General comments

This article presents a study of the atmospheric boundary layer over a heterogeneous surface. To do this, the first part of the article uses observational data from a field campaign, and the second part runs and explores a LES on the area of the same campaign. Overall, the article is well written. The article gives interesting results on the ABL dynamics over a heterogeneous surface and in particular on the blending height of atmospheric state variables. However, some shortcomings in the methodology affect the quality of the article. In particular, the analysis of a composite day created by averaging 3 radiosoundings seems questionable. In my opinion, it prevents the authors from fully understanding the meteorological situation they are trying to study and makes the comparison with observations more difficult. Some results (such as the presence of an IBL) are not based on objective quantification but on subjective judgement. The article is relatively long, with weak links between section 2 (observations) and section 4 (LES), and the structure is sometimes confusing. In addition, some results could be discussed in a more balanced way, with more discussion of the limitations of this work. I think Appendix A is important as it shows that the LES does not capture the observed ABL dynamics, which is probably the reason why Sections 2 and 4 are not so well connected.

Specific comments

• l. 4 -6: This sentence is a bit unclear/ hard to understand at first – reformulate? “strength” → “magnitude”?

• l. 8: Add that the sensible heat flux drives the buoyancy over the non-irrigated field, in opposition to the first part of the sentence.

• l. 9: “mean scalars” → what are the scalars discussed? Potential temperature and vapour mixing ratio?

• l. 11: “three-dimensionnality” → Isn't an LES 3-dimensional by default? Needed? Moreover the spectral analyses presented later are performed at given heights, with 2D fields.

• l. 13: “adjustment of ABL characteristics” → unclear

• l. 42 – 44: How does this depend on the intensity of the heterogeneity? If the sensible heat flux difference between the heterogeneous patches is relatively low, is the result still valid? What happens to secondary circulations when the scale of heterogeneity is larger than the ABL depth? I would still expect a secondary circulation to form. An idea of article discussing that is Segal & Arritt, 1992, BAMS (doi.org/10.1175/1520-0477(1992)073<1593:NMCCBS>2.0.CO;2)

• l. 103-104: Not clear. Is the cool and moist easterly wind the sea breeze?

• l. 132: The citation of Brilouet 2021 does not correspond to LIAISE data. Please refer to the right data paper.

• l. 136-140: This paragraph is key to the methodology and needs more detail in my opinion. How is the golden day calculated? Is it just an average of the 3 days? How are the wind fields generated? And after averaging potential temperature, specific humidity (and 3D winds?) separately, do the newly created vertical profiles still make sense physically? Is there a risk of having inconsistencies for some time step (e.g. strong negative vertical gradient of virtual potential temperature, divergence of horizontal wind speed and positive vertical speed w?)

  Has this approach been used in other LES studies?

  Sounds like it makes this study an idealised study? Does it?

  Why didn't the authors use a single realistic day (which would have made it easier to compare the LES results with the observations)?

• figure 2: What are the shaded areas? Standard deviations calculated over 3 days? In relation to the above comment, it would be better to plot the three radiosoundings used directly (e.g. with finer lines). Also, adding the wind speed and direction to the vertical profile would be very helpful in understanding the situation.

• l. 147: Below 500m there appears to be a well mixed boundary layer, but above 500m it appears to be quite stable. The mixing ratio does not seem to be well mixed either. Since you mention the fact that the atmosphere is stable at 10 UTC, this just looks like a classic boundary layer evolution in a stable atmosphere. This should be discussed in the text and perhaps related to the TKE profile of Figure 3 (see also the related comment).

• l. 151: “presence of a secondary circulation” → would be interesting to have it confirmed with vertical profiles of wind, as mentionned above

• l. 155: “unstable layers near the surface in the dry area” → Yes, it seems to be unstable near the surface, but it is not clear up to where it can be considered unstable. The mean profiles of potential temperature at 12 and 14 UTC show slightly stable boundary layers between 500m and 1500m, why that? Is it an effect of the three days averaging?

• l.155: “IBL in the wet area” → It is not clear why these are internal boundary layers and not just the classical ABL. This IBL could be an artefact of the averaging process. Further investigation of the characteristics of the atmosphere between 500 and 1500m would be interesting to clearly define IBL vs ABL here.

