A Case for a New Satellite Mission for Remote Sensing of Night Lights

Round 1

Reviewer 1 Report

Qite long, but nicely done, and clearly explained. Only small issue I saw is the Table 1 (line 589) going out of the right page margin, so please correct it.

Author Response

Reviewer 1: "Only small issue I saw is the Table 1 (line 589) going out of the right page margin, so please correct it."

We were unable to correct this in the LaTeX specialtable environment and ask the assistance of the journal production staff to fit the table onto the page.

Reviewer 2 Report

The manuscript by Barentine et al. makes a case for a new satellite mission focused on observing night lights. Key features of the mission include having multiple spectral bands, high spatial resolution, and higher low light sensitivity than typically available. The ability to measure different types of lighting using the visible bands seems particularly compelling. This work is interesting and very relevant to Remote Sensing. I feel the strongest part is the detailing of science questions related to night lights in multiple categories, while the part relating these science goals to mission requirements, as well as some technical details of the proposed instrument, need to be improved.

Main comments

1. While I understand the main focus of the work is developing a set of mission requirements, some specific recommendations are a made, and the paper makes it sound like the mission as proposed would already be technically feasible. While a full optical and hardware design is clearly out of the scope, the outline of the solution should at least be clearer:

a. In the abstract, the mission is defined as being a “smallsat (<500kg)”. While many large / multi-purpose satellites are over >500 kg, this is actually a fairly nondescript claim since there is often further delineation below the 500 kg. You can see this looking at the paragraph at L551, where these satellites mentioned are widely varying in design and mass:

2U CubeSat ~ 1 kg

Luojia 1-01 (6U CubeSat ~ 20 kg)

EROS-B ~ 350 kg

See e.g. https://www.nasa.gov/content/what-are-smallsats-and-cubesats/.

b. In section 5.5, a low Earth orbit (LEO) is mentioned as fitting the requirements. Since most Earth observation satellites are in LEO, this mainly just seems to exclude a geosynchronous orbit.

Satellites in LEO with an altitude of 400 or 500 km only have a lifespan of only a few years, so presumably it must be higher. Pinning down the orbit is a little more is important since this constrains what is actually achievable in terms of repeat time, field of view, number of satellites needed, cost, satellite lifespan, etc.

Also, the claim on L716 “A LEO will, by definition, provide observations over any one location multiple times within any one year and at various times of the night” is wrong. Many Earth observation satellites are in sun-synchronous orbits with overpasses times at the roughly same time of day/night.

c. Does the proposed mission include along track scanning? Parts of the writing seem like not, perhaps for cost and simplicity, but it should be mentioned. While spatial resolution is specified, I don’t think there is any choice but to also give some consideration to field of view / area coverage / swath (if applicable) since this affects (along with orbit design) the temporal aspect of the data. I think this is important since the mission is being presented as something that is already feasible and that should be built as soon as possible.

2. Repeat time:

Of the design requirements, having only a monthly satellite overpass frequency seems to be very restrictive, in particular for helping answer the policy questions. Not to mention, some nights will be cloudy so there will be even less data. I don’t understand why other key parameters seem to be specified around what would be desirable and a big advance over VIIRS-DNB, while the temporal aspect is relaxed.

3. Specification of spectral bands:

In my research, the most interesting “night” product of VIIRS is actually not even the visible lights per se, but the fire and flaring measurements analyzed also using infrared, since it provides quantitative information about combustion. Science question EV3 also specifically mentions oil flares (sidenote; suggest calling it ‘oil and gas’ flares). On L579-L580, it is briefly mentioned “The addition of a thermal IR band for cloud and fire detection is also highly recommended.” I think this should be discussed a little further. For VIIRS at least, it does not use what is typically called a “thermal” band, but rather multiple shortwave / near-IR bands.

Also missions often aim for continuity / comparability with previous data, in this case, a relevant one being VIIRS-DNB. Here the proposed panchromatic band is bluer (370-870 vs 500-900 nm), so it would not be directly comparable. However, 383-503 nm seems like it could be used to make the measurements more comparable. Thoughts on this aspect?

