Review Reports - Calibrating Nighttime Satellite Imagery with Red Photometer Networks

Round 1

Reviewer 1 Report

As the manuscript describes a radiometric cross-calibration study between satellite and ground measurements, it covers not only conversion between the radiometric units used for VIIRS and the astronomical units used for the ground-based photometers, but also the diametral difference in the direction of the measurements: downward from space for VIIRS and upward from the ground for the photometers. The derived regression equation is presented for a select range of the measurements. While the lower limit of the range is explained by VIIRS sensitivity, it is not clear what is responsible for the deviation above the upper limit. Is it non-linearity of VIIRS (or photometer) radiometric response or perhaps an effect of atmospheric scattering?

Another issue worth addressing is the global applicability of the results. Including a map with locations of all photometer measurements selected for the regression may help readers understand this issue. While the goal of the presented study is very beneficial: to create correlation between the “light pollution” measurements from space and on the ground, the readers may be interested in knowing whether the presented results are applicable to locations they are interested in.

The study uses VIIRS Level 3 product called “Black Marble” that is produced in several processing steps. One of the effects of that is the spatial sampling of 500 m while the actual VIIRS DNB measurements are acquired with both spatial sampling and resolution of about 750 m. That information may help in understanding results from the example shown in Figure 14.

Minor suggestions for language corrections:

Line 8: “photomettric" spelling

Line 347: word “move” unneeded

Minor suggestions for language corrections:

Line 8: “photomettric" spelling

Line 347: word “move” unneeded

Author Response

Borja Fernandez-Ruiz

Medical and Audiovisual Technology Group

Facultad de Telecomunicaciones

Universidad de Las Palmas de Gran Canaria

Paseo Blas Cabrera Felipe “Físico”, s/n

Las Palmas de Gran Canaria, 35016, Spain

August 1, 2023

Dear Editors,

We would like to thank the reviewers and editor for their valuable comments and feedback. We have made significant changes to the text, attempting to address all comments and enhance the overall clarity of the paper.

We have implemented significant changes to the paper's structure. Firstly, at the editor's request, the "Results and Discussion" section has been divided into two separate sections. Section 4 contains the results presented concisely, while Section 5 covers the discussion of these results and their errors. Additionally, in Section 5, a new subsection on future applications has been added.

Secondly, the "Data description and methods" section has also been divided into two parts: Materials (Section 2) and Methods (Section 3), following the feedback from the second reviewer. Moreover, Section 3 has been restructured to provide a more detailed description of the procedure, including a new Section 3.2 that outlines the filtering process with the workflow, and a Section 3.3 that describes the regression method used.

Next, we address each of the reviewers' comments in more detail. All changes and improvements to the paper are marked in blue in the manuscript.

Reviewer 1:

“As the manuscript describes a radiometric cross-calibration study between satellite and ground measurements, it covers not only conversion between the radiometric units used for VIIRS and the astronomical units used for the ground-based photometers, but also the diametral difference in the direction of the measurements: downward from space for VIIRS and upward from the ground for the photometers. The derived regression equation is presented for a select range of the measurements. While the lower limit of the range is explained by VIIRS sensitivity, it is not clear what is responsible for the deviation above the upper limit. Is it non-linearity of VIIRS (or photometer) radiometric response or perhaps an effect of atmospheric scattering?”

We are uncertain about the cause behind this apparent deviation. Within the magnitude ranges of the study, the photometers maintain linearity as shown in Alarcón et al. (2021). Some atmospheric effects cannot be ruled out. However, the sample size above 19.41 mag (10 photometers) is not significant enough to draw conclusions about the possible cause. Hence, it is of particular importance to deploy more photometers in highly polluted areas to investigate this phenomenon further.

As a result of this comment, lines 496 to 501 have been added.

“Another issue worth addressing is the global applicability of the results. Including a map with locations of all photometer measurements selected for the regression may help readers understand this issue. While the goal of the presented study is very beneficial: to create correlation between the “light pollution” measurements from space and on the ground, the readers may be interested in knowing whether the presented results are applicable to locations they are interested in.”