• Figure 3: add in the legend what are the error bars.

• l.162-163: Give the ABL heights found with this method for each time stamp of fig.2, maybe in a separate table or in appendix.

• l. 172: “there is a weak moisture flux near the surface” → rather seems to be virtually null

• l. 176: “strong TKE” → I would not consider a value of 0.5 m2.s-2 a “strong” TKE

• l.176 – 179: How do you explain having values of TKE of about 0.5 – 1 m2 s-2 in the free troposphere? The TKE values shown in Figure 3 give the impression that TKE does not vary with height. It would be interesting to show in appendix similar profiles but for the morning.

• l. 180: Why “conversely”? The text was already discussing the wet landscape before.

• l.196: Why not using TKE here, as for Fig.3? Comparison and understanding would be made easier.

• l.199: Why is the flight data is not average on 20 to 22 July like for fig. 2 and 3?

• Figure 4: y-axis is uTKE and but mean values on the top left corner are TKE… I think using only TKE here would be better. Also add the altitude a.g.l. or a.s.l. of this flight.

• l.215: “superadiabatic lapse rates in the lowest hundreds of meters” → Fig. 2 shows superadiabatic lapse rates closer to the surface in my opinion

• l.220: “evident” → do not agree, should be discussed more in a more balanced way, to relate with previous comments

• l.253: Add a few sentences or a paragraph to resume the results of the comparison between the LES and the LIAISE data (cf also comment on Appendix A below). Specifically you should mention the shortcomings of the LES outputs and justify why you don’t observations anymore in this section.

• l. 259: The most energetic eddies of the turbulence are smaller in the surface layer and in the entrainment zone. A vertical resolution of 25m means that the eddies close to the surface are not explicit in the model, and therefore the LES is not an LES in the near-surface troposphere. In the same way, eddies in the entrainment zone are smaller than in the well-mixed part, and your model may not be a LES in the entrainment zone. Do you have an idea at which heights your model can be considered an LES? As the IBL above the alfalfa local scale is rather low, can the model be considered an LES in this part of the atmosphere? Please quantify how much of the TKE is explicitly resolved in your LES at different height. This should be done ideally for the wet and dry landscape.

• l. 259: What do you call equilibrium with ABL in the context of a heterogeneous terrain? Do you have a reference for justifying it is large enough?

• l. ~264: Do you consider orography or a flat terrain? In the latter case, a map of orography could be added in Fig. 1. Also if orography is considered, mention the limitations inherent to the fact of using a single radiosounding to represent all the domain in such a complex terrain.

• l. 264: refer to comment on l. 136-140, and recall the limitations of such an approach based on an average day.

• l. 272: If I understand well, the same advection is considered at all heights of the domain, based on the advection measured near the surface? Please discuss the limitations of such an approach.

• l. 278: How the roughness lengths are determined in Mangan et al 2023a? I didn’t find much information on that in the article.

• l. 291: The choice of the parcel method should be more discussed, with the limitations and advantage of this method (e.g. doi.org/10.1016/S1352-2310(99)00349-0). I guess this method is particularly relevant for assessing the possible ABL top from a single field, and therefore to study the local scale here. However the actual modelled ABL top is also influenced by larger scales as you show later, and therefore other methods for the ABL heights could be used to make the most of the LES, specifically for the landscape and regional scales (e.g. gradient of TKE, of potential temperature, etc).

• l. 297: You consider the ABL height of the wet and dry landscape as the mean of the local scales included in each landscape if I understand well. Could you measure the height of the ABL with a different method more adapted to the landscape scale and check if you find the same value approximately?

• l. 300: This value of specific humidity is very low, are you sure the unit is the right one?

• Figure 7: What is the shade around the lines representing local scales (alfalfa and fallow)? If it is a standard deviation how do you calculate it? Also add to the legend that the standard deviation correspond to the shade.

• L. 290-300: each time you refer to Fig. 7 you actually refer to Fig. 7 (a). Please be more precise in the referencing of the figures by adding the letter when appropriate.

• l. 301: You didn’t comment the Figure 7 (b). Remove it or discuss it. Specifically it is interesting that you find EF > 1 for three scales including the largest one. It seems rather early for Spain.

