Using Hyperspectral Remote Sensing to Monitor Water Quality in Drinking Water Reservoirs
Round 1
Reviewer 1 Report
Manuscript Title
Using hyperspectral remote sensing to monitor water quality in drinking water reservoirs.
Authors
Clémence Goyens , Héloise Lavigne, Antoine Dille and Han Vervaeren
Review
Significance and innovative elements
The manuscript present the possible use of an automated ocean color monitoring system that was developed in a project. The system is another set of sensor systems that add to the off the shelf suite of radiometric systems among other prototypes. The application of the near-realtime use of the sensors is not well reported or maybe not operational but is hinted in the manuscript. One missed opportunity in the manuscript was to reported on some regional or international observatories or similar systems which could add value to the work e.g. NIOZ Texel has a times series platform, U Victoria, Canada has one system on ferry system, Germany several institutes have such a system, LOV likely has one and Rio de la Plata should have something similar. As much as the authors present the advanced system it would be useful to put that into context and state related works.
Main critical issue
There are several issues that need to resolved before the manuscript can be published.
Authors are recommended to only to respond to the suggested edits but thrive to implement the changes of concerns in the manuscript.
1. The text needs some polishing to improve flow and clarity.
-mix of past and present tenses
2. The hyperlinks seem to be down at the time of the review kindly check this and update accordingly.
-maybe provide more details about the sensor system briefly in the manuscript
-does it have both radiance and irradiance sensors?
-why have a camera system and not use the images nothing is discussed or reported about the output from the camera
3. The referencing styles seems to be mixed it would be good to use a uniform format.
4. The discussion present the online system but this was not well presented
-is the system online already?
-is the platform open or limited access?
-make this clear that it is a future direction or this is already online?
5. The figures seem to small considering the journal has no restrictions on size.
-the fonts in some figures also is hard to read
-some figures were crowded maybe change the symbols used
-contrast in figure could be improved or brightness improved
-in the caption and figure indicate the number or order eg 1a 1b etc
6. Testing the two algorithms for chl-a was not well justified, maybe present only the best performing algorithm.
-why test the two only and not other algorithms e.g. several in the GIOP set of algorithms.
7. Derivation of reflectance was not well explained and the missing point is why use Lookup Tables from reference 20 when the authors have developed own approach see reference 26 which also uses wind speed information.
-other approaches have been reported e.g. (Gould et al., 2001; Lee et al., 2010) why not explore these?
-it is not very clear why the Filters are used because there is no clear evidence or studies to fully support the use of the threshold nor have the authors done a detailed intercomparison
- the glint correction applied might not be appropriate since the authors state in Line 319 ‘spectra showing (non-physical) negative values between 400’ in indicator of overcorrection
-is a sequence of scans = to a single measurement? Clarify this better as in the discussion it becomes confusing
8. In the work, the values for variables or parameters used are missing
-what threshold was used or what value of that parameter was used A= 0.01 would be good to have in the manuscript for future studies to be able to replicate the work with minimal effort.
-some of the referenced works are not open-access which makes it hard to follow or understand the authors or verify the approach
9. Algae imaging was done but no image or possible identification of the species was reported so why was this presented and done?
10. It is stated that in Line 226 a hyperspectral algoritm is used and yet in line 236 it states ‘Chlorophyll a Retrieval using an Adaptive Two-band’
11. statistical metrics used are fine but an R² could be useful be it from linear regression or spearmann/pearson
Line by Line Comments
Line 4
What ‘strategies’ do the authors mean ‘monitoring’ or ‘cleaning’
Line 4-7
Revise the text not clear what the authors mean here.
Line 7
-revise ‘products’ with e.g. end-products or parameters or variables
Line 10
-What do the authors mean here ‘are processed to these two’, maybe revise
Line 14
-had to google the mean of grab sampling, maybe consider another term of at least define this for broad readership
Line 19-22
-sentence is too long can it be spilt for made concise
Line 22
-is there really an urgent need?
Line 24
-it could be useful to explain grab sampling and/or provide a reference to literature
Line 25
-What does coarse mean exactly can the authors be more specific?
Line 28
-What does coarse mean exactly can the authors be more specific?
Line 23
-‘…..aquatic biogeochemical processes.’ Any citations here?
Line 35-47
-what is the message here?
-reference to CEOS and PACE related papers could improve the message
Line 43
-could the recent works be relevant have a look at e.g. (Bracher et al., 2017; Dierssen et al., 2020)
Line 50
-who is ‘Their’
Line 61
-where is Blankaart? Which country can this be included for clarity
Line 69
-do the authors mean freshwater or drinking water?
Line 73
-‘ measurements may help in assessing’ would this not be best ‘continuously’
Line 74
-Why these two parameters? What is the relevance?
Line 96
-was installed at Belgian Blankaart WPC
-link not working ‘We can’t connect to the server at www.hypstar.eu.’
Line 99
-What is LT? is this different from UTC?
