Doppler Aerosol WiNd (DAWN) Lidar during CPEX 2017: Instrument Performance and Data Utility

Round 1

Reviewer 1 Report

This is a worthwhile addition to the literature on airborne Doppler lidars and should be published after minor corrections and some tightening of the text.

The DAWN lidar is considered as a reliable instrument, with its functions described at a top level, and with the emphasis on data products and intercomparisons of interest to the meteorological community (lines 45-56, 565-568). The paper is not designed to report recent laboratory research, and it is not a study of lidar physics in fine technical detail. The authors have intentionally summarised or omitted the details of large literatures e.g. on lidar modelling and on the “information peaks” in spectral estimates formed from the detector output.  

I have only these brief remarks and a few suggestions:

 

Line 41: “but, already, have” and then “complementary” (not “complimentary”).

Line 73: “high-pulse-energy”?

Table 2: “near-Gaussian”?

Line 130: perhaps add “i.e. from times when no reflected signal is being received”. Similarly, Witschas et al. (JTECH-D, June 2017) say “the signal after the ground return can certainly be considered to be just noise”.

Line 135: areas “of the order of square kilometres” instead of “defined in kilometers”?

Line 187: “referred to as”?

Line 188: “500 MHz”

Line 225: “at the beginning of”. Also “the objective” would be better as “one objective”; there are other objectives of processing, and line 428 / lines 584-587 mention some reserved for publication elsewhere.

Line 236: is “provides a solution to the single answer to the question” needed instead of “provides a single answer…”? Or perhaps “…ASIA manages the amount of sample integration to ensure a “good” estimate of the LOS wind component, and in so doing it defines the “spatial resolution” of the estimates…”

Lines 238-246: please check this and rewrite if necessary. “Parameters” 4 and 5 (“Define…Select…”) leave me unsure what are supplied to ASIA as fixed values and what can be varied / tested during ASIA operation. It looks as if steps 1-3 concern fixed “input parameters”, and steps 4-5 are parts of the “managing”, but I’m in doubt. In line 242 zero padding is mentioned, but data windowing and in general any supplementary or non-Fourier alternatives are not.

Table 4: why 34 degrees here, when 30 degrees is stated elsewhere?

Line 295: “frequent attenuation”

Line 348: “from the original location”?

Line 424: “heights…are”

Line 480: “Google Earth”

Line 505: “section 5.1” should probably be “4.1”. Given the great importance of precise attitude and beam pointing, and your exceptional efforts to measure and correct for the variations, you might comment on what is determining the current thresholds for “excessive” roll/turn/climb, and whether even greater effort would allow useful data collection throughout a normal flight envelope. Are the problems distinct, i.e. the roll/turn/climb does not temporarily disturb the lidar itself (alignment / frequency jitter)? You could say this.

Line 535: some help is needed here, because “while” might mean “although” or “during [the time that]”. Perhaps say “low level flow; although the clouds…”.

Line 588: “co-PI Pu” is terse. Your Utah colleague Zhaoxia Pu is joint principal investigator?

Line 594: “ranging from”

 

When describing what the Tables of flights and success rates mean, your wording is almost too careful. The central point that DAWN has become a reliable and valuable instrument is clearly made, but there does not seem to be – perhaps there cannot be – a simple measure of its “turnkey reliability”. Suppose the aircraft is aloft and the atmospheric conditions look reasonable. What is the “probability” that DAWN will now operate without failure through the stages of warm-up, settling, and data acquisition?

This may not be what Table 6 tells us, but it is plausibly what a customer wants to know. Does the text (lines 272-279, with “DAWN had begun taking data…”) mean that:

- in every such case (“Profiles attempted”) the result was either “Not Processed” (for a good reason such as turn or climb, not a bad reason such as breakdown) or “Successful”?

- and then these “Successful” runs yielded a subset of “Full” (very successful) profiles?

- that is, there were no cases where the plane was outbound etc., and DAWN “had begun taking data”, but it stopped or failed for whatever reason? And no cases where “taking data” – although intended – did not begin, because of some fault? That seems to be what you mean, and it would be worth stressing as impressive evidence of reliability.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

This is a very nice paper showing some interesting applications of lidar data to wind profile retrievals. there are only a couple of minor clarifications that are likely needed, otherwise the paper is basically ready for publication in my opinion.

A few sentences are particularly long, that I would have phrased differently, but that is a writing style choice only and doesn't affect the soundness and the quality of the paper. Similarly, the choice of not providing full equations for the statistical descriptors used here in the validation stages doesn't really impact on the manuscript. I would have liked a bit more details on the data processing though, or at least some references, to the DAWN data retrieval algorithm and the LM damped least-squares algorithm for instance. 

something that should probably be commented more thoroughly is the following. Figures 4 (and following) suggest that there is no direct correlation between low SNR values and the capability of inverting the signal to wind data. is there any particular threshold for which this occurs? I could not find any indications on the manuscript. How is SNR defined here?

Figures 8-9 shows the comparison with dropsonde data for wind speed and direction for two different scenarios. it is interesting to note that while the general shapes match relatively well, data are offset along the z-axis suggesting a lag possibly originating from data processing or timing delay. Interestingly, the delay seem in opposite directions between the undisturbed and the disturbed scenarios. is this something that can be explained? A similar pattern seems to be present in figure 11, too.

Table 8 compares different data sets for classes of velocities. How about comparing data by classes of SNR values instead?

Something else that needs to be clarified is the comparison with NDBC buoy wind data. This refers to 25-50m data from dropsondes and DAWN. I expect NDBC data have been converted to these levels too with a logarithmic profile law. is this the case or levels have not been adjusted properly?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf