pattern) regime. Measurements by Caravela (an AutoNaut) and three underwater gliders characterized the air–sea interface in a small, and spatially fixed, (ca. 10 km) region in this domain (Fig. 17). These measurements help untangle spatial from temporal variability, with both a secular (seasonal) cooling of surface waters over the course of the campaign and a variable, but at times pronounced, diel cycle (Fig. 17). In addition, CTD casts, lower atmospheric profiling (with a mini-MPCK and a quadcopter), and eddy-covariance measurements from an outrigger mast were performed by the R/V Meteor as it steamed up and down the 57.25^∘ W meridian bisecting the EUREC⁴A-Circle just upwind of Caravela's box. Rounding out the measurements in this region were low-level Twin Otter, ATR (as part of its “L” pattern) legs, and BOREAL UAS measurements, as well as airborne remote sensing of sea surface temperatures along the EUREC⁴A-Circle by HALO. Based on preliminary analyses, these measurements are proving useful in quantifying the diel cycle in both the upper ocean and in the lower atmosphere.

Effects of ocean sub-mesoscale processes on air–sea interactions were the focus of measurements in Region B (Fig. 1). On two occasions the R/V Ron Brown deployed six SWIFT drifters (spar buoys) in regions of surface heterogeneity: once in January near the NTAS buoy and again in early February near 55^∘ W. The deployments were performed and coordinated with further measurements by the R/V Ron Brown, as well as by the P-3, two Wave Gliders, and a Saildrone. The P-3 (see also Figs. 4 and 2) dropped AXBTs around the SWIFTS, quantified air–sea exchange with near-surface flight legs, and surveyed the near-surface wind and wave fields using remote sensing. Figure 17 documents how, during the February deployment, the SWIFTS sampled large 0.5 K mesoscale (ca. 30 km) variability in sea surface temperature (SST) features. This variability gives rise to air–sea temperature differences twice as large as the baseline, as inferred from the average of measurements over longer periods (i.e., as shown by the Saildrone data, orange lines) and is characteristic of the SWIFT data away from the local feature in surface temperatures (e.g., green solid line in Fig. 17).

In the Boulevard des Tourbillons (Region C), coordinated sampling between Saildrones and two research vessels aimed to quantify mesoscale and submesoscale air–sea interaction. Submesoscale variability and strong near-surface currents, with a circulation indicative of an NBC ring, were measured by the Saildrones (Fig. 17). These measurements were coordinated with the activities of the R/V Atalante and R/V MS-Merian, as well as three underwater gliders (e.g., Fig. 4). Extensive vertical profiling, also by high-speed underway CTDs, aimed to quantify the impacts of submesoscale fronts and filaments and mesoscale eddies on surface exchange processes, and vice versa. Being able to resolve the thermal structure of the upper ocean should also help quantify the importance of the O(0.3 K) cool-skin effect and diurnal warming just below the skin layer (Fairall et al., 1996) on ocean mixing and air–sea exchange.

Factors, including the role of meso- and sub-mesoscale variability, influencing air–sea gas exchange were also studied. pCO₂ measurements were made on the R/V Atalante, R/V MS-Merian, and the R/V Ron Brown (Fig. 18). In addition, both the R/V MS-Merian and R/V Meteor regularly sampled water at four different depths (selected based on chlorophyll concentrations) for N₂ fixation and primary production rates as well as potential aerobic methane production. DNA- and RNA-based sequencing will additionally be performed on these water samples to identify diazotrophic community members, potentially including so far unrecognized members. Furthermore, large floating mats of seaweed (genus Sargassum) were observed from all crewed platforms. On the R/V MS-Merian, to investigate if, and to what degree, this biomass and primary production can be supported by local N₂ fixation, incubation experiments including stable isotopes were conducted on seaweed samples that were collected underway. In addition to extending studies of air–sea interaction to incorporate chemical and biological processes, EUREC⁴A may also shed light on the role of meso- and submesoscale ocean circulations on these chemical and biological processes.

Figure 19Global 2.5 km mesh simulations performed by ICON as part of DYAMOND-Winter for the EUREC⁴A period. The snapshot, with a magnification over the study region to show the degree of detail in the simulations, was taken from 2 February of a simulation initialized on 20 January and allowed to freely evolve thereafter.