Line 101
-this is a bit confusing of ‘0.5 (3) nm’
Line 102-104
-revise or rephrase this is not clear
-‘ and embedded camera’ the purpose of the camera is not discussed nor any image for the camera system was presented
Figure 1
-coordinates might be useful in the map
-the use of two satellite images is not well explained why not use the GeoEye image to present message and crop to region of interest panchromatic or RGB
-using the GeoEye then mark the location of the sensor and maybe orientation
-components of the sensor system must also be highlighted since hyperlink is down
-why GeoEye and not other sensors, any high resolution image can be used so why these specific dates what is the message
Line 118-184
-revise the text is confusing a flowchart might be simpler to understand
-what does this mean’ every 15 a 30 minutes’
-for further clarity author add the zenith and nadir angles for viewing
-what are the blacks? Does the sensor have dark correction integrated or?
-is there some open-access component to the processing/processor RHYMER?
-can the threshold values be enumerated? Was it 1 or 10 ?
Line 193
- The supernatant was…
Line 200-205
-provide the sources as reference
Line 226
Equation 3 and 4
-maybe provide superscript terms to differentiate them
Line 256
-so what values was actually used in the analyses ‘l2 is chosen in the 704-740 nm interval.’
Equation 5 and 6, 9
-what values were used for A and B
- HYPSTAR retrieved Chla, ChlHa , can these be define as simis and crat or some format consistent in the whole manuscript
-the values are not provided for equation 9
Line 274 and 284
-what value for ‘with a constant a_ ph(l)’
- AN2010 and CN2010 why not state the actual values here since paper is not open-access?
Line 287-290
-why, give a reason to support this text
-what is sufficient?
-is it C12 or CI2 subscript the numbers or
Equation 11
-Check the equation and also add brackets for the last term
Figure 2
-a bit crowded maybe find other ways to present or make it bigger or split into seasons
-text no clear
Figure 3
-four subplot 2x2 for clarity and bigger font
-why are there strange peaks at round 680 and 770nm? Can these be explained?
Line 334
-red-edge feature
-why not also do an NDVI analyses as simpler quantitative approach
Line 347
-not sure this is evident in Figure 4 highlight for better visualization
-also the relationship between spectra and Chl-a is not easy to see maybe subplot or other presentation approach
Line 360-362
-the statement is not supported here, did the authors apply any statistical analyse to confirm?
What do the authors mean here ‘are processed to these two’, maybe revise
Line 373
-Which plot a or b? (i.e., see points below grey dashed line in Fig. 6).
Line 379
-where are these parameters coming from A and B and what are the values? Equation 5?
Line 384
-why not provide the range min-max values of the ‘variable a_ phy(l)’
Figure 6, 8
-the grey lines not easy to see in printed copy also chose other colours or line styles for easy visuals
Figure 7
-challenging to understand revise for better visualization
-why the specific date ‘Orange vertical bars points to specific dates, i.e., 2021-03-06, 2022-02-07, 2022-01-10, and, 2022-07-17.’
Figure 9
-the dot are too large and make the figure hard to follow or understand
Figure 10
-satellite imagery is relatively dark improve brightness and make large also crop to area of interest
-plot with spectra font size is hard to read colors and line width can be improved
-highlight the features in the satellite imagery for clarity
-why not present the TSM or turbidity maps from the applied ACOLITE processing
Table 1-what is the message exactly?
Line 419
Where are the gustwinds in Fig 10? Highlight them
Figure 11
-y-axis label, limits need to be updated and font
-derivative plot add the vertical lines or points to highlight the peaks or dips of interest to the CI algorithms
Line 459 paragraph
-it is not very clear what the message is
-the authors could have explored the various glint correction approaches why was this not done?
Line 477
-does production mean treatment or cleaning?
Line 484
-how large should the number of datapoint be to be sufficient?
-are the authors or the authorities making an effort to increase data points or what is the best way forward as a recommendation by the authors?
Line 491 paragraph
-for an online system that means data is accessible from source to end-user and is this the case here?
Line 500
-so what is the way forward to improve atmospheric correction? Is this the Hypernets goal or any examples of ways to advance in this direction?
Line 543
-is this correct….check?
References
Bracher, A., Bouman, H. A., Brewin, R. J. W., Bricaud, A., Brotas, V., Ciotti, A. M., Clementson, L., Devred, E., Di Cicco, A., Dutkiewicz, S., Hardman-Mountford, N. J., Hickman, A. E., Hieronymi, M., Hirata, T., Losa, S. N., Mouw, C. B., Organelli, E., Raitsos, D. E., Uitz, J., Vogt, M., and Wolanin, A. (2017) Obtaining phytoplankton diversity from ocean color: A scientific roadmap for future development. Frontiers in Marine Science, v. 4, p. 55(51-15), doi:10.3389/fmars.2017.00055.
Dierssen, H., Bracher, A., Brando, V., Loisel, H., and Ruddick, K. (2020) Data needs for hyperspectral detection of algal diversity across the globe. Oceanography, v. 33, no. 1, p. 74-79, doi:10.5670/oceanog.2020.111.
