The authors present multi-spectral fluorescence backscatter profiles measured with a lidar. The study is of high importance to the lidar community as the fluorescence topic is still not sufficiently explored with lidar systems and the aerosol intensive properties measured here will most probably be used in future aerosol classification studies. Therefore, it is important that the values and respective uncertainties are well reported.

I would recommend the publication of this paper after some revisions regarding the following points:

-- Ratios of the fluorescence backscattering are calculated over a broad spectral region (438-614nm) but the authors do not specify how they treat the atmospheric attenuation in the different wavelengths. These spectral effects are not simplified by dividing the two backscatter coefficients and are certainly not negligible. The aerosol extinction cross-section at 614nm is 0.5 times less compared to 438nm for an Angstrom exponent of 2 (typical for biomass burning). Likewise, the molecular extinction cross-section at 614nm is 0.35 times less than the 438nm one. This introduces optical-depth dependent effects that become more and more important as the beam goes deeper in the atmosphere. For instance, the backscatter ratio among 472 and 614nm will be biased by ~4% above an aerosol layer with 0.1 AOD and ~12% above an aerosol layer with 0.3 AOD without including the molecular contribution at all. The authors must include a correction for the atmospheric attenuation in their technique (if not already applied), at least for molecules and for typical Angstrom exponent values. They should also address the error introduced when aerosol with different Angstrom exponent values are present. In Fig. 11 we may probably already see such an effect in the ratios B472/B513 and B472/B560 as the backscatter (and therefore the extinction) increases with humidity.

-- Protective windows for the telescopes are usually deployed in lidar systems. Their spectral reflectance, that also dependends on the incidence angle) must be taken into account when calculating backscatter ratios over such a broad region. Are there protective windows installed in this system? Is their effect measured and subtracted, or at least included in the calibration with the lamp? The authors must address any such potential issues and if present, correct for them or include them in the error calculation.

-- Uncertainties are not sufficiently addressed in the manuscript. There are some error bars in the figures but they are only there for some of the lines. The authors should include them for all. It is also not clear whether these correspond to random noise errors from the signal analysis or from some other source uncertainty as there is no relevant discussion. The authors should comment on how they performed the error estimation.

Minor comments are also included directly inline in the manuscript. The use of English can be in general improved. I would recommend the authors to go through the document again and correct minor grammatical/phrasing issues.

The paper is well written and describes a new approach of an interference-filter-based aerosol fluorescence lidar with five channels. An aerosol fluorescence lidar is rather helpful to clearly identify pollen, wildfire smoke, and other fluorescing aerosol components and to separate them from mineral dust (that does only weakly fluoresce).

I have only minor remarks. The second reviewer provided a good list of comments to the authors so I can be short.

line 71: What is the reason for fluorescence in the case of urban haze? Why is there a difference to wildfire smoke? What about pollen in this context? A few more words would be helpful.

line 76: The Adam et al. (2021, ACPD, Part 2) paper was reject and thus should not be cited. Adam et al., Part 1, is available (published).

line 148: At favorable conditions, aging of smoke particles is completed within 2 days. So, the particles were probably aged. However, smoke from North America is much older (10 days), and then may show different properties.

line 185: With only one wavelength (355 nm) there is no (good) way to categorized smoke based on lidar ratio. 40 or 60 sr was found for smoke as well as for urban haze. Again, please use alternative citation. Adam et al. (2021) is not a good reference.

lines 197-200: Maybe one should mention that the presented aerosol typing is not optimum. Optimum would be dual-wavelength (355, 532nm) depolarization and lidar ratio observations TOGETHER with the fluorescence observations as well as with humidity observations.

lines 289-290: This is a valuable message of the work. Fluorescence observations at wavelengths < 532 nm are sufficient to distinguish fluorescing urban haze from wildfire smoke. That means, three wavelength lidar observations can be combined with fluorescence lidar observations.