3.6 Benchmarks for modeling and satellite retrievals

The range of scales and types of processes that can presently be captured by both satellites and models, and the extent to which they were integrated into EUREC⁴A's experimental design (see Bony et al., 2017), allows EUREC⁴A to address questions that could not be addressed with data from earlier field studies. For instance, what resolution is required for atmospheric models with an explicit (fluid-dynamical) representation of clouds and convection to represent the vertical structure of the lower troposphere, and its interaction with mesoscale vertical motion and upper-ocean variability, within the observational uncertainty? The fine scale of the EUREC⁴A measurements also makes it possible to quantify satellite retrieval uncertainty, for instance for measurements of small-scale precipitation features, cloud microphysical properties, or column energy budgets (Illingworth et al., 2015).

Figure 20Geostationary satellite image showing the cloud field in the measurement area on 5 February 2020. Snapshots of cloud fields over 50 km × 50 km subdomains labeled “A” to “D” are compared to large-eddy simulation. The large-eddy simulation employed a 100 m mesh, doubly periodic horizontal boundary conditions, and horizontally homogeneous mean forcing is estimated from measurements in the vicinity of subdomain “D”.

For these purposes EUREC⁴A was closely coordinated with efforts to develop and test a new generation of Earth-system models. Recently, following the pioneering efforts of Japanese colleagues (Tomita et al., 2005), a number of groups in other countries have demonstrated – within the DYAMOND⁵ project (Stevens et al., 2019c) – the capability of performing kilometer-scale simulations on global (atmospheric) grids (Satoh et al., 2019). A follow-up, called DYAMOND-Winter, is extending this capability to also include coupled global models and has been coordinated to simulate the EUREC⁴A period. DYAMOND-Winter simulations are being initialized from observational analyses on 20 January and run for at least 40 d. With grid scales of a few kilometers in the atmosphere (and ocean for coupled runs) the simulations explicitly represent scales of motion similar to those observed, all as part of a consistently represented global circulation. This enables investigations of processes influencing the mesoscale organization of fields of shallow convection, including the possible role of surface ocean features, as well as a critical evaluation of the simulations.

An example of an uncoupled DYAMOND-Winter simulation using ICON is given in Fig. 19. The simulated cloud fields exhibit rich mesoscale variability whose structure, while plausible, begs a more quantitative evaluation. The combination of the field measurements and simulations with realistic variability on the mesoscale will aid efforts to test retrievals of physical quantities from satellite radiances. This should make it possible to establish a self-consistent and quantitative understanding of controls on cloudiness.

In addition to the global coupled modeling activities, coordinating modeling activities using much-higher-resolution (meters to tens of meters) simulations of the ocean, atmosphere, and the coupled system over a limited area are ongoing. These include idealized simulations with doubly periodic boundary conditions, atmospheric simulations designed to track the Lagrangian evolution of the flow, and simulations with open boundaries matched either to meteorological/oceanographic analyses or the free-running global simulations. Few if any field studies have benefited from such a rich complement of modeling activities.

Some of the challenges to evaluating these simulations are illustrated with the help of preliminary, but idealized, large-eddy simulations with the forcing specified based on preliminary data in a manner similar to what has been adopted in past studies (e.g., Stevens et al., 2005; vanZanten et al., 2011), albeit (in the present case) over considerably larger domains. Figure 20 shows, with the help of a satellite image, the degree of mesoscale cloud variability. This apparent whimsicality suggests that, given the imprecision in the forcing and the cloud retrievals, assessing the magnitude of systematic biases in the simulations will be a challenge. In this case, the simulations performed for the mean conditions in the vicinity of “D” seem implausible. The challenge will be to assess to what extent this reflects imprecision in the forcing, of the sort that differentiates the different marked regions in the figure.

Given a demonstration that fine-scale models can quantitatively represent the macro-structure of the observed clouds, EUREC⁴A measurements are expected to provide benchmarks for the simulation of cloud microphysical process. This would allow the first ever evaluation of the ability of microphysical models, which depend on a variety of parameterized processes, to quantitatively represent precipitation formation processes in realistically simulated cloud fields. Previous attempts (Ackerman et al., 2009; vanZanten et al., 2011) at making such an evaluation have highlighted large differences in models, but it remains unclear to what extent these differences are due to the representation of cloud macrophysics versus microphysics. Greater confidence in the fidelity of these simulation approaches will also greatly benefit their application to questions in remote sensing.

