We are grateful for the comments made by the reviewer.

We are pleased that the manuscript was generally well received, and we have improved the parts that were not understood and not detailed enough (especially regarding the gas and CO2 aspects), following the reviewer's comments. We also added a new reference (De Moor et al. 2019 GRL) to support a newly addressed point in the conclusion.

The revised version of the manuscript will contain all the new points addressed in the review.

Line 80:

We provide more details in this part. Note that a full description of the drone characteristics is provided in the following paragraphs. Thus, we adapted the structure of this part to make it clearer.

Line 102-109:

Thanks for your comment; we have detailed this part following your suggestions. All equipment described in this study can be transported by airplane without any restrictions. This point has been added in the manuscript at the beginning of the section (for each UAV battery, the energy rating is 131.6 Wh, which does not exceed the 160 Wh limit allowed on commercial flights). For the payload capacity, higher capacities indeed exist. However, we selected this specific drone because our scientific objectives required only a modest payload capacity (under 300 grams). Consequently, larger drones, capable of carrying kilograms of equipment, would have offered no real advantage while introducing unnecessary weight, complexity, and cost. Moreover, the chosen drone’s other characteristics—such as flight endurance, ease of use/transportation, and overall reliability—made it well-suited to our project’s needs. We explain the choice of the drone, based on multiple criteria, in section 2.1.

Line 124-125:

Thanks for noticing that. The drone was piloted manually to adapt to any environmental changes during volcanic activity. This point has been added to the manuscript in section 2.3.

Line 204:

We added some details on this point (we have access to PM, SO2, and CO2), which is already good for gaining insights into the plume intersection.

Line 210-211:

Thanks for your comment. We have added more details about the multigas instrument in the manuscript. It was originally designed to be carried by this UAV and is calibrated every year. As you said, these sensors are exposed to data drifting over time. Thus, the instrument is checked and calibrated by the manufacturer on an annual basis, as recommended (using proper calibration chambers for all sensors). As explained in the method section and by the manufacturer of this instrument (which was originally designed for drone measurements), changes in atmospheric conditions are directly taken into account by the device to mitigate and compensate for the effects of P, T, and RH. We clarified this point in the manuscript. Of course, we have to avoid sudden jumps in T, P, and RH as much as possible to be as precise as possible. Also, the flight shown in this study using the multigas device was vertically fixed (stable attitude), thus with stable P (cf. supplementary file S7 for an exhaustive data description acquired by the multigas). Detection limits and sensitivities are also well described in the manuscript (For CO2: detection limits between 0.01 (100 ppm, which is also the sensitivity) and 5%, with a max deviation of 2%; for SO2: detection limits between 0.000075% (750 ppb, also the sensitivity) and 0.01%, with a maximum measurement deviation of 4%). This has been clarified in the method.

Line 226:

We had a specific concern about this point. In fact, the device automatically synchronizes the data (on-chip proprietary algorithm), taking into account the different response times of the sensors. As an example, we can see in the multigas data that PM and SO2 are perfectly correlated in time (peaks and variations occur at the same time, apart from the few natural exceptions described in the manuscript, where peaks of PM occur independently of the gas concentrations), which confirms that data are synchronized considering the different response time of the different sensors.

Lines 246-249:

Indeed, regulations vary depending on the location, country, etc., and we are careful about that. Field campaigns were performed at Sakurajima and Etna. At Sakurajima, flight regulations allowed flying 500 m above the takeoff point (because of the relative proximity of two airports). At Etna, this limit was increased to 1500 m. Regulations are always checked on the online international database, which is regularly updated on the drone.

Line 257:

Of course, we have to make this approximation, which in the end is not impactful for this kind of measurement, especially for these fine grain size fractions (fine particles with high aspect ratios). This approximation is widely assumed for OPC measurements.

Line 364-378:

We totally agree with your comment. In the presented case, the CO2 sensor is not able to detect variations (which are probably only a few ppm above background) because it is not sensitive enough to detect such small changes in gas concentration (the sensor resolution is 100 ppm). This is a limitation of this device’s CO2 sensor. However, as you also mentioned in your comment, higher CO2 fluxes in some cases could be detectable. We clarified this point in the manuscript, specifically in the discussion section.

Regarding the CO2/SO2 ratio: indeed, this is an important parameter to consider, and we developed that point in the perspectives, as we were not able to measure it properly. We also specified the possibility (and necessity) of upgrading the wide-range CO2 sensor (100 ppm resolution) to a more precise (1 ppm) sensing module that can be installed on the payload instrument. That will be the next upgrade to allow reliable CO2/SO2 measurements on specific targets (we also add the reference of De Moor et al. 2019 to support this perspective). Note that the multigas measurements were also performed at ground level and show similar patterns to the data acquired at airborne levels, though with lower gas and PM concentrations because the volcanic plume was very diluted at these levels.

Line 420:

Aggregates are close to being spherical and do not affect the overall GSD. Note that all particle sizes reported in the analysis are equivalent spherical diameters. We add this point in the method section. Also, as explained in the perspectives, the next step of image analysis is to count all the particles inside and outside the aggregates to have a real and total GSD, with the information on the number of particles inside and outside the aggregates. This analysis is currently being carried out for another study focused on the ash aggregation mechanism.

Line 493-497:

Thanks for your suggestion. We have added this point in the perspectives.