the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 13 Aug 2026
Assessing evaporative cooling from seawater spraying in the marine boundary layer
Abstract. Increasing frequency and intensity of extreme heat events motivate improved understanding of processes that may modify near-surface thermal conditions. This study investigates seawater spraying as a method for direct atmospheric cooling. Using large-eddy simulations (LES), we quantify plume dynamics, cooling response, and impacts on heat stress across a range of atmospheric conditions. Simulated spray rates of 50 L s-1 produce near-surface temperature reductions of approximately 0.5–1 °C within the first few kilometers. Although theoretical analysis suggests that evaporative cooling potential can exceed -20 °C under hot and dry conditions, atmospheric mixing and advection substantially reduce this potential under realistic conditions. Cooling is strongest near the source due to evaporatively driven downdrafts, but decays rapidly downstream. We compare deployment from wind turbines with ship-based platforms. A key finding is that injection height governs plume structure and cooling response. Ship-based spraying produces strong, localized near-surface cooling, whereas wind turbine deployment redistributes cooling more effectively within the lower boundary layer. The turbine wake modifies local mixing, but vertical redistribution is mainly governed by evaporatively driven buoyancy perturbations. Extending the analysis to human thermal comfort reveals that temperature reductions alone do not determine effectiveness. Changes in Heat Index are non-linear and depend on background humidity, with benefits near the source that may diminish or even reverse downstream. Overall, seawater spraying can provide measurable cooling, but its performance is highly sensitive to background atmospheric conditions.
- Preprint
(20263 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-4092', Anonymous Referee #1, 17 Sep 2026
-
RC2: 'Comment on egusphere-2026-4092', Michael Diamond, 27 Sep 2026
In this manuscript, the authors explore the possibility of cooling the atmosphere locally with seawater injection and subsequent evaporation. They first provide a theoretical parcel treatment and then perform LES experiments with wind turbine and ship-like sprayers. I have some concerns about the realism/representativeness of some of the profiles that can be addressed by either showing example soundings of where those conditions are relevant or better framing the results as highly idealized. The presentation of the figures also can be improved. The manuscript will be publishable if suitably revised.
Major comments:
A) Realism of case profiles: The high-temperature cases are quite extreme, although conditions approaching these values may occur in some coastal Middle Eastern and South Asian environments. More importantly, it is not clear whether the complete combinations of temperature, humidity, boundary-layer depth, and inversion strength used here are representative of observed atmospheric profiles. Because the magnitude of evaporative cooling depends strongly on the background thermodynamic state, it would be helpful to provide example observed soundings or reanalysis profiles demonstrating environments in which these cases are relevant. Alternatively, the manuscript should more explicitly frame these cases as idealized sensitivity experiments rather than representative atmospheric states.
B) Figure span and colorbar issues: Figures 6, 8, 9, 10 are difficult to interpret due to the choice of color scaling and the truncation at 3 km despite having a 6 km domain (with the 3-6 km results only shown in Figure 7). Figure C1 helps a bit on the former. For Figures 6, 8, and C1, the sharp cutoff at -0.01 or 0 degC seems to highlight negative values of “noise” while omitting positive values of the same magnitude. It also gives the impression of a sharp cutoff in the plume and I cannot tell if this is real or an artifact. Figure 9 artificially inflates the positive values for effect, but it produces the misleading impression that the positive overshoot is as strong as the initial cooling. Readers would be better served by a simpler colorbar, like the diverging RdBu_r colorbar from Figure 9 but linearly varying from +5 to -5 degC or something along those lines.
C) Seemingly missing turbulence/height results: Results from the Spray and Turbine cases are never shown, despite Discussion section 5.2 seemingly being based on them.
Specific comments:
- Figure 3 and related discussion of “20 K cooling” (e.g., Lines 6, 360): Adding 10 g kg-1 to an air parcel would require extremely large spray rates. I agree this is a physically plausible upper bound, but I would de-emphasize the discussion of this extreme corner as it really is not practically relevant.
- Lines 125-126: The number of vertical levels and the vertical domain height/resolution are redundant. It would be clearer to just specify the domain size and horizontal and vertical resolutions.
- Line 232: 6 km domain?
- Line 240: +2 to -2 degC does not seem “neutral”… +2 is actually twice the maximum value of your colorbar.
- Line 252: “Strong” dependence on height is a bit overstated… I would argue the responses are more similar than they are different between Turbine Spray and Ship Spray.