3.7 Scientific outreach and capacity building

A core, and hopefully sustainable, feature of the EUREC⁴A field campaign was the rich human and scientific interactions with the Barbadian public, the regional research community, and the larger community of scientists from outside of the region. Activities that permitted these exchanges included operational support for flight planning, inclusion in flight teams and on-ship data collection teams, weekly seminars, a larger symposium, and scientific outreach to schools and to the general public. A total of more than 25 researchers from the region as well as representatives of regional governments contributed to the data collection. This participation was essential not just for meeting EUREC⁴A's original objectives, but it also expanded the scope of activities to include the development of regional climate resilience and capacity building in the use of advanced weather and climate early warning systems.

3.7.2 Symposium and scientific seminars

Knowledge transfers of immense value were facilitated by the organization of regular scientific presentations that provided an opportunity for exchange among EUREC⁴A participants and researchers at the Caribbean Institute for Meteorology and Hydrology (CIMH). It is expected that such exchanges will sustain collaborations well beyond the campaign. Keynote presentations at the Barbados Museum and Historical Society brought to a general audience the goals of the EUREC⁴A campaign, the very early history of meteorology on Barbados, and issues or relevance to climate change and related adaptation.

Campaign participants also celebrated the 50th anniversary of the Barbados Oceanographic and Meteorological EXperiment (BOMEX) field campaign with a 2 d public symposium entitled “From BOMEX to EUREC⁴A”. The symposium brought together a varied audience, including regional and international scientists, EUREC⁴A participants, and students from the University of the West Indies Cave Hill Campus (Fig. 21). The symposium provided an opportunity to reflect upon the evolution of climate research during the past 50 years. From fascinating speeches by BOMEX veterans, to presentations describing the state of present-day understanding as expressed in EUREC⁴A's objectives, the symposium helped contextualize the efforts being made as part of EUREC⁴A.

In December 2019, prior to the start of the EUREC⁴A campaign, the Caribbean Meteorological Organization (CMO) Headquarters Unit and the University of Leeds, with the assistance of CIMH, organized a Caribbean Weather Forecasting Initiative workshop at CIMH to promote knowledge exchange between researchers and forecasters from CMO members and other Caribbean states. This activity supported the EUREC⁴A forecast testbed, in which many of the regional forecasters who participated in the workshop provided some of the daily forecasts previously discussed.

Figure 21“From BOMEX to EUREC⁴A” symposium participants (photo by Frédéric Batier). Pat S. R. Callender (front, third from right) and Clyde Outram (front center, with cane) participated in BOMEX.

3.7.3 Scientific outreach in schools and facility visits

Scientific outreach activities, such as school visits and scientific open houses, sought to sensitize the public, in particular school children, to the EUREC⁴A program and its important role in addressing issues of severe weather and climate change. Together with the CIMH, and the Barbados Ministry of Education, 10 visits to primary and secondary schools in Barbados were arranged. Simple experiments were designed and performed with the children to help build intuition as to the underlying atmospheric and ocean processes relevant to EUREC⁴A's scientific objectives, not to mention the weather phenomena that surrounds them on a daily basis. These outreach efforts aimed to raise awareness in ways that would increase the resilience of the region to weather and climate extremes, to support citizen science, and to expose young people to scientific career paths.

The open houses consisted of guided tours of many of the measurement platforms. This included tours of the research ships; visits to the BCO at Deebles Point, St. Philip (where visitors could help launch radiosondes); the aerosol measurement facility on nearby Ragged Point; BOREAL, Skywalker, and CU-RAAVEN drone launches at Morgan Lewis Beach, St. Andrew; and tours of the POLDIRAD radar at Colleton, St. John. As an informal complement to the symposium, the outreach activities provided a window into the daily life of the campaign and gathered a diverse audience, from local Barbadians to the scientists involved in EUREC⁴A. The success of EUREC⁴A's outreach efforts is perhaps best exemplified by the ad hoc team of young engineers (Fig. 22) that helped flight proof the drones before their launch from Morgan Lewis Beach.

