We use in situ and ground-based remote sensing data from the BCO , which has been operated by the Max Planck Institute for Meteorology together with the Caribbean Institute for Meteorology and Hydrology since April 2010. The BCO is located atop a 17 m cliff on an eastward promontory of Barbados called Deebles Point (13.16∘ N, 59.43∘ W) and samples nearly undisturbed Atlantic trade-wind conditions. We have used surface meteorology and micro-rain radar (MRR) data since January 2011, cloud radar data since January 2012, and Doppler lidar data from March 2016 until March 2021. All data are aggregated into 1 min averages. The instruments used and meteorological variables derived are explained in the following. More details about the BCO and its instrumentation can be found in and .

The pattern of mesoscale cloud organization at the BCO for the period January 2018 to March 2021 is classified by a neural network algorithm applied to infrared satellite images from the Geostationary Operational Environmental Satellite 16 (GOES-16). We use brightness temperature retrievals every 30 min from the 10.35 µm channel at a spatial resolution of 2 km from the Advanced Baseline Imager (ABI) Level-1b data product , over a large domain including Barbados (45–66∘ W, 9.3–23.3∘ N).

The neural network based on the RetinaNet algorithm was initially trained on and applied to visible images in and later retrained and applied to infrared images by . The use of infrared images also allows study of the diurnal cycle of the mesoscale organization . The classifications of the neural network are rectangles of various sizes that belong to either the Sugar, Gravel, Flowers, or Fish pattern. We select every classified rectangle that overlaps with the BCO location. Periods without a classification are labelled as “No”. The BCO data at 1 min resolution are considered contemporaneous with the nearest 30 min neural network classification. If a given pattern is present for more than 75 % of the duration of a cold pool, the cold pool is categorized by this pattern.

At any given time, multiple rectangles of different sizes of the same and different patterns can occur. Multiple rectangles of the same pattern are combined and counted only once, while multiple rectangles of different patterns are counted separately. This leads to locations in the images, and hence data in the meteorological time series, being classified e.g. as both Gravel and Flowers. Excluding situations with multiple patterns only marginally influences the results but reduces the sample size considerably as previously noted in. Ambiguities in the classification can be physical – for example due to regime transitions or similarities between patterns – or related to ambiguities introduced to the neural network by disagreement in the human classifications. The occurrence of multiple patterns can be reduced if a stricter threshold is used for the agreement score representing the confidence of the neural network prediction here set to 0.4 as in, but this again reduces the sample size.

We detect cold pools by identifying abrupt drops in the BCO surface temperature time series following . We first filter the 1 min averaged temperature time series with an 11 min running average. We then classify all filtered 1 min temperature drops δT=Tfil(t)-Tfil(t-1)<-0.05 K (per minute) as a cold-pool candidate (see Fig.  for an illustration). For every candidate cold pool, we detect the time of the cold-pool front onset (tmax), the time of the minimum temperature (tmin), and the end of the cold pool (tend) as follows.

tmax: the onset of the cold-pool front tmax is defined as the last instance of δT>0 K within 20 min before the initial abrupt temperature drop with δT<-0.05 K. If the temperature is falling continuously in this period, tmax is chosen as the time of the maximum temperature (that is, 20 min before the abrupt temperature drop). We refer to the smoothed temperature at tmax as Tmax.

The period between tmax and tmin is referred to as the cold-pool front and the period between tmin and tend as the cold-pool wake.

Our cold-pool detection algorithm is similar to the one used by but with the important modification that we only identify cold pools for situations with abrupt temperature drops exceeding our threshold of δT<-0.05 K. With our algorithm we thus filter out both turbulent fluctuations and advective or diurnal patterns of temperature variability. The threshold of δT<-0.05 K is subjectively chosen based on visual impression and represents distinct variations in temperature. For an 11 min averaging window, a δT of -0.05 K corresponds to about 2 % of the data. Figure  shows example cold pools for all patterns and illustrates the algorithm. In the next subsection we briefly discuss the strengths and weaknesses of the algorithm based on these examples.