We have included a map with the locations of the photometers in Section 2.1.

As a result of this comment, lines 107 to 108 have been added.

“The study uses VIIRS Level 3 product called “Black Marble” that is produced in several processing steps. One of the effects of that is the spatial sampling of 500 m while the actual VIIRS DNB measurements are acquired with both spatial sampling and resolution of about 750 m. That information may help in understanding results from the example shown in Figure 14.”

Indeed, the product used has a higher spatial resolution than the base VIIRS data. We are uncertain whether this change in resolution caused by the VNP46A2 algorithm could introduce errors in Figure 15.

As a result of this comment, lines 79 to 81 have been added.

“Line 8: “photomettric" spelling”

Corrected, line 15.

“Line 347: word “move” unneeded”

Corrected, line 435.

Reviewer 2:

- Abstract:

“ l.7-9: This sentence should be at the end of the abstract, as these are results.”

Corrected, lines 7 to 9 and 14 to 16.

“l.12: “Encoder and Sky Sensor WIFI (TESS-W) types, located at different ground-based locations”

Corrected, line 12.

- Introduction:

“The objective of the paper is not clearly stated. Lines 47-49 state that “studying the correlation between satellite and photometer network data is of particular interest, so that we can convert one into the other in order to carry out studies with both integrated sources”. So the objective of the paper is to study this correlation? However, the title mentions the calibration of the nighttime satellite imagery. Moreover, most of the results deal with the calibration of the VIIRS images based on the regression equation derived from the correlation analysis. The objective must be formulated in accordance with the work and results presented in the paper. The integration of the satellite and ground data is more of a perspective. It places the work according to the current scientific locks. Why is it important to study this correlation/calibration? Please clarify the scientific context and expectation in the field.”

The main objective is to obtain a correlation and regression that allows calibrating the measurements of the VIIRS with the measurements of the photometers SG and TESS. In the example of Valverde de Burguillos, this calibration is used to generate a map of the area in magnitude units. We have tried to clarify this throughout the text.

“l.29: “the deployment of large ground-based networks that””

Corrected, line 32.

“l.31: Explain why there is a lack of precision of the satellite in areas of high darkness”

Satellites have limited precision and sensitivity. That's why in very dark areas, they cannot measure many values. Being a linear precision, the relative uncertainty is higher in lower values. For example, the VIIRS has an uncertainty of 0.1 and a sensitivity of up to 0.5, insufficient to measure in places with more than 21.5 mag.

As a result of this comment, lines 34 and 86 to 87 have been changed.

- Data description and methods:

“This section should be divided into two sections: (1) data description (or material) and (2) Method. In (1) data description, both datasets should be described in more detail. All information related to the data should be gathered in this section and not scattered throughout the paper. For the photometers, please provide the instrument footprint and the characteristics of each sensor type (this could be put in a table). For the data acquisition, it should be specified whether the data are acquired instantaneously or as an average integrated over a time interval (this is not specified in Table1). For the satellite imagery, more details on the VNP46A2 products should be provided.”

We have divided Section 2 into two distinct parts: Section 2 Materials and Section 3 Methods. The entire Section 2.1 has been revised to add additional information about the photometers and the VIIRS, including a comparative table among the photometers.

“In (2) Method, a first subsection should present the global workflow developed in the study to achieve the objective of the paper. This should include an overview of the analyses performed to prepare the photometer data and present the procedure applied for the correlation with satellite data (like info provided in lines 300-302). It should specified how the data are correlated, is it based on ground data acquired at satellite overpass time only or nightly average? Line 276 annual averaged are mentioned, how does it fit in this correlation? In the following subsections, all photometers filtering tests can then be presented. Please clarify the objective of each test at the beginning and specify whether you intend to discard entire photometers datasets or just outliers and anomalies in the data.”

We have restructured the methods Section 3 to provide greater clarity. It is now organized into three parts: data analysis, data cleaning, and regression. The first subsection, 'data analysis,' could be separated into a region before the methods if you find it clearer. In the second subsection, we have attempted to summarize the data cleaning process based on the previous analysis, specifying the details as requested and adding a diagram to enhance clarity. Finally, in the regression subsection, we have included those parts of the results that you indicated should be in the methods. We have also tried to provide detailed motivation for each analysis.