• Figure 8: Why is fallow scale is not shown for 08:00 UTC?

• l. 305: “and z/zi ∼ 0.5 at all times of the day” I don’t understand this part.

• l. 309: “the increase OF heat flux”?

• l. 310: z/zi ∼ 0.2

• l. 310: What are entrainment fluxes? Fluxes in the entranment zone? Then it is even way above 0.5.

• l. 317: High moisture fluxes over the dry landscape are also found below the entrainment zone, so the entrainment of free troposphere dry air may not be able to explain such high moisture fluxes… Also it is not only the strengh of the entrainment that determine the moisture flux but also the gradient of humidity between the ABL and the free troposphere. And in the case of the dry landscape I would expect a relatively weak gradient in comparison to the wet lanscape. Could it be the signal of a secondary circulation that you develop afterward?

• l. 326: Fig. 5?

• l. 338: Can you please develop the physical meaning of your characteristic length scale?

• Figure 9: At what time are the spectra calculated?

• Figure 9 a, b, c, d: Add to the legend what is the gray dashed line. For panel d, the spectrum of the vertical velocity is expected to follow a law in k-2/3 of the inertial subrange of the turbulence. Can you please represent such a law on this panel (e.g. https://doi.org/10.1007/BF00122327) and discuss it to assess the behaviour of the turbulence in your LES?

• Figure 9 e: change y-scale to better see the low values of the integral length scale.

• l. 340: “1111m at 12 UTC” for which area? The regional domain? If it computed over the regional domain, then the zi considered is the same everywhere, and you could add that it is at 555m agl. It would make more sense to relate it to the landscape scale zi, and in this case you should give the zi values for each landscape.

• l. 341: “the gray DASHED (?) line” If you indeed refer to the gray dashed line, please add it to the legend of figure 9.

• l. 344: It is indeed likely that the microscale peak is linked with the microscale of H and LE, but this very likely due to the parcel method that uses the near-surface temperature to compute the ABL height. I think it just shows that the parcel method gives ABL height that strongly depend on the field scales. If you do the spectra of ABL height calculated with a different method not so dependent on the surface, would this peak still appear?

  Then, the upturn for the high wavenumbers of a LES can also be due to numerical filtering issues (for ex. cf fig 10 of doi.org/10.1175/MWR2830.1), how to be sure this microscale peak is not due to that?

• l. 349: “IS” more classical? What is the classical spectra of spatial surface fluxes? It does not necessarily follow a law in k-2/3

• l. 358: “lessons the length scale”?

• l. 359-360: Why the length scale of vertical velocity does not decrease abruptly in response to the sea-breeze like for the length scale of zi?

• Figure 10 a,b,c: Place the alfalfa and fallow field on the x-axis. Add in the legend that the transect corresponds to the box of Fig 6 c. The distance 0 does correspond to the border between wet and dry landscape from figure 6, why is that? What is the white line in panels a, b, c? More generally for figures, better describe all features of the figure (colors, lines, etc) in the legend, and no need to repeat it in the text then.

• Figure 10: in legend: “dashed lines” → “dotted lines”

• Figure 10 d,e,f: The fluxes in the entrainment zone are not readable. Put it aside in a different panel or remove it.

• l. 365: Fig 6a → Fig 6c

• l. 366: “wind direction is predominantly from the west” This wind direction should have been described earlier when discussing the weather situation on the golden days (l.136-140). More specifically one could expect from a thermal low situation on the iberian Peninsula to have a eastern wind in the Ebro basin. Why is the wind westerly here?

• l. 367 to 371: You describe the features of the figure here. This should be done in the legend, and the text should be used to interpret what is shown in the figure.

• l. 372: “increase in zi”: How do you see that? Is the white line the ABL top? If so, I would expect from Fig 6c that the ABL goes from about 500m to 1500m.

• l. 376: It's tricky to compare the measurements with the LES in view of the comparison results shown in Appendix A... It should at least be pointed out that the LES results are not directly comparable with the measurements.

• l. ~371: It is not said how TKE from the LES is calculated. From the figure 10, we can have the feeling that the small whirl at distance=800m and z=700m should be considered as turbulence more than mean flow. More detail should be given on the separation between mean flow and TKE in the LES. Also do the values of TKE include the sub-grid TKE or not?