Time series of example cold-pool days along with corresponding satellite images are shown for every pattern in Fig.  and shed some light on the differences in the cold-pool characteristics of the four patterns. The two Sugar cold pools stem from isolated precipitating deeper cumuli. The satellite image captures the deeper cloud over the BCO at the time of the first cold pool and also indicates some organization of the cumuli in lines upstream the BCO, while the canonical Sugar fields of shallow cumuli pass further north. The textbook-like Gravel example day is characterized by many short and often weak cold pools quickly following each other, interspersed by stronger cold pools. The stronger cold pools are associated with the presence of strongly precipitating deeper clouds (note that the radar did not work prior to 12:00 LT). The many cold pools present on this day clearly imprint their signature on the satellite image in the form of mesoscale arcs.

The cold pools on the Flowers day are associated with the large cloud system whose stratiform layer reaches the BCO at 10:00 LT. Three cold pools are directly associated with the large system, with the first one starting at 11:00 LT, showing a very strong ΔT of -3.85 K. The large system has rain rates up to 3.6 mm h-1 and is preceded by a weaker cold pool at 09:30 LT associated with the very thin mesoscale arc visible in the satellite image. This first weak cold pool goes along with a strong increase in humidity of 1.3 g kg-1. The Fish day features a 6 h-long cold pool associated with steady and intense rain (maximum RR of 11.6 mm h-1), continued strong downdrafts, and very high humidity throughout its entire duration. The temperature fully recovers within about 20 min of the cold-pool end, and 3 h later two subsequent pronounced cold pools follow that are again characterized by continued precipitation and downdrafts. The satellite image shows the fish-bone-like cloud band typically associated with the Fish pattern, which is strongly connected to trailing cold fronts of extratropical origins . The more front-like character of the Fish cold pools with steady showers and downdrafts is clearly evident. While most of the cooling is expected to stem from the evaporating precipitation, we cannot rule out a small cooling contribution related to a larger-scale temperature contrast between the south and north of the Fish cloud band see also.

The example cases highlight how well the detection algorithm works in these diverse situations. Abrupt strong temperature drops are reliably detected, successive fronts sensibly combined into one single cold pool, and even the 6 h-long cold pool with frontal character on the Fish day is correctly identified.

The examples also indicate some challenges of the cold-pool identification. Although they look like cold pools, some temperature drops on the Gravel and Sugar days are not identified as cold pools because they are either not abrupt enough (δT≥-0.05 K) or not strong enough (ΔT≥-0.4 K). The difficulty in defining the end of the cold-pool wake is illustrated in the Fish case: the cold pool starting shortly before 16:00 LT lasts until well after 18:00 LT, but the temperature drop near 17:00 LT causes a premature end of the cold pool. A temperature drop of this magnitude could also be caused by the diel cycle in temperature. The cold-pool end definition could be improved by an additional rain or downdraft requirement to more robustly distinguish between cold-pool activity and other processes. Because most analyses and diagnostics computed in this study focus entirely on the cold-pool front (see next section), not fully representing the wake of rare long-lasting cold pools is a minor issue that could only influence the overall cold-pool fraction and the duration statistics.

As mentioned in Sect. , the organization pattern definition is somewhat ambiguous. Also among the example days shown in Fig. , multiple cloud patterns pertain to some cold pools. For the Flowers case, the 2 h at the beginning and end of the period shown are also classified, respectively, as Gravel and Fish. In the Sugar case, only the period between 09:00 and 16:00 LT is exclusively classified as Sugar, while the periods before and after are also partly classified as Gravel. Most surprisingly, the textbook Gravel day is also entirely classified as Flowers, and also setting a stricter agreement score of 0.5 leaves half of the day co-classified as Flowers. This indicates that distinguishing Gravel from Flowers can be particularly challenging as also shown in. The Fish day is very confidently classified and no other pattern is detected for the entire day.

