The paper examines the Boundary Layer over marine and West African regions. It utilizes various instruments to compute the Planetary Boundary Layer Height (PBLH) using both space-based and ground-based measurements, along with ECMWF outputs. The study discusses the similarities and differences among the various technologies and retrieval algorithms employed. It effectively characterizes the horizontal variability of the PBLH in marine and land regions during September and compare it to ECMWF retrievals. However, questions about the methodology and robustness of some analyses are problematic.



Major comments: 

- Why did you choose to use Relative Humidity (RH) to calculate the PBLH from sonde measurements, instead of using potential temperature profiles? Potential temperature is typically the more common variable utilized to derive the PBLH from radiosondes. RH measurements are often avoided due to their higher uncertainty, making them less reliable compared to potential temperature (see Liu and Liang, 2010).

- Is there any reason why you chose to do these analyses in September? It would also be worthwhile to evaluate other months, especially when the SAL activity typically ramps up between mid-June and mid-August.

- While the slope of a linear regression and the correlation coefficient are related, they are not same. The authors argue that there is a significant correlation among the different retrieved PBLHs. However, they fail to include any correlation coefficients, basing their claims solely on the slope. Additionally, the analyses presented in Sections 3.1 and 3.2 would benefit from the inclusion of scatter plots and correlation coefficients (or the R² value from the linear regression) to better illustrate the agreement between the CALIPSO and ECMWF data. Although the mean values may be similar, the variability from day to day can differ significantly.

- Constraining your analyses to only September limits your investigation's robustness, especially in section 3.3. In the radiosonde measurements, you only have 3 cases, which does not allow you to even make any meaningful conclusion about the relationship between the retrieved PBLH from the radiosonde and CALIPSO.

-  I think the paper would benefit from a richer discussion about the improvements needed in ECMWF in representing the PBLH, especially during the continental region of West Africa, where you observed the highest differences.

Minor comments

- Although the authors mentioned the instruments and techniques used to compute the PBLH, the description of the methodology is missing some information that will allow its reproduction. In particular, the WCT is very sensitive to the dilation factor in the Haar function, but this term is not included in the methodology section, nor are the integration limits. In addition, people typically preprocess lidar data before computing the PBLH, either by interpolating between pressure intervals or horizontally averaging to increase the SNR. I wonder if the authors made anything like this

- L50:  Spaceborne lidar signal not only attenuates as it approaches the surface due to the presence of clouds. The weakened return signals result from longer travel distances from the satellite platform to the earth’s surface, which lead to a lowered SNR.

- Figure 3: The authors mentioned that the PBLH was retrieved using both the WCT and the gradient methods for Lidar measurements. However, in Figure 3, the authors only showed the WCT for the PollyXT and the gradient for CALIPSO.

- In line with the previous comment, you should also include in Figures 5 and 6.

 the retrieved PBLH from the WCT for the CALIPSO data.

- Why did you choose to use Relative Humidity (RH) to calculate the PBLH from sonde measurements, instead of using potential temperature profiles? Potential temperature is typically the more common variable utilized to derive the PBLH from radiosondes. RH measurements are often avoided due to their higher uncertainty, making them less reliable compared to potential temperature (see Liu and Liang, 2010).

Summary:

The paper describes boundary layer (BL) height characteristics over the subtropical northern Atlantic off the coast of Africa, as derived from CALIPSO observations, ECMWF/IFS reanalysis, and ground-based observations comprising of two ground-based lidars and radiosonde observations. Ten-year BL height climatological values over two regions are analyzed and intercompared using CALIPSO observations and ECMWF reanalysis. The ground-based lidars and their respective BL height retrieval algorithms are evaluated and compared against CALIPSO using data from Cabo Verde. Furthermore, two test cases over Cabo Verde are evaluated showcasing distinct interactions between the BL and the Saharan Air Layer (SAL). The first case shows stronger BL inversions and suggests clear separation between SAL and BL, whereas the second case exhibits weaker inversion and shows dust aerosols mixed throughout the BL.

I commend the authors for assembling and performing analysis of several different datasets. The topic is interesting, and the figures are engaging, although the figure fonts should be substantially enlarged. The writing is largely clear and understandable, with only sporadic improvements of style required. The paper shows potential, although in my opinion the paper it falls short on meaningfully investigating the impact of dust on the Atlantic BL. It seems to me rather showcasing a collection of measurements and datasets, with little and inconclusive analysis of their strengths and disadvantages. Here are a few specific complaints:

1. Climatological analysis of collocated CALIPSO and ECMWF (sections 3.1 and 3.2).

The results section starts with analysis of climatological values of BL height in Area 1 and 2. Over Area 1, CALIPSO and ECMWF are in general agreement, with ECMWF being slightly higher than CALIPSO. However, in Area 2, and especially over land, we see very large differences in BL heights. The authors argue, that “CALIPSO in some cases detects the mixing layer height rather than the residual layer and the entrainment zone (Liu et al, 2018)”, an explanation that is vague and unsatisfactory. The large differences between datasets require more in-depth analysis. For example, how many profiles were used in each of the bins in Fig. 6? How often CALIPSO misidentifies BL height in these cases (the error bars on CALIPSO data suggest it is a systematic bias rather than occasional misidentification)? Why two over-land bins agree within the error bars, but six bins do not? For nighttime data CALIPSO is systematically higher than ECMWF. This is very interesting, but it is not mentioned in the manuscript and no explanation is provided.

1. Correlations between CALIPSO and ECMWF, PollyXT Lidar, and Halo Lidar over Cabo Verde (Section 3.3).

In Section 3.3 the authors compare CALIPSO BL height retrievals against ECMWF, PollyXT Lidar, Halo Lidar, and Radiosonde datasets over Cabo Verde. Even though the number of data points is rather small (e.g. 13 for CALIPSO collocations with PollyXT), this still would be an interesting opportunity to evaluate the strengths of different BL height measurement methods. Instead, the analysis part (lines from 261 to 274) is rather short and often seems inaccurate. The slopes of 0.66 and 0.63, in my view, do not indicate good agreement between datasets. There is no Halo lidar data that would suggest overestimation at lower values of BL height (the data fits are simply inconclusive). ECMWF does not retrieve BL height (it uses a parameterization based on vertical profiles of atmospheric parameters). It would be interesting to see how distance from ground observations (PollyXT, Halo, Radiosonde) and CALIPSO affects comparisons (the islands affect BL structure and CALIPSO measurements can be as far as 150 km away).

1. Two case studies over Cabo Verde

The two cases presented in the manuscript are quite interesting and properly illustrate different interactions between the BL and free atmosphere. However, I find the atmospheric profiles and BL height measurements rather inconsistent in these two cases, and the authors do not provide a satisfying analysis and explanation of the datasets. The authors refer to the virtual potential temperature in Figs. 9a and 11a, but it clearly appears to be the regular temperature (it drops systematically with height within the BL). In the first case (dust above BL), the BL height is well characterized by radiosonde observations (Fig. 9a), but all the other datasets indicate lower, sometimes considerably, BL heights (Fig. 9b, c). Halo lidar measurements are almost half the radiosonde value. There is no attempt to reconcile these discrepancies. In the second case (desert dust within BL), the BL height determination is more complicated and the discrepancies between methods could be more justifiable. The radiosonde BL height should be included in both Figs. 9b and 11b.