Further activities included members of the scientific team planting trees as part of Barbados' “We Planting” initiative. In parallel to the scientific campaign, two French filmmakers also visited Barbados for the duration of the EUREC⁴A field campaign to shoot a documentary combining scientific, cultural, and historical elements of the island of Barbados. From their material an additional short scientific documentary of the campaign was created and is provided as a Supplement (video asset) with this paper.

Figure 22Local children helping to evaluate air worthiness of CU-RAAVEN UAS prior to launch from Morgan Lewis Beach (photo by Sandrine Bony).

4.1 Data

The data collected during EUREC⁴A will, in different stages of development, be uploaded and archived on the AERIS data center. The AERIS data center is part of the French Data Terra research infrastructure, which has the objective to facilitate and enhance the use of atmospheric data, whether from satellite, aircraft, balloon, or ground observations, or from laboratory experiments. It generates advanced products and provides services to facilitate data use, to prepare campaigns, and to interface with modeling activities.

In addition, emphasis is being placed on the publishing of datasets through a special collection of articles in Earth System Science Data. Many of these data papers will involve the construction of cross-platform datasets, for instance for the upper-air network, or dropsondes, isotope measurements, or classes of remote sensors. At the end of the data collection phase, all data on AERIS will be mirrored by the Caribbean Institute for Meteorology and Hydrology in Barbados.

4.2 Environmental impact

EUREC⁴A was motivated by an interest to better anticipate how Earth's climate will change with warming. This makes it all the more relevant to ask how EUREC⁴A exacerbates the problems it attempts to understand, or more pertinently, whether it made appropriate use of scarce resources. The first step in answering this question is to estimate the magnitude of its environmental impact. We do so here mostly in terms of EUREC⁴A's carbon footprint, which we estimate (see Appendix A for details) to be 5000 t of CO₂. The marginal increase – EUREC⁴A took place at the expense of other campaigns – is of course much less. The main contribution to the carbon footprint was from fossil fuels (kerosene and diesel) used to power the research platforms. The travel of the participating scientists contributed non-negligibly (5 %) to EUREC⁴A's carbon footprint and provides context for the total emissions.

When learning about EUREC⁴A many people become concerned about the environmental impact of the dropsondes – a concern shared by some of the present authors. The sondes have been designed to sink to the ocean floor after descending to the sea surface. As elaborated upon in Appendix A, this along with the choice of materials (including batteries), and their small size (which with planned modifications may be reduced by a further factor of 2), results in the environmental impact of the use of even a very large number of sondes themselves being minimal. One hesitates to call any environmental impact negligible, but compared to many of the other activities – let alone the initial emotional response to the idea of throwing objects out of an aircraft – this is probably an apt description.

The potential of using yet smaller sondes, or smaller platforms in general, to further reduce environmental impacts was vividly illustrated by EUREC⁴A's extensive use of robotic sensors (uncrewed aerial systems, UASs; autonomous ocean-observing platforms, AOOPs). In many cases these provided more agile and less energy intensive ways of sampling the environment. For instance, the Saildrones, Wave Gliders, and AutoNaut (Caravela) make use of renewable energy sources for their propulsion (e.g., wind for Saildrones, waves for the Wave Gliders and AutoNaut), and for their scientific sensors (solar panels).

Validation for EUREC⁴A's use of the resources was experienced not just through the data collected, but also through the social interactions that the campaign enabled. These, as discussed in Sect. 3.7, were expressed in bonds of friendship that were established and through the many opportunities that were presented through the outreach and capacity building activities (Fig. 22). Further validation of EUREC⁴A's use of resources depends on how the gained data advances scientific understanding to help humanity. This ultimately depends on what is done with the data – something over which we, the authors, have considerable influence and responsibility. We very much hope this realization will motivate a determination to learn as much as possible from the EUREC⁴A measurements and that the importance of supporting such efforts is recognized by funding agencies. We also acknowledge the imperative this creates to make the hard-won data easy to access and use, e.g., through data papers, and to also communicate what we learned from our efforts, as widely and freely as possible.