• l. 382: In the surface layer, with your resolution, the model is very likely not an LES… How is your TKE computed here? Is it mainly a sub-grid TKE here?

• l. 383: Similarly in the entrainment zone, the model is likely not an LES either

• l. 385: Clarify the nature of the counter gradient and what it means.

• l. 389: What is the reason for this small circulation? Or could it be a large eddy of the CBL? In the latter case, it could be considered as one of the large eddy of the turbulence more than as a secondary circulation. More details must be given on what is considered turbulence and what is mean flow in Figure 10.

• l. 390: “One a stronger secondary circulation” I don’t understand this sentence.

• l. 392: Interpreting something on IBL from the absence of IBL signal in the figure 10 seems problematic…

• l. 410: Specify the unit of the blending height and meaning of it different values. As I understand the unit is [1] and a values of 1 means that the blending height is only found at the ABL top. Is that right?

• l. 411: C_{v,b} is the field averaged at the surface then?

• l. 421: “we analyze at the heights in which the local” to reformulate

• l. 424: “In Fig. 11, we show…” the general description of figure 11 comes after the specific description of each panel. To reverse.

• Figure 11: Change style for markers between variables. When dots overly, it is not possible to see the one in the background. Also change increase y-scale for panels c and d, points around zi are hidden.

• l. 449-450: “This is because entrainment introduces air masses of different properties into the ABL” This sentence is not clear. The entrained air is supposedly the same in the wet or dry landscape. As I understand it it is more the entrainment rate which modify more or less the ABL depending on the landscape scale.

• l. 458: I don’t really agree on the fact that it is a realistic case. It mixes weather situations of three days without proving that it correspond to a case that happened in reality. Moreover this methodology makes it hard to compare the LES results to the observations. The only part discussing this is put in appendix, and the results shown in appendix are not convincing.

• l. 462: What is the “land-surface scale”? Do you mean “land-surface” only?

• l. 464: The IBL is not that clear. More generally the authors mention the existence of an IBL several times, but this IBL is subjective, it is not clear from the figures. The existence of this IBL and its stability should be better substantiated, by objective means, for the averaged day as for the individuals 3 days.

• l. 467: “evident”: avoid this type of subjective judgement. Moreover I would expect negative surface heat fluxes for a stable layer… this should be discussed in the previous 4.2.1 section. Do not add interpretation of the results in the Discussion section.

• l. 469: “Do not cross defining the landscape scale” Unclear, reformulate.

• l. 478: You mention a secondary circulation for the dry landscape when you discuss Figure 10. Must be clarified.

• l. 486 “greater than 2.5 ms−1 reduces the formation of secondary circulations if is not normal to the boundary” This should be discussed in the literature review section. In the introduction you mention that 5ms−1 of background wind is needed to prevent secondary circulations. How this threshold of 2.5ms-1 relates to the the other threshold of 5-7 ms-1?

• l. 506: “to represent the LIAISE.” The LIAISE campaign? To reformulate.

• l. 507: You use a realistic averaged surface flux map, but since atmosphere and surface are not coupled and the turbulence in the surface layer is not explicitly resolved, I don’t think you can say that you “explicitly resolve the surface in the LES”.

• l. 520: Didn’t you say the opposite at line 474 (and in the corresponding previous section)?

• l. 527-528: missing word?

• l. 535: The stable surface layer is not that clear either from observation or model. To discuss in the previous corresponding section. Characterize objectively the stability of this surface layer (e.g. positive gradient of potential temperature, Richarson number, or other)

• l. 541: “The regional ABL characteristics fall between the extremes of the two landscape scales” When you discuss it (Fig 7 and 8) it feels like you just average the features (profiles, ABL top) and don’t use independent measures. This sounds trivial if you just average the two landscape scales.

• l. 550: Concerning this second unblended layer due to entrainment, it is due to the fact that ABL is never blended for vertical velocity in the ABL (to deplace)

• l. 554: “bi-directional land-atmosphere interactions” The word bi-directionnal gives the feeling you studied this bi-directionnal interaction, whereas it is uni-directionnal in your case as I understand it.