“l.71: “VIIRS takes an image of the…””

Corrected, line 78.

“l.72: “VNP46A2 is a level 3 satellite product with…”.”

As a result of this comment, lines 80 and 81 have been added.

“Figure 1 : What are the Johnson–Cousins [23] BVRI filters?”

Johnson--Cousins BVRI filters are a widely used photometric system in astronomy introduced by Johnson (1966). We have added this new reference. As a result of this comment, Figure 1 has been changed.

“l.79-80: “For this reason, SG and TESS photometers are much more suitable when looking for correlation with satellite data.”. More suitable than what?”

SG and TESS are much more suitable than SQM photometers. As a result of this comment, line 92 has been changed.

“l.140: “such as interference from nearby artificial lights”. It is not clear why it is a problem to measure the contribution of nearby artificial light. Aren’t photometers designed for that? This sentence is confusing, please clarify this point.”

The purpose of photometers is to measure the skyglow at the zenith. Poor placement of the photometer can result in receiving direct light from a nearby lamp, which hinders taking accurate measurements.

As a result of this comment, lines 21, 22 and 176 have been changed for a better explanation. The introduction has also been modified.

“l.139-145: This is not clear what is the purpose of this test. Even if the sub-section title is explicit, this should be clearly expressed in the text. It should also be mentioned what is discarded (data or entire dataset from a given photometer?).”

As a result of this comment, lines 181 to 183 have been changed. The first paragraph of Section 3.1 has also been added. In section 3.2, details about data cleaning are provided.

“l.142: On what is computed the median P50? On the entire year of 2022? If yes, it should be clearly stated.”

As a result of this comment, lines 178 to 180 have been changed.

“l.147: Explain why subsampling of the original sample are used.”

Bootstrapping is a method that involves creating subsamples from the original sample, similar to how conventional statistics take samples from a population. This allows us to estimate how the statistics of the original population behave.

“l.160: “as it approaches 1000 data points and stabilizes thereafter””

Corrected, line 198.

“l.179: “date” should be “data””

Corrected, line 218.

“l.183: “Photometers that have less than 500 data points have not been considered.” The link of this statement with the test on the Milky Way effect is not clear (as this limit of 500 data points was already mentioned in the previous section). Please, clarify.”

Indeed, this sentence is redundant with the previous section it was meant to refer to. Therefore, we have removed it to avoid confusion, lines 222 and 223.

“l.185: Specify how is estimated the partial presence of the Milky Way.”

Using the galactic latitude. When the absolute value of the galactic latitude is less than 40º, it is considered to have a sufficient impact on the skyglow to be taken into account. The corresponding reference has also been added.

As a result of this comment, lines 225 to 227 have been added.

“Section 2.4: It is not clear if the objective of this test is to discard or to correct the data of the Milky Way effect.”

The photometer filters always discard the points where they are affected by different phenomena, as subtracting their contribution in the skyglow is very complex.

As a result of this comment, lines 146 to 147 have been added.

“l.208-210: I don’t understand this sentence and its link with the test on the Milky Way effect. Please, clarify.”

We compare the effect of the galaxy with the brightness of the location to determine which photometers need to apply this filter. For example, photometers located in cities will not require filtering because their effect will not be significant.

As a result of this comment, line 232 and 235 have been changed.

“Figure 5: What are the dashed lines in the left figure?”

As a result of this comment, Figure 6 has been added.

“Section 2.5: Develop why the Gaussian approximation is needed for, what is the final objective of this test? Clarify step by step this test and what is its objective.”

We describe the characteristic distribution type of the photometers. This enable us to identify those photometers with anomalous distributions and thus unreliable measurements.

As a result of this comment, lines 254 to 256 have been added. The first paragraph of Section 3.1 has also been added.

“l.276 : “comparing an annual average with satellite data”, this is the first time it is mentioned that the analysis will be performed on annual average. This should be mentioned earlier, in the description of the global workflow.”