- Section 5.4: It might be worth noting global constraints on evaporation and precipitation and the atmospheric energy budget here, as these mean this type of spraying can only work as a localized intervention.
- Lines 337-339: Figure 8c does not seem to support the idea that turbine spray would be much better than ship spray for MCB or MSB.
- Lines 383-383: I’m not sure what this means. How do the results show that direct evaporative cooling “belongs” to some class of theoretical techniques?
- Lines 390-391: I don’t believe this is compliant with ACP’s data policies. At the minimum, outputs to recreate the analyses in the paper should be provided in a public repository (e.g., Zenodo). It seems like this might already be included in the author’s Zenodo repository with the input files, in which case the statement here just needs to be clarified.
Citation: https://doi.org/10.5194/egusphere-2026-4092-RC2
Data sets
Dataset Supporting Assessing evaporative cooling from seawater spraying in the marine boundary layer, ACP 2026 Annelot Broerze, Stephan R. De Roode, Herman Russchenberg https://doi.org/10.5281/zenodo.21239333
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 135 | 72 | 35 | 242 | 30 | 32 |
- HTML: 135
- PDF: 72
- XML: 35
- Total: 242
- BibTeX: 30
- EndNote: 32
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
General
This manuscript covers an interesting subject, potentially of interest for the readership of ACP. The study into cooling the marine atmospheric boundary layer by means of sea spray establishes a link with thermal comfort, which seems to provide the main motivation of the study. The effect on the heat index is significant near the sea spray sources, but diminishes quickly with distance from the source. It shows that near-realistic first-order estimates are possible by using simplified concepts.
As expected, the strongest effects are found near the sources, and the effect quickly diminishes with distance from the source. However, there will be probably nobody or hardly anybody near the source, at places with the potentially largest benefit from the cooling in terms of human thermal comfort. Even in this study for idealized conditions, which provides an estimate of maximum effects at for large quantities of injected seawater, cooling will be hardly felt at some distance, in places where it is most needed. The study considers processes in the marine boundary layer, but – again in terms of human thermal comfort – we need cooling mainly over land areas. These are the main reasons we consider this a nice, but purely academic study, with little to no impact in practice. In fact, it supports the idea that significant cooling of cities requires a comprehensive urban design plan.
The discussion is disappointing in this regard. So, why not present it as an academic study, driven by interest in boundary-layer processes, and stick to a thorough discussion on physical aspects? If the authors insist on the link with thermal comfort, the discussion on the use in practice (development towards such a use) should be improved.
The manuscript is -in general- well written, but some additional explanation / small changes will further improve the text.
Main comments
Thermal comfort indices
The authors choose HI to diagnose effects on thermal comfort (TC). Whereas this index nicely fits the study in illustrating the trade-off between cooling and humidification, it is not a good TC indicator. The reason that this index is often used is because it is relatively simple, which facilitates communication to the general public. Since TC is presented as the link to practical applications on the long term, the limitations of this index should be more thoroughly discussed. For example, it is well-known that (solar) radiation and ventilation (wind) are strong drivers of TC. The authors seem to be aware of this, but the discussion is a bit disappointing in this regard.
The spraying events would have to be organized at some distance from the coast. Given the quickly declining cooling effect, it is questionable if such a “measurable cooling” would ever be really felt by humans, in places where it is actually needed most: urban areas on land. Over land, turbulence is enhanced (during daytime at least), which will further diminish the expected impact. Yes, there are other aspects of cooling that are not yet covered by standard TC indices. However, are these really relevant in the context of the present study? E.g., direct (evaporative) cooling of the skin by spray droplets is ignored. But how many droplets would reach actually reach human targets, again given the fact that the cooling diminishes quickly with distance. And who wants to be exposed to them and be covered with (polluted) sea salt?
If the study remains to be presented as a study of TC, additional aspects of TC should be considered in more detail, in the framework of more widely accepted indicators (PET, UTCI, WBGT). E.g., when taking into account radiation, there will be some kind of a “baseline stressor”. Can anything be said about the contribution of the computed cooling and moistening in that context? How might effects of wind (including sea-breezes) affect conclusions regarding TC? What can be said about changes in the wet-bulb temperature?