4. Relating satellite measurement from the mission to surface conditions:

Night sky research traditionally has been the domain of the astronomy community, however, oddly it is not mentioned at all. Are any research questions about stargazing / seeing conditions relevant for this mission? The ability to see a non-light polluted night sky could be considered an environmental or human concern.

On a related note, presumably the ecology questions also depend on surface conditions (including artificial lights and natural lights, modulated by weather conditions), not necessarily the satellite-sensed upward radiance. While one question does get a this, question C7, I think more could be said about it.

5. Discussion section, especially the last paragraph:

L789 I wouldn’t really say a ‘modest solution’ is proposed (at least not yet). The manuscript details mainly science questions and a set of mission requirements that would meet them, with the specifics of the “smallsat’ solution being unclear

L794 ‘The technologies are available.’ – again, I don’t see evidence that the feasibility has really been thought through enough to claim this

L791 WorldView-3 is complemented here, but its drawback is not yet mentioned

6. Summary section

- ‘aided and abetted’ : suggests criminal conduct; this is inappropriate here

- ‘We find that the NITESat concept is highly cost effective…’ : cost effectiveness has not been evaluated or presented in any way

- ‘Considering its potential social and scientific benefits, the time for the mission we describe here is now, and the potential return on investment for the effort is high.’ : this could go in an opinion / commentary piece, but should not be used in this research article. Additionally, the Summary as written mainly talks about the motivation for the mission. It should also summarize the mission requirements that were determined and any other key findings of the work.

Additional:

L19 is a little misleading, since VIIRS in general has spectral information from multiple bands, but the DNB that is designed for night time research is broadband. Also on L20, VIIRS-DNB not only has better spatial resolution than OLS but much better dynamic range and quantization.

L38 just to check, where does the 0.7% area comes from? The cited paper, frustratingly, just seems to discuss trends.

L43 while the cited reference does mention light pollution, it seems to be in the opposite sense of how it is used in the sentence. The ref discusses artificial lighting not as a “threat” to public safety, but rather that artificial lighting enhances safety and light pollution is a negative externality of these efforts that needs to be studied

L138-146 these top level NASA goals are very broad. I don’t think they need to be reproduced here

L266, EV2 – what does “visible spectrum as a ratio of total emissions” mean here? What is “total emissions” here, and is this something that is designed for in the mission specs?

Table 1 is chopped off

Figure 4, in the Economy (EO) box – “o local” should be “of local”

L748 delete second period

Author Response

We quote Reviewer 2 below in italics and give our responses in plain text (except where quoting directly language we have added or changed, which we also note in italics). 

1. While I understand the main focus of the work is developing a set of mission requirements, some specific recommendations are a made, and the paper makes it sound like the mission as proposed would already be technically feasible. While a full optical and hardware design is clearly out of the scope, the outline of the solution should at least be clearer:

1. In the abstract, the mission is defined as being a “smallsat (<500kg)”. While many large / multi-purpose satellites are over >500 kg, this is actually a fairly nondescript claim since there is often further delineation below the 500 kg. You can see this looking at the paragraph at L551, where these satellites mentioned are widely varying in design and mass:

2U CubeSat ~ 1 kg

Luojia 1-01 (6U CubeSat ~ 20 kg)

EROS-B ~ 350 kg

See e.g. https://www.nasa.gov/content/what-are-smallsats-and-cubesats/.

We have updated language in the manuscript (last paragraph of Section 1; first paragraph of the Discussion section) to suggest that while NITESat is "achievable" as a small-satellite (<500 kg) mission based on a survey of existing facilities, it may also be equally achievable in a considerably smaller mission. The remarks about NITESat being a “smallsat” mission were intended to suggest an upper mass limit without precluding, e.g., a cubesat or even a nanosat mission. The mission may even be achievable as an instrument flying on an existing platform such as the International Space Station. 

We believe this addresses the reviewer's comment without requiring a detailed engineering analysis that is beyond the scope of this paper.

5. In section 5.5, a low Earth orbit (LEO) is mentioned as fitting the requirements. Since most Earth observation satellites are in LEO, this mainly just seems to exclude a geosynchronous orbit. Satellites in LEO with an altitude of 400 or 500 km only have a lifespan of only a few years, so presumably it must be higher. Pinning down the orbit is a little more is important since this constrains what is actually achievable in terms of repeat time, field of view, number of satellites needed, cost, satellite lifespan, etc.