We have tried to clarify this throughout Section 3.1 in the first paragraph. In section 3.2, details about data cleaning are provided.

“l.288: Could you provide a distribution of the photometers between dark and bright areas (global percentage are enough) “

I think I don’t understand you. The proportion between dark and bright photometers is shown in Figure 3.

-Results:

“l.307: Explain why the VIIRS measurements are expressed in log?”

As a result of this comment, lines 392 to 393 have been added.

“l.352-353: “The highest concentration of photometers from the EELabs and STARS4ALL networks is located in the municipality of Valverde de Burguillos (Extremadura, Spain),””

Corrected, lines 440 to 442.

“l.370-371: Provide some explanations for this observation.”

We are unaware of the reason for the large variability in the VIIRS data. One possible cause could be the component of upward ascending direct emissions that the photometers do not measure.

As a result of this comment, lines 460 to 461 have been added.

- Conclusion:

“As for the introduction, the objective should be clarified and the conclusions written to answer them. The word “calibration” is only use in the title, the abstract and then it appears again l.290 and in the conclusion. There is a confusion in the text between calibration and correlation. The conclusion should also open on future applications of the presented work.”

We have added more references to the calibration in an effort to clarify the text. The mention in line 290 referred to the calibration of the photometers as an instrument, and it has been modified accordingly in line 335.

The future applications are discussed in the discussion section as indicated by the editor. They could also be included in the conclusions if deemed necessary.

Reviewer 3:

“Great paper - just check for typos - on line 70 adjust "Suaomi" to "Suomi" and on line 91 I think you mean "Macaronesia" and not "Macronesia" correct?”

As a result of this comment, lines 77 and 104 have been added.

Reviewer 4:

“This manuscript focuses on the calibration of nighttime satellite imagery using red photometer networks. The method and data are well described, and the results were well analyzed and discussed. Overall, the readability of the article is good and can attract readers' attention. However, the introduction section does not fully introduce other existing research work, and it is recommended to make modifications and supplements to allow readers to have a better understanding of the overall situation of the research work.“

There are not many studies on this correlation; however, we have added a reference to a study from Indonesia where, similar to Madrid, they observed a correlation between SQM-VIIRS.

As a result of this comment, line 54 has been changed.

Thank you for your reviews and comments.

Sincerely,

Borja Fernandez-Ruiz, Miquel Serra-Ricart, Miguel R. Alarcon, Samuel Lemes-Perera, Idafen Santana-Perez and Juan Ruiz-Alzola

Author Response File: Author Response.pdf

Reviewer 2 Report

The submitted paper presents an approach to calibrate VIIRS nighttime satellite imagery by investigating the correlation with ground-based photometers data in the context of nighttime artificial light pollution studies. A large part of the work is dedicated to the analysis and filtering of the data acquired by the photometers. The correlation between ground and satellite data is analyzed to derive a regression equation for the calibration of VIIRS images. This regression is then applied to a real case over Valverde de Burguillos.

The idea of studying the correlation between nighttime satellite images and ground data to be able to combine them in the future is very interesting, especially considering the potential spatial and temporal benefits. The availability of a large amount of ground data is a valuable point of the work. However, the objective of the study is not clearly stated and its future impacts are not sufficiently described. Section 2 is confusing and some information is missing to fully understand the methodology and experimental choices made during the study. Therefore, major revisions are recommended. The introduction should be expanded to clarify the objective and context of the study. The description of the data should be more detailed and the methodology section should be reorganised and clarified. Below are some specific comments on the article:

- Abstract:

o l.7-9: This sentence should be at the end of the abstract, as these are results.

o l.12: “Encoder and Sky Sensor WIFI (TESS-W) types, located at different ground-based locations”