For the subsequent analyses, we apply a number of selection criteria to make the comparison of cold pools more robust. That is, we only consider cold pools with ΔT<-0.4 K and less than two missing values in the filtered temperature time series during the entire cold-pool duration (set all with 9234 cold pools). For the analyses of the cold-pool properties we further apply a criterion of no non-recovered cold pool in the hour prior to the cold-pool onset (set noprev with 8772 cold pools), which selects cold pools moving into an initially undisturbed atmosphere that is not modified by previous convection. Except for Appendix B, we also focus on the dry winter regime from December to April (set noprevWI with 3889 cold pools), which is characterized by steady easterlies, subsiding large-scale motion in the free troposphere and the predominance of shallow trade-wind convection .

All these selection criteria reduce the cold-pool sample size considerably. They represent a trade-off between ensuring a robust and unbiased sample to address our research questions while not being unnecessarily strict and removing too many cold pools. The selection criteria are thus somewhat subjective and differ among studies. For example, used the criterion of no rain in the hour prior to the cold-pool onset to select cold pools unmodified by previous convection, whereas we achieve the same goal with the criterion of no non-recovered cold pool in the prior hour, which excludes less cold pools in our case (i.e. about 2500 additional cold pools would be discarded with the criterion of ). Instead of focusing on the winter regime, we could have also set a criterion based on the cloud-top height to focus on trade cumulus cold pools. However, as this would restrict the analysis to periods when the radar is running and – as we are relying on single-site measurements – the parent convection might not move over the BCO in its entirety, we would likely exclude too many cold pools with a CTH criterion without even being sure that periods of deep convection are really excluded. Despite the rather strict criteria applied here, the long time series leads to a much larger number of cold pools analysed than in previous studies.

Another potential sampling issue regarding the single-site measurements is that it is not clear at which stage of its life cycle we sample the cold pool and where we sample it with respect to its centre. Assuming azimuthally symmetric wind variations around the cold-pool centre, which in case of little wind shear is a good approximation , the change in wind direction from the mean direction prior to the cold-pool onset could give a hint as to the location relative to the cold-pool centre. Due to our large sample size, a potential random bias is likely to be small. The influence of wind shear on the propagation direction and characteristics of cold pools is an interesting topic for a future study.

If not mentioned differently, diagnostics for each cold pool are computed either as the minimum difference (ΔXmin) or maximum difference (ΔXmax) of a variable X across the cold-pool duration, e.g. ΔXmax=max(X((tmax+1):tend)-X(tmax)). If the cold-pool wake lasts longer than 20 min, the diagnostics are computed only until 20 min after tmin (instead of until tend) to prevent problems in case of a poorly defined cold-pool end. Similarly, Xmean or Xmax are the mean or maximum of variable X over the same analysis period (indicated in dark red in Fig. ). For the Doppler lidar vertical velocities, we diagnose wmaxSCL (wmax⁡450) as the maximum wSCL (w450) in the first half of the front (including the last 10 min before tmax) and wminSCL as the minimum wSCL in the second half of the front (including the first 10 min after tmin). Unless otherwise stated, the surface meteorology diagnostics are computed from the 11 min filtered time series.

Along with most diagnostics and composites, we show the standard error (SE), which measures how well the median or mean of a given sample can be estimated. The SE of the median is computed as IQR/n, where IQR represents the inter-quartile range and n the sample size, and the SE of the mean as σ/n, where σ is the standard deviation. As not all instruments were running all the time, some diagnostics are only available for a subset of the cold pools, and the sample size is adjusted accordingly when computing the SE.