• l. 557: “a maximum background wind of 2 ms−1” if the maximum wind is so low it should allow for a secondary circulation to develop. I think it is more an issue of wind direction. The circulation expected would go from wet to dry landscape, and it is already the direction of the background wind, so the circulation may just be included in the mean westerly wind. For the return flow of the circulation, it is indeed not visible, but it is known that this features is not systematically observed. Even though I am not an expert, I think these resources may give some clues on that: doi.org/10.1175/1520-0450(1994)033<1323:TIOLSW>2.0.CO;2 and doi.org/10.1007/s10546-010-9517-9

• l. 559: “combining observations with an LES case study” The article gives the feeling the observations and the LES are treated very separately, and indeed it seems it is needed since the LES does not capture all the observed ABL dynamics (Appendix A).

• l. 559: “increased buoyancy flux from the dry landscape influences the state variables” can you recall how?

• Appendix A: the cool IBL observed in the 500m agl is not reproduced by the LES, even at the alfalfa local scale. Why is that? More generally, I miss a critical description and interpretation of the LES results. Or maybe stress that the LES results should be considered as an idealized atmospheric setup but with a realistic surface flux map, and therefore that the comparison to the observation is not useful.

  There is very few comparison between the LES and the observations along the article. I guess the reason for this lies in this appendix. Maybe this figure deserves its place in the body of the article?

• l. 573: Please develop the surface representation issue and the assumptions you mention.

• Figure A1: What does the shading correspond to? Would it be possible to increase the size of the figure to be able to see the details? By rotating it if necessary.

• Appendix B: According to the shaded areas, there is very few area where the atmosphere is blended, even above the ABL. This result is surprising and may call into question your methodology for the blending height calculation. Explain why you find such behaviour above the ABL.

Technical corrections

• l.10: “a” stable …

• l. 23: large eddy simulation → add acronym here for the first mention apart from abstract

• l. 108: “~10 km”

• l. 170: “This is consistent”

• l. 179: unit issue, variances are in m2 s-2

• l. 219: “At midday”

• l. 236: “to complement”

• l. 290: (Fig 7)

• l. 300: Khr1 → K h-1 (add a space between K and h and exponent should be negative)

• l. 352: “in” repeated twice

• l. 376: Two different tense used in a same sentence.

• l. 393: add a space before your subsection “Blending height”

• l. 415: repetition of where

• Equation 4: put the a of Da in subscript

• l. 457: too much “in”

• l. 503: “implications ON how…”

• l. 525: “placed ON how”

• l. 533: “Internal boundary layers” → IBL

Comments are attached.

Mangan et al. present an interesting study on atmospheric boundary layer dynamics over a heterogeneous landscape including irrigated and fallow/dry landscape units. Their main interest lies in the interactions of spatial scales and how they jointly affect atmospheric boundary layer development. In their study, they combine detailed field observations from the LIAISE campaign and Large Eddy Simulations (LES) to better understand these interactions. They use a composite day that is based on measurements during three days in July. Their approach is innovative and the detailed characterisation of boundary layer dynamics over a thermally heterogeneous landscape is unique. Overall, the manuscript is well written, and the presentation of results is clear. I think the discussion would be strengthened if the authors could discuss some of the limitations of the experimental setup. They use a composite day in the summer for their study, which allows them to discuss in detail different processes. However, how representative is this setting for other seasons/conditions? It could be discussed what would be needed to extend observations or modelling to larger timescales to investigate how representative the study conditions are.

I do not have any major comments. However, a few sections could benefit from clarification:

Line 2: A geographical location would be useful here.

Line 10: Do these results apply to afternoon conditions? If so, please specify.

Figure 2: Could atmospheric boundary layer heights be added to this figure? I was also wondering if the layering could be due to the existence of a residual layer from the day before. In general, it would be good to support the interpretation of "well-mixed" with quantitative metrics. How was the mixing state determined? Did the authors apply thresholds for lapse rates?

Figure 11: The presentation of the blending height is insightful. However, I was wondering if the authors could calculate uncertainties for the blending heights. There is quite some variability, and I was not sure how much this variability is due to changing uncertainties in the estimates. Why did the authors decide to show blending heights for both scalar fluxes and concentrations? Why would the heights differ?