- Introduction:

o The objective of the paper is not clearly stated. Lines 47-49 state that “studying the correlation between satellite and photometer network data is of particular interest, so that we can convert one into the other in order to carry out studies with both integrated sources”. So the objective of the paper is to study this correlation? However, the title mentions the calibration of the nighttime satellite imagery. Moreover, most of the results deal with the calibration of the VIIRS images based on the regression equation derived from the correlation analysis. The objective must be formulated in accordance with the work and results presented in the paper. The integration of the satellite and ground data is more of a perspective. It places the work according to the current scientific locks. Why is it important to study this correlation/calibration? Please clarify the scientific context and expectation in the field.

o l.29: “the deployment of large ground-based networks that”

o l.31: Explain why there is a lack of precision of the satellite in areas of high darkness

- Data description and methods:

o This section should be divided into two sections: (1) data description (or material) and (2) Method. In (1) data description, both datasets should be described in more detail. All information related to the data should be gathered in this section and not scattered throughout the paper. For the photometers, please provide the instrument footprint and the characteristics of each sensor type (this could be put in a table). For the data acquisition, it should be specified whether the data are acquired instantaneously or as an average integrated over a time interval (this is not specified in Table1). For the satellite imagery, more details on the VNP46A2 products should be provided.

In (2) Method, a first subsection should present the global workflow developed in the study to achieve the objective of the paper. This should include an overview of the analyses performed to prepare the photometer data and present the procedure applied for the correlation with satellite data (like info provided in lines 300-302). It should specified how the data are correlated, is it based on ground data acquired at satellite overpass time only or nightly average? Line 276 annual averaged are mentioned, how does it fit in this correlation? In the following subsections, all photometers filtering tests can then be presented. Please clarify the objective of each test at the beginning and specify whether you intend to discard entire photometers datasets or just outliers and anomalies in the data.

o l.71: “VIIRS takes an image of the…”

o l.72: “VNP46A2 is a level 3 satellite product with…”.

o Figure 1 : What are the Johnson–Cousins [23] BVRI filters?

o l.79-80: “For this reason, SG and TESS photometers are much more suitable when looking for correlation with satellite data.”. More suitable than what?

o l.140: “such as interference from nearby artificial lights”. It is not clear why it is a problem to measure the contribution of nearby artificial light. Aren’t photometers designed for that? This sentence is confusing, please clarify this point.

o l.139-145: This is not clear what is the purpose of this test. Even if the sub-section title is explicit, this should be clearly expressed in the text. It should also be mentioned what is discarded (data or entire dataset from a given photometer?).

o l.142: On what is computed the median P50? On the entire year of 2022? If yes, it should be clearly stated.

o l.147: Explain why subsampling of the original sample are used.

o l.160: “as it approaches 1000 data points and stabilizes thereafter”

o l.179: “date” should be “data”

o l.183: “Photometers that have less than 500 data points have not been considered.” The link of this statement with the test on the Milky Way effect is not clear (as this limit of 500 data points was already mentioned in the previous section). Please, clarify.

o l.185: Specify how is estimated the partial presence of the Milky Way.

o Section 2.4: It is not clear if the objective of this test is to discard or to correct the data of the Milky Way effect.

o l.208-210: I don’t understand this sentence and its link with the test on the Milky Way effect. Please, clarify.

o Figure 5: What are the dashed lines in the left figure?

o Section 2.5: Develop why the Gaussian approximation is needed for, what is the final objective of this test? Clarify step by step this test and what is its objective.

o l.276 : “comparing an annual average with satellite data”, this is the first time it is mentioned that the analysis will be performed on annual average. This should be mentioned earlier, in the description of the global workflow.

o l.288: Could you provide a distribution of the photometers between dark and bright areas (global percentage are enough)

- Results:

o l.307: Explain why the VIIRS measurements are expressed in log?

o l.352-353: “The highest concentration of photometers from the EELabs and STARS4ALL networks is located in the municipality of Valverde de Burguillos (Extremadura, Spain),”

o l.370-371: Provide some explanations for this observation.

- Conclusion:

o As for the introduction, the objective should be clarified and the conclusions written to answer them. The word “calibration” is only use in the title, the abstract and then it appears again l.290 and in the conclusion. There is a confusion in the text between calibration and correlation. The conclusion should also open on future applications of the presented work.