In total we detect 3889 cold pools that meet the criteria of ΔT<-0.4 K and less than two missing values in Tfil in the winter seasons considered. We find that cold pools are very frequent at the BCO, and on 73 % of days at least one cold pool is detected. The BCO is on average affected by cold pools during 7.8 % of the day (i.e. 112 min) and by a cold-pool front during 4.4 % of the day. The median daily cold-pool fractions are about one-third smaller than the means mentioned, indicating that there are some days with a very large cold-pool fraction. The mean daily cold-pool fraction of 8.6 % for January and February 2011–2021 is also very close to the 7 % found by during the EUREC4A campaign in January and February 2020, despite their very different method defining cold pools in atmospheric soundings based on a mixed-layer depth criterion.

Table  presents statistics of the most important cold-pool properties for the set of winter cold pools with no non-recovered cold pool in the prior hour (noprevWI). It shows that 50 % of the cold pools have a temperature drop exceeding 0.9 K across the front (the unfiltered temperature drop is 0.3 K stronger), a Δqmax exceeding 0.2 g kg-1 and a Δqmin below -0.43 g kg-1, decreases in θe and θv exceeding -2.1 K and -0.96 K, respectively, a Δpmax exceeding 9 Pa, and a ΔUmax larger than 1.14 m s-1 (with the unfiltered anomaly being more than twice as large). The median rain intensity measured by the MRR is 0.9 mm h-1. Furthermore, 50 % of the cold pools are associated with a maximum cloud-top height exceeding 3 km and wmaxSCL and wminSCL of 0.9 m s-1 and -0.55 m s-1 near the onset and end of the front, respectively.

The average cold-pool duration is 33 min, of which a bit more than half of the time pertains to the front. Multiplying the duration by the surface wind speed yields a median cold-pool length larger than 13.3 km. This median cold-pool duration and length may seem small compared to satellite imagery, in which mesoscale cold-pool arcs can easily span 100 km. Also, the largest 2 % of cold pools are hardly larger than 40 km. The smaller cold-pool sizes found here are likely due to the algorithm sampling mostly the edge of the cold pools and due to the challenges of defining the cold-pool end purely based on the surface temperature time series (see discussion in Sect. ).

The IQR shows that all these medians are associated with substantial variability, especially for the humidity and rain variables. However, focusing on the winter regime generally reduces the IQR of the diagnostics compared to all seasons (not shown), suggesting that this criterion indeed results in a more homogeneous cold-pool sample representative of the trade cumulus regime.

Table  also compares the median ± IQR of the 25 % strongest and weakest cold pools in terms of ΔT. The strongest cold pools last longer, follow each other more quickly (lower Δtnextcp), and are associated with deeper clouds, more rain, stronger downdrafts, humidity drops and wind gusts, and larger positive vertical velocities at the beginning of the front compared to weaker cold pools. This agrees well with the conceptual picture of deeper clouds producing more rain and having a larger potential for rain evaporation, which drives stronger downdrafts that bring down more dry air from further aloft and which induces a stronger cooling and a stronger gust front that is associated with stronger rising motion at its leading edge. Similar but slightly smaller differences between stronger and weaker cold pools are found when comparing cold pools associated with the 25 % strongest versus weakest downdrafts or the 25 % deepest versus shallowest CTHmax (not shown).

The rain duration in the front is the diagnostic that explains most variability in ΔT (R2=0.364). Rain duration is well correlated (R=0.47) with the front duration, which itself explains a comparable amount of variability in ΔT (R2=0.357). That the accumulated rain amount in the front explains less variability in ΔT (R2=0.21) than the rain duration indicates that the rain intensity is of secondary importance. Another important predictor of ΔT is the downdraft strength wminSCL (R2=0.23), which together with the front duration explains 50 % of the variability in ΔT for the noprevWI set. The CTH usually scales with the precipitation amount for trade cumuli , and CTHmax also explains some variability in ΔT (R2=0.10). That the rain duration, the downdraft strength, and the maximum CTH also distinguish the cold-pool properties well indicates that the parent convection triggering the cold pool is sampled well by the single-point measurements.