Also, the claim on L716 “A LEO will, by definition, provide observations over any one location multiple times within any one year and at various times of the night” is wrong. Many Earth observation satellites are in sun-synchronous orbits with overpasses times at the roughly same time of day/night.

We agree with these comments. We did not consider the full range of orbits. Our intent was to describe low-Earth orbits that are specifically not Sun-synchronous, but that meaning was not connoted by the language we used. Therefore, in the revised manuscript, we clarified this point by changing the sentence to read: "A satellite in non-Sun-synchronous LEO can provide observations over any one location multiple times within any one year and at various times of the night.” We made the same change to other instances in the manuscript in order to note the more specific types of LEO we have in mind. Furthermore, we added language indicating that our preferred orbit is "near" 60 degrees in order to cover much of the inhabited portion of the Earth as mentioned at the start of the paragraph in Section 5.5.

1. Does the proposed mission include along track scanning? Parts of the writing seem like not, perhaps for cost and simplicity, but it should be mentioned. While spatial resolution is specified, I don’t think there is any choice but to also give some consideration to field of view / area coverage / swath (if applicable) since this affects (along with orbit design) the temporal aspect of the data. I think this is important since the mission is being presented as something that is already feasible and that should be built as soon as possible.

We respectfully disagree that this manuscript is the right place to discuss such topics. We believe that, in describing the mission concept and how it addresses various science cases, certain engineering decisions are beyond the scope of this paper by virtue of the fact that those decisions do not necessarily interact with the science requirements. While along-track scanning, which we understand to be the same as the "push broom" or "whisk broom" scanning method, may or may not be a preferred design approach, we don't believe such recommendations belong in the current presentation. 

2. Repeat time:

Of the design requirements, having only a monthly satellite overpass frequency seems to be very restrictive, in particular for helping answer the policy questions. Not to mention, some nights will be cloudy so there will be even less data. I don’t understand why other key parameters seem to be specified around what would be desirable and a big advance over VIIRS-DNB, while the temporal aspect is relaxed.

Note we have already clarified that we mean non-Sun-synchronous LEO elsewhere in the revised manuscript. We further updated the text to refer to "temporal sampling" rather than "return time" in order to reduce potential confusion. We have updated the language accordingly throughout the manuscript and in the tables and figures.

3. Specification of spectral bands:

In my research, the most interesting “night” product of VIIRS is actually not even the visible lights per se, but the fire and flaring measurements analyzed also using infrared, since it provides quantitative information about combustion. Science question EV3 also specifically mentions oil flares (sidenote; suggest calling it ‘oil and gas’ flares). 

We changed the wording of EV3 as suggested to read "oil and gas flares" per the reviewer's suggestion.

On L579-L580, it is briefly mentioned “The addition of a thermal IR band for cloud and fire detection is also highly recommended.” I think this should be discussed a little further. For VIIRS at least, it does not use what is typically called a “thermal” band, but rather multiple shortwave / near-IR bands. Also missions often aim for continuity / comparability with previous data, in this case, a relevant one being VIIRS-DNB. Here the proposed panchromatic band is bluer (370-870 vs 500-900 nm), so it would not be directly comparable. However, 383-503 nm seems like it could be used to make the measurements more comparable. Thoughts on this aspect?

We understand this to be two comments that we addressed individually. One was to better motivate the suggestion of a thermal-IR channel, which was also mentioned by Reviewer 3. The other was a suggestion to enable a kind of historical continuity with VIIRS-DNB passband by enabling some kind of more or less direct comparison between the two. We have added text to address both of these issues.

4. Relating satellite measurement from the mission to surface conditions:

Night sky research traditionally has been the domain of the astronomy community, however, oddly it is not mentioned at all. Are any research questions about stargazing / seeing conditions relevant for this mission? The ability to see a non-light polluted night sky could be considered an environmental or human concern.

We do not consider such issues within the scope of the proposed NITESat mission and therefore do not discuss them.

On a related note, presumably the ecology questions also depend on surface conditions (including artificial lights and natural lights, modulated by weather conditions), not necessarily the satellite-sensed upward radiance. While one question does get a this, question C7, I think more could be said about it.

To address this concern, we have added some text to Section 2.1.1: "However, remote sensing data alone are insufficient to address many ecology research topics. Rather, a range of information sources, including ground-based measurements, is needed to validate biologically relevant ALAN exposures. While upward radiance sensed from orbit may represent a proxy for the total light emissions from ground-based sources, the relationship between upward radiance and illuminance of surfaces on the ground remains poorly constrained. If the relationship is well understood in the future, we may obtain a more reliable calibration between upward radiance and biologically relevant ALAN."

5. Discussion section, especially the last paragraph:

L789 I wouldn’t really say a ‘modest solution’ is proposed (at least not yet). The manuscript details mainly science questions and a set of mission requirements that would meet them, with the specifics of the “smallsat’ solution being unclear

We changed the word "modest" to "feasible", given that there is no technological barrier to flying the mission we proposed. It is an issue of cost, an estimate of which is beyond the scope of this paper. The intent, rather, is to make the science case for the mission, leaving the detailed engineering specifications and a cost estimate to a future funding proposal. We further dropped a mention to the smallsat class in the abstract for the same reason.

L794 ‘The technologies are available.’ – again, I don’t see evidence that the feasibility has really been thought through enough to claim this

The imaging technologies are available and have been proven in past missions. We clarified this, adding the word 'imaging': "Regardless, the results of some of these commercial missions demonstrate that the imaging technology to observe nighttime lights is mature, commercially viable, and capable of significantly advancing the science of ALAN."

L791 WorldView-3 is complemented here, but its drawback is not yet mentioned

To make clear that we do not propose a platform like Worldview as a solution, we changed the end of the last paragraph before the end of Section 3 to read "While themselves not suited to addressing the science questions articulated here, the results of some of these missions demonstrate that the imaging technology to observe nighttime lights is mature, commercially viable, and capable of significantly advancing the science of ALAN." Also, in that section, we now point out the drawbacks of these commercial missions: "Often the data from these missions are not freely available to the public, are limited in coverage, or require expensive commercial requests for targeting, which severely limits their utility for research."

6. Summary section

- ‘aided and abetted’ : suggests criminal conduct; this is inappropriate here

This is now changed to "resulting in part from”.

- ‘We find that the NITESat concept is highly cost effective…’ : cost effectiveness has not been evaluated or presented in any way

This is a fair criticism. We have changed this sentence to read "We find that the NITESat concept can be effective in opening new fronts…” so as to avoid putting our comments in a financial or cost context.

- ‘Considering its potential social and scientific benefits, the time for the mission we describe here is now, and the potential return on investment for the effort is high.’ : this could go in an opinion / commentary piece, but should not be used in this research article. 

We have deleted the language about "return on investment" in the strict sense that we did not formally estimate the cost of NITESat. That is beyond the scope of this paper. 

Additionally, the Summary as written mainly talks about the motivation for the mission. It should also summarize the mission requirements that were determined and any other key findings of the work.

Without simply reiterating the contents of Section 5, we added a new penultimate paragraph to the Summary section that we think captures the important takeaway messages of the paper: "We have here described the science goals and mission requirements for a Nighttime Imaging of Terrestrial Environments Satellite, or NITESat. This mission leverages existing technologies deployed in previous missions in which the remote sensing of nighttime lights was at best a secondary objective. Such mission designs were therefore not optimized for the purpose of observing night lights; we propose a new mission with this as its primary mission. NITESat in particular improves on the ground scale resolution and spectral coverage of previous missions and would add varied overpass times to address a broader set of science questions than was possible in earlier cases. We argue that these capabilities constitute a new way to see the phenomenon of night lights."

Additional:

L19 is a little misleading, since VIIRS in general has spectral information from multiple bands, but the DNB that is designed for night time research is broadband. 

In order to make the intended meaning clearer, we have changed all relevant instances of "VIIRS" and "OLS" throughout the manuscript to "VIIRS-DNB" and "DMSP-OLS", respectively. The only exceptions are where we refer to either the VIIRS or DMSP instruments generically and without respect to any particular sensor or passband.

