Both of the re-reviewers of the manuscript agree that the manuscript is sufficiently improved to be published after minor revisions. I attach the comments of both reviewers so that you can make these minor revisions to the manuscript. Thank you for addressing these minor changes and congratulations.


Referee 1:

"Intra-urban spatial variability of surface ozone in Riverside, CA: viability and validation of low-cost sensors", Sadhighi et al.
2nd version.

This manuscript has been improved dramatically. The decision to remove the CO2 analysis allows more focus and precision on the O3 testing, validation, and interpretation. I have very few mostly wording fixes/comments below.

Specific/technical comments:

P 3, L8: "these sensors" - is this referring to metal oxide sensors specifically, or any low-cost ozone sensor? Just repeat and make clear.

p3, L13: Seems odd to say "regional air quality body" without naming it. Is it the EPA region or the SCAQMD?

P3, L16: Also odd to mention funding source here - should this be better located in the acknowledgements at the end of the paper?

p3, L20, perhaps say "low-cost metal oxide ozone monitors"? Or low-cost ozone monitors at least, to differentiate from pm2.5 which is mentioned right before.

P6, prior to the results, I was expecting another short section on "deployment",explaining which sensors were deployed and where. Presumably the three sensors noted in the "validation" section were kept near reference monitors for "validation", but the rest were deployed throughout the region as in Fig.1. A few sentences could explain this and refer to the timeline in Fig.1 .

p8, L15, this is the first mention of Linear 4, 4T, etc. Perhaps what they are should mentioned here, or prior to Eq 1? (is Eq 1 Linear 4T?). [this was a comment in the first round, which the authors say they addressed, but not sufficiently].

p8, L18l should be Masson et al.

p9, L18 should have a comma after "contant pressure value"

p9, L16 - P10 L2. This seems odd and a bit arbitrary, to filter the UPod data just based on a maximum value seemed unrealistic. Do you have reason to think the sensors fail for some reason and read higher values than they should? Did this occur during the calibration phase, or for the validation data? Are they discontinuities from a time series (i.e. perhaps the ozone reading jumps suddenly)? At 170 are they correct?

I see the response to my original comments explain this a little more, but the amount of data points affected and final effect could be mentioned in the text as well for the other readers.

p10, lines 4-9 - this seems reasonable, so I wonder why the absolute magnitude was needed to be filtered as well. Did the high (> 171 ppb) values occur in the paired sensor in the same location as well? If not, then perhaps some agreement between pairs could be used as a data quality indicator?

P14 L5 & L9, should be "Gao et al." in both places.

P14 L9 should refer to Fig. 7

Here (Fig 7 and text) and elsewhere you should be clear with the word "concentration". I believe these are mole fractions or mixing ratios (not sure which). A concentration is a mass (or number of moles, if molar) per unit volume and would be in units of moles per liter or grams per liter. At the very least, they should define that they will be using the word concentration to mean moles of ozone per moles of dry air (or whatever is appropriate).

p15, L6-10: Although now I am clear that "collocated" = calibration period and spatially distributed = deployment period, this paragraph could be even clearer, for example:
"U-Pod ozone measurements are more correlated to each other during the calibration period (when they were collocated) than during deployment, when they were spatially distributed through the region. The R2 values between collocated pods are very high, with their medians varying from 0.92-0.99 ppbv. Conversely, during the deployment period, the spatially distributed pods were less correlated with each other, leading to R2 distribution medians between 0.52 and 0.86. "

P16 L23 should be "indicating"

Supplement:
Fig S1 typo on y-axis (magnitude)


references:

In the text, often references refer to the authors and then include them again in the citation, that should be fixed. Example p7, line 10, "Barsan and Weimar (Barsan and Weimar, 2001)" should just read Barsan and Weimar (2001) ....

There seem to be some references in the final list that are not referred to in the paper; most seem to be left over from the original manuscript version which included CO2 (Kort, Hutyra, and others). They should be checked thoroughly and removed.


Referee 2:

The revised version of the manuscript is significantly improved from the original submission. It seems like the authors have addressed all of the reviewer comments from the first round of comments and significantly revised the manuscript. I have several minor comments below.

(1) The description of the humidity term for Equation 1 is a bit confusing (I do not quite know what a "global absolute humidity" is). It seems like there should be a simpler way to express that all U-Pods, even during deployment, use the AQMS humidity measurement.

(2) Figure 1 should include the year of deployment somewhere near the timeline.

(3) Figure 3 needs a higher resolution image.

(4)Last paragraph of section 3.2 - I understand that the authors want to be totally transparent, and therefore describe the situation with the U-Pods that had electrical issues and therefore the data were not used for analysis. However, including these U-Pods throughout the rest of the manuscript (e.g., in the map in Fig 1) hampers readability. Perhaps it is appropriate to describe the calibration of these units, exclude them from Fig 1, and briefly describe why the data were not used in analysis.

(5) Figure 6 - Are the red lines in (i) and (ii) a best-fit or 1:1 line? Visually, it would help to rescale the axes so that, at the very least, the axis labels show the same numbers.

(6) The poorer performance on the right of Fig 6 is interpreted purely as time drift, though Eq 1 does include a term for the time drift. Would performance at week 9 be improved by a longer calibration co-location? This could be tested for the U-Pods that remained at an AQMS station for the entire deployment.

(7) Figures 10 and 11 focus on differences between D7 and D3. In Figure 1 it looks like D3 is co-located with D4 and DF. Do those U-Pods also see lower O3 than D7? I realize that these two U-Pods have the issue with the electronics board (Section 3.2), but based on Table 2, it seems like the electronics issue introduces a relatively stable bias that can be corrected. Do D4 and DF consistently read lower than D7 even after a bias correction?

Thank you for the revisions. Congratulations!