Figure  shows the composite mean temporal structure of the perturbations associated with the cold-pool passages. To facilitate the comparison of different cold pools, we use a normalized time coordinate within the cold-pool front with values after tmax and before tmin mapped onto 20 points (the median front duration), similar to previous studies .

The temperature of the composite-mean cold pool, after increasing slightly before tmax, decreases rapidly in the front to -1.15 K and recovers by ΔT/e within 16 min after tmin (not shown). The temperature remains about 0.5 K below Tmax in the hour after the frontal passage. The temperature drop in the front of the 25 % strongest cold pools is by definition stronger but with a mean tendency of -0.070 K min-1 also more than twice as abrupt compared to the weakest cold pools. The strongest cold pools also take longer to recover than the weakest ones.

The temporal structure of the specific humidity response is intriguing. The composite-mean humidity starts to increase already 8 min before tmax and increases by about 0.2 g kg-1 until tmax. In the first quarter of the front, the humidity increases by another 0.2 g kg-1 before it drops to its minimum of -0.25 g kg-1 at tmin, which is hardly lower than the pre-front value. The specific humidity response of the strongest cold pools only differs significantly from the weakest cold pools at tmin, with the humidity drop at tmin being about -0.4 g kg-1 and thus about twice as strong as the drop for the weakest cold pools. If the entire set of cold pools including the summer season with deeper convection is used, the strongest cold pools have a significantly weaker positive humidity anomaly at the beginning of the front and a significantly faster and stronger humidity reduction at tmin compared to the weakest cold pools (see Fig. c–d).

The humidity recovers much more quickly than the temperature and remains slightly elevated compared to its pre-front value in the hour after. The fast humidity recovery might be due to the trapping of surface moisture fluxes in the anomalously shallow mixed layer typically associated with cold pools . Another reason in cases of more strongly precipitating cold pools might be continued evaporation of precipitation, which would cool and moisten the air in the cold-pool wake and thus speed up the humidity recovery but slow down the temperature recovery.

The initial increase in humidity at the edge of the front at the BCO might be explained by enhanced surface fluxes due to the strengthening winds , by moisture advection , or by an accumulation of moisture from evaporation of precipitation of the parent convection, which was pushed to the edge of the front . Analyses of the various isotope measurements made during the EUREC4A field campaign might help elucidate the origin of these moisture rings, as suggested by idealized large-eddy simulations . This could also help understand why cloud-resolving models seem to have difficulties in representing the humidity structure in the cold-pool front correctly . As discussed by , the humidity increase just before tmax might be mostly due to the increasing saturation-specific humidity associated with the increasing temperature before tmax (as seen by the relative humidity anomaly in panel d being slightly below zero) and as such likely also related to the way we identify Tmax. A potential dynamical reason for the pre-front humidity increase could be moisture convergence ahead of the front .

The temporal structure of the equivalent potential temperature (Fig. c) is similar to the humidity structure but with a stronger drop across the front and a stronger difference between the weaker and stronger cold pools governed by the temperature drops. The relative humidity signal in the front is mostly governed by the temperature decrease, with RH being 8 % larger at tmin for the strongest cold pools.

The in-front wind speed increase has a maximum in the middle of the front, confirming that the interval between tmax and tmin catches the front characterized by a vortical overturning internal circulation. After the frontal passage, the wind speed decreases slightly below the pre-front level. As the cold pools spread into a strong background easterly flow (mean wind direction at tmax is 86∘, not shown), the wind speed anomalies show that the cold pools on average push forward into the wind until tmax, and backward after tmin. For some cold pools, tmin might thus mark the center of the divergent flow and indicate that the parent convection passed over the BCO. The strengthening winds in the front and the slackening winds in the wake are again significantly more pronounced for the strongest cold pools, with a maximum of 1.5 m s-1 and a minimum smaller than -0.5 m s-1 in the front and wake compared to the value at tmax.

Figure f–g show the composite mean Rfreq and RR measured by the MRR. Both rain variables increase rapidly after the onset of the cold pool, peak towards the middle or end of the front, and start to decrease shortly before tmin. The strongest cold pools have much larger rain rates and rain frequencies during the entire front compared to the weakest cold pools, and the rain frequency of the strongest cold pools also remains strongly elevated until more than an hour after tmin.

The last panel of Fig. h shows the Doppler lidar vertical velocity averaged over four 30 m range gates with mean height of 450 m (w450). The mean w450 peaks at the edge of the front with about 0.25 m s-1 and decreases rapidly to -0.3 m s-1 near the end of the front, reflecting updrafts triggered at the cold-pool gust front and downdrafts driven by the evaporating precipitation inside the front, respectively. The median wmax450 at the gust front edge (see Table ) is at 1 m s-1 much larger than the averaged hourly in-cloud vertical velocities near cloud base measured by the BCO Doppler radar, which has a peak density at 0.2 m s-1 and maxima of 0.6 m s-1 . 1 m s-1 also marks the upper tail of cloud-base averaged updraft vertical velocities at the BCO see Fig. 4b of. This suggests that the gust-front vertical velocity maxima are very relevant for triggering new convection in the trade cumulus regime. The strongest cold pools have significantly stronger downdrafts and also updrafts compared to the weakest cold pools (see also Table ), the latter highlighting the potentially enhanced triggering of new convection by stronger cold pools. For the vertical velocity averaged over the entire sub-cloud layer (wSCL), the picture is similar, but the peak wmaxSCL is slightly smaller for the strongest cold pools and more similar compared to the weaker cold pools (Table ).

As already shown in Table , Fig.  shows that the strongest cold pools are also the driest and the rainiest, and have the strongest wind and vertical velocity anomalies in the front. The relationships and timings discussed are mostly the same when considering all cold pools meeting the noprev criterion (i.e. also including summer periods), just with larger anomalies and the differences mentioned above for the humidity structure. The mean temporal structure for all variables – except for the specific humidity and partly for the wind speed – is also similar to previous observations of tropical deep convective cold pools during the DYNAMO field campaign , just with slightly larger mean across-front temperature and humidity decreases (-1.3 K and -0.6 g kg-1 during DYNAMO compared to -1.15 K and -0.25 g kg-1 at the BCO) and larger mid-front wind speed increases (about 1.5 m s-1 compared to 1 m s-1) during DYNAMO due to the deeper convection. Furthermore, during DYNAMO the increases in specific humidity at the beginning of the front are hardly present, and the wind speed remains elevated by 0.4 m s-1 in the wake of the DYNAMO cold pools , whereas at the BCO the wind speed decreases below the value at tmax in the wake. We hypothesize that this difference can be explained by the stronger cold pools during DYNAMO travelling further away from their parent convection, such that they continue to push forward into the mean wind. What strengthens cold pools in the trades despite the shallower parent convection is the drier cloud layer and free troposphere compared to the deep convective regions, which facilitates evaporation of precipitation and can strengthen downdrafts .

The cloud radars at the BCO also allow study of how the cloud properties change across the cold-pool passage (Fig. ). The mean cloud-top height (CTH) increases rapidly by ∼500 m after the cold-pool onset and peaks at the end of the front. CTH remains elevated by about 300 m compared to the pre-front value in the following hour. The 25 % strongest cold pools are associated with significantly deeper clouds throughout the entire period shown, especially so at the end of the front, when the CTH is on average higher than 3300 m. The cloud-base height (CBH) starts to decrease already slightly before tmax and reaches its minimum near the end of the front at ∼500 m. This decrease is due to the more frequent precipitation with very low echo-base heights and is most pronounced for the strongest cold pools.