Also on L20, VIIRS-DNB not only has better spatial resolution than OLS but much better dynamic range and quantization.

We have added words to that effect in the abstract.

L38 just to check, where does the 0.7% area comes from? The cited paper, frustratingly, just seems to discuss trends.

While we believe that we calculated this from data in Falchi et al. (2016) and Kyba et al. (2017), none of us could find the basis for the figure. Given that it has no bearing on the rest of the paper, we have therefore deleted this sentence.

L43 while the cited reference does mention light pollution, it seems to be in the opposite sense of how it is used in the sentence. The ref discusses artificial lighting not as a “threat” to public safety, but rather that artificial lighting enhances safety and light pollution is a negative externality of these efforts that needs to be studied

We agree that the original wording could be misleading. In the revised manuscript, we separated out 'public safety' from the list in this sentence and put it into a new sentence that puts it into a context of 'concern' of the interaction between outdoor lighting and public safety.

L138-146 these top level NASA goals are very broad. I don’t think they need to be reproduced here

We agree that listing these items is not strictly required for the benefit of the rest of the paper and that reviewing them can be left as an exercise for the reader. We deleted the bulleted list in this paragraph.

L266, EV2 – what does “visible spectrum as a ratio of total emissions” mean here? What is “total emissions” here, and is this 

something that is designed for in the mission specs?

To clarify our meaning, we changed this to "What is the inefficiency of outdoor lighting as the fraction of total light that is emitted at upward angles?”

Table 1 is chopped off

As mentioned above in the response to a similar note from Reviewer 1, we were unable to correct this in the LaTeX specialtable environment and ask the assistance of the journal production staff to fit the table onto the page.

Figure 4, in the Economy (EO) box – “o local” should be “of local”

L748 delete second period

We made the suggested corrections.

Reviewer 3 Report

The manuscript is well written. The scientific goal of night time remote sensing has been summarized very well and deeply discussed. Also the main specificatio of the satellite has been proposed. There are two suggestions:

1) The infrared channel has been mentioned, using for clound detection in nighttime. This very important, so the GSD and bandwidth of the infrared are suggested to added in the specification list;

2) Many application using VIIRS data faced a lot of stray lights effects, which is very harmful. The elimination requirement of stray light should be disscussed or mentioned in the manuscript, for such an advanced scientific satellite. 

Author Response

Here we quote Reviewer 3 in italics and write our responses in plain text (except where we quote language we have changed or added to the manuscript, also in italics.)

There are two suggestions:

1) The infrared channel has been mentioned, using for cloud detection in nighttime. This very important, so the GSD and bandwidth of the infrared are suggested to added in the specification list;

See our response to Reviewer 2’s similar comment. We have added to Table 2 an indication that a thermal IR channel should be included while not specifying any particular passband. We also updated the 10 m/px GSR figure to indicate that it only applies to the optical passbands.

2) Many application using VIIRS data faced a lot of stray lights effects, which is very harmful. The elimination requirement of stray light should be disscussed or mentioned in the manuscript, for such an advanced scientific satellite.

We appreciate this comment, and acknowledge that we didn't specifically take this into account in our analysis. However, to the extent that there are instrumental systematics, we specify a S/N ratio in our radiance sensitivity requirement that we believe implies a limit on allowable stray light effects. We would then leave it as a task to engineers to determine a stray light solution that yields the required radiance sensitivity at the specified S/N. To the end of Section 5.3 (Sensitivity) we have added the following text: "In specifying a radiance sensitivity threshold and minimum SNR, we indirectly take into account issues such as stray light that are known to have affected earlier missions." 

Round 2

Reviewer 2 Report

The revision was appropriate, and the work very interesting. I enjoyed reading this study and believe it is a good contribution to night light research.

Note a few details in the proof:

L100 and L861 ‘National Academy of Sciences, Engineering, and Medicine’ => ‘National Academies of Sciences, Engineering, and Medicine’

L861 under DMSP, ‘Satellilte’ => ‘Satellite’

Ref 20 start of author name is cut off

Author Response

We thank the reviewer for pointing out these copy editing and typesetting issues. They have all been corrected in the resubmitted version.