The Dark Target Algorithm for Observing the Global Aerosol System: Past, Present, and Future

Round 1

Reviewer 1 Report

This paper provides an review of dark-target aerosol retrievals, centered on the developments for the MODIS sensors. It provides information on the algorithm principles, historic development, and perspectives. In my view, this is a well-written, mature and immensely useful overview produced first-hand by a group of authors that drove and drive the developments in the field.

I consider this manuscript well worthy of publication, and only have a few very minor suggestions, which the authors may or may not consider. Congratulations on this summary, and on the work behind it!

- Abstract: Briefly stating what 'dark target' means might be useful for orienting a reader without prior knowledge.
- A (very) short overview of aerosol optical properties, and hence the physical basis of detecting aerosol from space relative to surface properties and wavelength would be useful addition in my view.
- The developments on MODIS are at the center of this paper, and this seems justified to me, given that algorithm innovations were made here. However, since the title is not platform-specific, a small overview on how MODIS developments inspired algorithms on other platforms would be of interest to this reader.

Author Response

Response for Reviewers.

Reviewer’s comment is in italics, and our response in plain text.

Reviewer #1.

This paper provides an review of dark-target aerosol retrievals, centered on the

developments for the MODIS sensors. It provides information on the algorithm

principles, historic development, and perspectives. In my view, this is a wellwritten,

mature and immensely useful overview produced first-hand by a group of

authors that drove and drive the developments in the field.

I consider this manuscript well worthy of publication, and only have a few very

minor suggestions, which the authors may or may not consider. Congratulations

on this summary, and on the work behind it!

Thank you for the kind words.  We are glad that you found the work useful.  We will be honest; writing this manuscript took quite a bit of work, and we were not sure how it would be received.  The three suggestions do indeed improve the manuscript and we have tried to incorporate each one.  Line numbers refer to lines in the  revised manuscript.

- Abstract: Briefly stating what 'dark target' means might be useful for orienting a

reader without prior knowledge.

We have added a second sentence to the Abstract to define dark target.

Lines 35-37. ‘The algorithm is based on measurements of the light scattered by aerosols towards a space-borne sensor against the backdrop of relatively dark Earth scenes, thus the name “Dark Target”.’

- A (very) short overview of aerosol optical properties, and hence the physical

basis of detecting aerosol from space relative to surface properties and

wavelength would be useful addition in my view.

We have added a paragraph early in the Introduction.  It is not easy being brief on this subject. We hope that this new paragraph is helpful despite its brevity. 

Lines 72 -88. Satellites provide the best means of observing and characterizing the global aerosol system [1]. These space-borne sensors point at Earth scenes and measure radiation that has been reflected or emitted by the Earth’s surface and atmosphere. Satellite aerosol remote sensing requires deconvolving the signal measured by the sensor into a component originating from the Earth’s atmosphere from that originating from the surface beneath, and then isolating the aerosol signal from other atmospheric constituents. The process requires extensive knowledge of aerosol optical properties: the degree to which particles emit, absorb or scatter incident radiation. For the purpose of this paper, we will consider only the absorption and scattering of solar radiation, and not emission at longer wavelengths. Particle properties determine the fraction of radiation scattered in a particular direction (i.e. towards the satellite), and this information is wavelength dependent. Higher particle concentrations increase the measurable radiation scattered towards the satellite. The range of radiation levels measured by the sensor are translated into a quantification of aerosol loading in the observed scene. Particle optical properties depend on their chemical composition as well as their size and shape. If the satellite sensor measures sufficient angular and/or spectral information, and if the aerosol signal can be sufficiently separated from the signals due to the surface and other atmospheric constituents, then other information about the aerosol physical properties can be inferred from the measurements.

- The developments on MODIS are at the center of this paper, and this seems

justified to me, given that algorithm innovations were made here. However, since

the title is not platform-specific, a small overview on how MODIS developments

inspired algorithms on other platforms would be of interest to this reader.

We have enhanced the 7^th paragraph of the Introduction to preview some of the statements and summaries found in later sections about applying the DT algorithm to VIIRS and geosynchronous sensors, and also some of the influence that the DT algorithm has had on other algorithm families.  The new 7^th paragraph now reads:

Lines 42-61.  The impact of the DT aerosol algorithm on remote sensing science, on Earth science and on applications has been immense. Initially applied to the MODIS sensors on the Terra and Aqua satellites, the algorithm grew to be much more than “the MODIS aerosol algorithm”. The algorithm is now retrieving aerosol properties from new sensors such as the Visible Infrared Imaging Radiometer Suite (VIIRS) on the Suomi-National Polar-orbiting Partnership (S-NPP) satellite and has been adapted non-operationally for a fleet of geosynchronous sensors. In addition to direct adaptation the DT algorithm has inspired other algorithm families in a variety of ways. including introducing techniques for cloud and snow masking, and the development of best-practices for the validation of satellite-based aerosol products. These influences will be discussed in Section 7.5. Meanwhile, many advances in understanding how aerosols affect climate, clouds, biological systems and air quality can be traced directly to research using DT products. We understand the global aerosol system now, much better than we did twenty years ago, in part due to the influence of this one algorithm.