The total hydrometeor cover (CC) increases rapidly at the beginning of the cold-pool front, remains about 25 % larger compared to the pre-front value inside the front, and decreases slowly in the wake. The mean CC of the 25 % strongest cold pools reaches nearly 100 % at the end of the front and is significantly larger than the CC of the weakest cold pools during the entire period shown, especially so in the wake. In Fig. d–f the cloud cover is split into contributions from cloud segments with different CBH by considering all 1 min hydrometeor fraction profiles independently. This shows that the enhanced CC of the strongest cold pools in the prior hour is entirely due to cloud segments with CBH above 1 km (CCaloft), whereas the enhanced CC in the front and wake of the strongest cold pools is mostly due to precipitating cloud segments with CBH below 300 m. The rapid increase in CClcl up to its peak at tmax strongly contributes to the CC increase at the edge of the front. This peak is also larger for the strongest cold pools, consistent with their larger w450 at tmax. CClcl and CCaloft are lower at the end of the front for the strongest cold pools, as the lowest CBH is mostly below 300 m and the cloud segments thus count to the CCprcp category. (Note that a given time can only count to one of the three categories.)

Overall, Fig.  indicates that clouds and rain are nearer tmin in the strongest cold pools and nearer tmax for the weakest cold pools. A potential explanation for this observation is that stronger cold pools are running further ahead of their stronger parent convection, while the weaker parent convection of the weaker cold pools might have already dissipated. However, drawing such conclusions from single-point observations is tricky, as the influence of the life-cycle stage and the overall cold-pool strength on the observed cold-pool characteristics cannot be disentangled. Some information about different cloud types populating the cold-pool front or wake can be derived by analysing the CC contributions grouped by the overall CBH of the segmented cloud objects (CBHID; see Appendix A and Sect. ). We find that the peak in CClcl at tmax is mainly due to edges of precipitating clouds that have CBH > 300 m. Assuming that this cloud population represents the clouds evident as mesoscale arcs in satellite imagery, this suggests that the cloudiness at the gust front is mostly characterized by well-developed precipitating clouds. The cloud-type analysis also shows that more than half of the CCaloft in the cold-pool wake is part of large precipitating clouds and is not from detached stratiform layers. This is also suggested by the time–height plots of the composite-mean hydrometeor fraction shown in Fig. g–i. These panels nicely summarize what was discussed in the previous paragraphs and again highlight the differences between the 25 % strongest and weakest cold pools in terms of the cloud response.

Figure a–f also indicate the respective mean CTH, CBH, and CC for all the winter months of the period 2012–2021. They show that cold-pool periods are much cloudier than the average winter conditions at the BCO, with the average in-front CC being twice as large as the 10-year climatological mean. Cold-pool periods also have much deeper clouds than the climatological mean of about 2 km, which is expected as it needs deeper precipitating clouds to form cold pools. The enhanced CC in the wake of cold pools compared to the long-term mean is nevertheless surprising, as convection might be expected to be suppressed in the cold-pool wake. Mesoscale arcs encircling vast decks of deeper cumuli with stratiform layers therefore seem more representative for periods of cold-pool activity than the more classical picture of trade cumulus cold pools as mesoscale arcs enclosing broad clear-sky areas.

Despite the various significant differences between the strongest and weakest cold pools highlighted in the previous paragraphs, there is a lot of variability among individual cold pools. The variability is illustrated in Fig. , which shows the temporal structure of the most important variables for individual cold pools ranked according to their ΔT. Especially the individual differences in humidity and wind in the front and beginning of the wake can by far exceed the mean differences among the strongest and weakest cold pools shown in Fig. . Notable is again the occurrence of pronounced negative values of the wind speed anomaly after tmin, which suggests that some cold pools push backward into the mean wind. Tendencies of more frequent (and intense) rain, deeper clouds, and stronger downdrafts near tmin of the stronger cold pools are nevertheless clearly evident. Especially the downdraft strength seems to be systematically increasing for stronger temperature drops. Besides showing the CTH, Fig. e also gives an indication of the CC, again illustrating how cloudy the cold-pool periods are.