Gould, R. W., Arnone, R. A., and Sydor, M. (2001) Absorption, scattering, and, remote-sensing reflectance relationships in coastal waters: Testing a new inversion algorithm. Journal of Coastal Research, v. 17, no. 2, p. 328-341,
Lee, Z., Ahn, Y.-H., Mobley, C., and Arnone, R. (2010) Removal of surface-reflected light for the measurement of remote-sensing reflectance from an above-surface platform. Optics Express, v. 18, no. 25, p. 26313-26324, doi:10.1364/OE.18.026313.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Summary:
HYPSTAR system installed at a drinking water reservoir is used to derive water quality parameters, Chlorophyll-a and SPM over a period of 2 years. Collocated grab water samples were processed to validate the performance of the HYPSTAR derived water quality parameters. The study is interesting in terms of using hyperspectral radiometers to derive water quality parameters frequently. Authors presented the most of the details corresponding to data collection, sensor information, procedures adopted, performance of the algorithms and discussed various reasons leading to the observed water quality parameter variation. While most of the information is clear, few major comments (see below) need to be addressed before publication of the manuscript.
Minor comments:
Line 3: Do the authors mean water treatment?
Figure 1: Add a subsection in materials and methods corresponding to satellite images from Landsat 8, Sentinel-2, GEOEye and ACOLITE. Reference to ACOLITE is not provided until discussion section. Also provide the dates and times of image acquisition.
Line 187: Are the grab samples collected during the same period as 2021-01-27 to 2022-08-03 or limited to one of the four seasons mentioned in Figure 3?
Line 297: Provide reference to calculation of as in Eq.2.3.1 or if any other method is used.
Line 313: Follow the same date format in all figures and in all text. Figure 2 has different date format compared to Figure 7 or 8.
Line 322: Please remove “Obviously”
Line 326: Figure 2 caption indicates four subparts whereas the legend shows only three. Either change the legend or the caption. Further, the caption indicates four checks, whereas the text in lines 312 – 326 and before indicates three. Please modify accordingly.
Figure 5 shows that more than half of the measured Chl-a fall below 10 and are not included in statistics calculation for CRAT. Indicate this information in the caption. Are these 11 points used in calculation of statistics for SIMIS as well? Indicate the N value in the figure. Are the blue and red lines fitted excluding or including the Chl-a data below 10 ?
Line 369: rephrase this sentence
Line 378: What does ‘8’ mean here? Stations? A*ph?
Figure 6: What does double star indicate? Eq. 13 indicates that ** correspond to power, please specify this in the caption of Figure 6. Also, change the color of Paavel or make it a dashed line. What does a dashed line in right figure indicate? If this refers to line 381, the dashed line should be at 10 .
Line 405: Is it 2021-05-05? Figure 8 shows only data from year 2021, whereas the text refers to 2022, please clarify.
Line 406: there is no early April data in Figure 8, what are the authors referring to? Is it, early June?( Figure 8 right says so).
Figure 8: Can the authors also present phycocyanin concentrations along with Chl a in left figure? May be on secondary y-axis?
Line 415: Throughout the manuscript the sentences such as “Figure x shows..” can be avoided, instead the figures can be referred within brackets at the end of the sentences.
Line 417: the sentence indicates its Landsat-8, whereas title of the middle figure in Figure 10 says its L9, please check.
Line 420: Is it Fig. 9? There is no gust wind data in Fig 10.
Table 1: Provide grab sample measured Chl-a concentrations if they exist for these dates.
Line 432: Add information about PANTHYR data in materials and methods. Is it installed next to HYPSTAR? Does spectra in Figure 11 correspond to PANTHYR sensor? How many reflectance spectra were collected using PANTHYR? What is the wavelength range? What are the dates of acquisition? Is the Phycocyanin feasibility study performed with PANTHYR sensor data only? If so, HYPSTAR data is not used and parts of the manuscript should be modified accordingly.
Figure 11-> Show a line at 600 nm spectral band and it seems like the second derivative is positive for all reflectance spectra at 600 nm. Is it 600 nm? Or are the authors referring to any other wavelength?
Line 478: The range of Chl-a specified here for CRAT is different from the range specified in Line 351.
Major comments:
If the phycocyanin feasibility study is carried out with PANTHYR data, it should be clearly specified in the abstract. It is unclear if HYPSTAR data is used for the phycocyanin feasibility study at all. It is also unclear in subsection 2.3.3.
Until subsection 3.4, the authors mentioned about HYPSTAR only. There is no information provided about PANTHYR and the years of acquisition (2019 and 2020), which do not lie in the time range of the data presented before (2021 and 2022).
Line 454: The first paragraph of discussion says HYPSTAR is usable for detection of algal blooms, whereas the phycocyanin feasibility study doesn’t provide information if HYPSTAR is used. Altogether, the subsection corresponding to phycocyanin feasibility study is confusing.
Line 481: if HYPSTAR data is not used in testing CI’s, how can it be proven to be useful for detecting cyanobacteria?
Grammatical Corrections
Line 65: Change ‘there’ to ‘their’
Line 190: change ‘0,8’ to ‘0.8’
Line 383: Change a*phy to a*ph
Figure 9: correct Timeseries spelling.
Comments for author File: Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
The authors have addressed all my comments and the manuscript can be accepted in the present form.