Whereas the authors conclude that there is a potential for cooling (in terms of TC), this conclusion cannot be inferred from this study. The optimistic conclusion relates to ignoring the fact that the reported (TC) effects will be small (hardly felt) in places where they are actually needed, and possibly unwanted consequences when using the method.
Please, avoid using / referring to “Hittekracht”. Apart from the fact that this is a Dutch word, it is not suitable for using it in an international context because this index has specifically been designed to be used in the present Dutch context (i.e., climate), for communication with the general public. Please, refer to the well-known WBGT instead, which has been used worldwide for a long time, even for legislative purposes.
Other comments
L12: reveals -> confirms (even “confirms” might be too strong, because it has since long been known that temperature alone does not determine thermal comfort).
L13: may diminish -> quickly diminishes
L24: In Europe as well, extreme summer heat is becoming increasingly hazardous -> Extreme heat is also becoming increasingly hazardous in Europe
L44: perceived thermal stress: this is a weird formulation. Use “thermal comfort” or “thermal stress” [or “perceived temperature”] instead.
L46-47: Statement is doubtful. This index does, however, provide a convenient diagnostic in the present study (the fact that it is used by national weather services does not guarantee it is a good TC indicator, since relatively simple indices like HI may have been selected mainly for the purpose of communication).
L50: depends on plume evolution in a turbulent atmosphere-> depends on turbulent transport in the [atmosphere][atmospheric boundary layer].
L73: RHatm is the only variable that is not italicized of the parameters presented in this section
Figure 2: The upper panel shows the temperature decrease toward Tmix -> temperature (Tmix) decreases towards Tmin, as schematically illustrated in the upper panel.
L90: some discussion on the effects of salt on the thermodynamic principles used here would also be appropriate.
Figure 4: Units for Heat Index are missing in the small text boxes
L109 – 110: at high temperatures, humidity can increase the Heat Index by up to 9°C. -> at high temperature (here 37°C), an increase of the relative humidity from [45?]% to [65?]% increases the Heat Index by 9°C.
L145: please, provide ranges as numbers.
Fig. 6: please, specify “near-surface” [height].
Figs.6,8,9: please, discuss the lateral effects in front of, and just behind the sources, outside the main plume. Is this an artifact introduced by the periodic boundary conditions? If so, have the consequences been assessed?
L133 and 161: Spin-up time is first referred to in seconds (7200s) and then in hours (2hr). Is this referring to the same spin-up time? Either way, the units should be kept consistent.
Figs. 6 and 7: Clarify if the statement “The in-plume region is defined as the coldest 20 % of ∆T” Influences the plotting of the figure. Does this line up with the -0.01˚C on the color scale?
L 174-175: statement is difficult to check / infer from figures. Redo color scaling?
Fig. 9: color scaling of left-side panels may exaggerate the humidification effect, relative to the cooling effect, because of asymmetry between positive and negative values.
L225-226: immediately after release -> immediately after moisture injection
L235 – 236: Unclear analysis of Figure 9. The gradient with distance seems sharper, and approaches 0 quicker for the turbine than for the ship in figure 9. What is the definition of “recovery” here?
L238-239: yes, there may be this shift, but is it really relevant?
L281-282: “As a result, … ,evaporative cooling”: could this be a consequence of the chosen simplified setup to describe effects of droplet evaporation?
L294-295: “A key outcome…proportional.” This is not a surprise, but inherent to the definition of HI. Hence calling this a “key finding” is an exaggeration.
L296: “…Heat Index response is strongly modulated by humidity and may recover…” : can a response recover? Or does HI recover?
Section 5.3: given the focus on TC, the discussion in this section is somewhat disappointing. More discussion on TC effects that are truly felt by humans, when the spraying systems like investigated here would be used in practice, would be expected (along with a related summary in Section 6). Also see General remarks and Specific Comments indices.
Section 5.4: how likely is an impact of limiting factor [in practice]? Isn’t the realized (=computed with DALES) cooling disappointing, since it is much smaller than the “theoretical one”, even in this idealized study? Especially in terms of TC?
Section 5.5: it seems that (most of) these comparisons are outside the scope of the study. So, how relevant are they?
L360: It is unclear if the -20°C has been derived from the research of the authors or from former research.
L365: large eddy simulations -> All other places refer to large-eddy simulations
L378 – 379: This is the first time TC metrics are mentioned in this way, along with this reference. This could be introduced earlier, perhaps under the priorly suggested discussion on these metrics.
L385: This potential is not evident from this study.