The long time series also allows study of the variability of the cold-pool frequency and characteristics at the daily timescale. Figure  shows the diel variability of cold-pool properties for the noprevWI set. There are clearly fewer cold pools and a lower hourly cold-pool frequency between 16:00 and 22:00 LT compared to the rest of the day. Just before midnight, the cold-pool frequency strongly increases in response to the nighttime increase in cloud cover, cloud depth, and rain rate (see green lines in panels e–i and ). The cold-pool occurrence remains strongly elevated between 02:00 and 15:00 LT, with a peak at 14:00 LT.

Also, most cold-pool diagnostics show a pronounced diel variability. During nighttime between about midnight and 04:00 LT, cold pools are associated with significantly deeper clouds, stronger mean rain rates, stronger downdrafts and updrafts, larger CC, and slightly stronger humidity drops and weaker wind gusts compared to daytime cold pools between about 08:00 and 16:00 LT. There is also a hint of slightly stronger ΔT during nighttime compared to daytime, but neither in the median nor in the 25 % quartiles is this diel cycle significant. It is somewhat surprising that we find no pronounced diel cycle in ΔT, despite the diel cycle of e.g. wminSCL and CTHmax. There is a climatological background diel cycle in temperature of about 1.2 K due to the daytime solar heating of the ground at the BCO (minimum and maximum temperatures near 05:00 and 12:00 LT, respectively, not shown), but this should not affect the cloudy cold-pool periods much and would contribute to lower ΔT in the morning. Other diagnostics like Δqmax and Rint do not show a pronounced diel variability (not shown).

The pronounced diel variability in the cold-pool frequency and most diagnostics is not surprising given the distinct diel cycle in trade cumulus cloudiness discussed in detail in based on realistic high-resolution simulations and both BCO and satellite observations. The diel cycle of trade cumuli is characterized by larger CC and deeper clouds at the end of the night and smaller CC and shallower clouds in the afternoon. This is evident in the background climatological diel cycles indicated in Fig. e–i. The diel cycles of most cold-pool diagnostics have a similar phase and also amplitude to their background diel cycles but are shifted to much larger values (as indicated in the respective legends). For the vertical velocity diagnostics, the amplitude of the diel cycle is also much larger compared to the background climatology.

The period of enhanced cold-pool occurrence between 23:00 and 15:00 LT and its peak at 14:00 LT extends much further into the day compared to the period of enhanced mean background rain rate between about 03:00 and 06:00 LT typical for the trades , suggesting that cold pools help extend the diel cycle of shallow convection into the early afternoon. Also, the diel cycle of cloud cover seems to be slightly extended in cold-pool periods, with CCtot decreasing below the diel mean about 4 h later compared to the climatological CC. The likely reason for the extension of convection into the afternoon is that the cold pools are successful in triggering new convection, which can again induce new cold pools that trigger further convection. Thereby, cold pools reinforce each other, which is supported by the shorter median interval between subsequent cold pools of 121 min between 07:00 and 14:00 LT compared to 182 min between 22:00 and 04:00 LT.

The importance of cold pools in triggering new convection is well established in the literature and can occur through mechanic lifting or moisture accumulations, both of which are favoured when multiple cold pools collide . also found a diel cycle of cold-pool occurrence near Barbados in realistic simulations with the ICON model at 2.5 km grid spacing, however, without the pronounced extension into the afternoon observed at the BCO (see their Fig. 16d). This might be due to gust front vertical velocities being poorly resolved at kilometre-scale resolutions, which leads to too little convective triggering and deficits in rain and convective organization, as simulations over Germany showed .

also find the diel cycle of trade cumuli to be strongly linked to the diel cycle in the occurrence frequency of the mesoscale organization patterns. Whether the cold-pool characteristics and their diel cycles are related to the pattern of mesoscale organization will be discussed in the next section.