Reviewer 2 Report

This work is a high quality general review of atmospheric remote sensing from space/ground beyond explaining the DT-algorithm. It’s very useful for young researchers and not young myself. Especially the references would like to be put as an annotated bibliography on the right.

Author Response

Response for Reviewers.

Reviewer’s comment is in italics, and our response in black.

Reviewer #2

This work is a high quality general review of atmospheric remote sensing from

space/ground beyond explaining the DT-algorithm. It’s very useful for young

researchers and not young myself. Especially the references would like to be put

as an annotated bibliography on the right.

Thank you, we are happy that you found it useful, and especially happy that you noted that the references provide an annotated bibliography for the community.  The 272 references were quite the challenge to implement.

It appears that this reviewer did not make any suggestions.

Reviewer 3 Report

This paper will make a major contribution to scientists working in the fields of aerosol climatology and air quality monitoring. The paper represents a milestone in the 3 decades of development of the Dark Target AOD algorithm, as it summarises the huge efforts to date that have gone into creation and refinement of the algorithm.

This lengthy paper (2-3 times longer than most) is well organised and written, thus easy to read. The sections describe logical aspects of the DT development and permit readers to focus on aspects of interest.  The  concise and detailed descriptions allow data users to quickly grasp how the AOD images have been generated so users can evaluate their relevance and/or levels of uncertainty, as well as which DT product to select.  The paper will make a valuable contribution to users of DT product. The enormous (and frankly impressive) reference list is also an invaluable resource for the remote sensing technical and scientific community. The up-to-date figures such as 11 and 12 well illustrate the significance and impact of DT on the discipline.

It is appreciated that the authors focus not only on the achievements of DT but also on the challenges, many of which remain.  The paper is timely, as the way forward will be influenced by incorporation of DT into more new sensors, whereas past development had focussed mainly on MODIS.

I have a few minor comments:

Abstract, Line 42. ‘Rock-solid’ is an unusual term for scientists to use, as uncertainty is always recognised.  What about AERONET sensor drift? do annual calibration and age of the sunphotometers really eliminate all error?

Line 46-47. Either delete ‘between’, or change ‘from’ to ‘and’

Line 137. ‘spatially’ should be ‘spatial’

Line 365. ‘to’ should be ‘at’

Line 541. 400 AERONET sites today – are all of these currently operating? I know that the temporary nature of many AERONET sites is discussed later in the paper, but if 400 are not currently operating it is a bit misleading.

Line 609. ‘than’ should be ‘from’

Line 735- delete comma

Line 753. ‘overall’ is one word

Line 783-4. The four geographical regions referred to would be of interest to users – it would be good to state which regions

Line 810.  ‘validation on a global basis’

Line 1101 Results.  This section describes the main impacts of the DT algorithm on the science community and on our understanding of global aerosols. The heading ‘Results’ usually follows a Methods section where an experiment or hypothesis for testing is set up. Therefore a more precise section heading such as   ‘Major impacts of the DT algorithm’ is suggested

Line 1185 ‘databases’ is one word

Author Response

Response for Reviewers.

Reviewer’s comment is in italics, and our response in plain text.  Line numbers in our response refer to the new line numbers of the revised paper.

Reviewer #3

This paper will make a major contribution to scientists working in the fields of

aerosol climatology and air quality monitoring. The paper represents a milestone

in the 3 decades of development of the Dark Target AOD algorithm, as it

summarises the huge efforts to date that have gone into creation and refinement

of the algorithm.

This lengthy paper (2-3 times longer than most) is well organised and written, thus

easy to read. The sections describe logical aspects of the DT development and

permit readers to focus on aspects of interest. The concise and detailed

descriptions allow data users to quickly grasp how the AOD images have been

generated so users can evaluate their relevance and/or levels of uncertainty, as

well as which DT product to select. The paper will make a valuable contribution to

users of DT product. The enormous (and frankly impressive) reference list is also

an invaluable resource for the remote sensing technical and scientific community.

The up-to-date figures such as 11 and 12 well illustrate the significance and

impact of DT on the discipline.

It is appreciated that the authors focus not only on the achievements of DT but

also on the challenges, many of which remain. The paper is timely, as the way

forward will be influenced by incorporation of DT into more new sensors, whereas

past development had focussed mainly on MODIS.

Thank you for the very generous statements.  As we wrote to the other reviewers, this was not easy, and especially the commitment to providing an annotated bibliography was quite the challenge. It is very gratifying to receive such positive reactions from the first set of readers of the manuscript.

We appreciate the identification of the items below.  Each change to the manuscript includes the new line number for reference.

I have a few minor comments:

Abstract, Line 42. ‘Rock-solid’ is an unusual term for scientists to use, as

uncertainty is always recognised. What about AERONET sensor drift? do annual

calibration and age of the sunphotometers really eliminate all error?

I have a tendency to use colorful language when I write or make presentations (This is L. Remer writing).  Colorful, not meaning expletives, but language that is not often found in science.  One of the co-authors made it her mission to strike all of my colorful language out of this paper.  She missed this one.

I have simply removed the “rock-solid” so that the sentence now reads:

Lines 41-44.  Contributing to that understanding were the observations and retrievals of the growing AERONET network of sun-sky radiometers, which has walked hand-in-hand with MODIS and other aerosol algorithm development, and after launch providing validation of the satellite-retrieved products.  

Line 46-47. Either delete ‘between’, or change ‘from’ to ‘and’

‘between’ removed so that the sentence now reads

Lines 46-56.  Then as the Terra and Aqua MODIS sensors aged, the challenge was to monitor the effects of calibration drifts on the aerosol products and to differentiate physical trends in the aerosol system from artifacts introduced by instrument characterization.

Line 137. ‘spatially’ should be ‘spatial’

Line: 186: ‘spatial’ has replaced ‘spatially’

Line 365. ‘to’ should be ‘at’

Line 469:  ‘at’ has replaced ‘to’

Line 541. 400 AERONET sites today – are all of these currently operating? I know

that the temporary nature of many AERONET sites is discussed later in the paper,

but if 400 are not currently operating it is a bit misleading

Today (20 August 2020) I went to the AERONET site and counted the number of stations reporting at least some Level 1 data in August 2020.  The number was exactly 370, so 400 is a bit of rounding up.  We have made the change to:

Line 721-722: The network of a few AERONET sensors in 1998 has grown to at least 370 active sites today (August 2020).

Line 609. ‘than’ should be ‘from’

Line 743: ‘from’ has replaced ‘than’

Line 735- delete comma

Line 863: comma deleted

Line 753. ‘overall’ is one word

Line 895: ‘overall’ is now one word

Line 783-4. The four geographical regions referred to would be of interest to users

– it would be good to state which regions

It took a whole new paragraph to provide that information:

Lines 930-944. Specifically, the four models included three that were dominated by fine mode particles and one dominated by a coarse mode.  The three fine mode models had small differences in their size distributions, but were distinctive in their light absorption properties.  The model with the most light absorption, with midvisible single scattering albedo (SSA) of 0.85, was found seasonally in tropical and southern hemisphere biomass burning regions and over particular high population density zones of countries with developing economies.  The model exhibiting the least light absorption with SSA = 0.95 was found in the post-industrial countries of the northern hemisphere. The remainder of the globe was assigned a model with a moderate level of light absorption (SSA = 0.90). These three models with dominant fine modes were matched within the retrieval with the one model with dominant coarse mode. As with the over ocean algorithm, the inversion mixed the two models to find a solution that best matched the measured top-of-atmosphere reflectances.  The same coarse-dominant model was applied globally.  In regions with significant coarse mode dust, the inversion uses more of the available coarse-dominant model, and regions without dust would minimize the amount of coarse-dominant model mixed into the retrieval. Details of the models and the algorithm are found in [18] and [19].

Line 810. ‘validation on a global basis’

Line 973. Now has ‘on’

Line 1101 Results. This section describes the main impacts of the DT algorithm

on the science community and on our understanding of global aerosols. The

heading ‘Results’ usually follows a Methods section where an experiment or

hypothesis for testing is set up. Therefore a more precise section heading such as

‘Major impacts of the DT algorithm’ is suggested

We were trying to adhere to the journal’s prescribed format, as best we could.  We prefer greatly the reviewer’s suggestion and hope that the journal allows for this flexibility.

Line 1191:  7. Major impacts of the Dark Target algorithm

Line 1185 ‘databases’ is one word

Line 1381 ‘databases’ is now one word