The Use of Multisource Optical Sensors to Study Phytoplankton Spatio-Temporal Variation in a Shallow Turbid Lake
, Monica Pinardi 1, Gary Free 1, Giulia Luciani 1, Semhar Ghebrehiwot 2, Marnix Laanen 2, Steef Peters 2, Valentina Della Bella 3,4, Rosalba Padula 3,4 and Claudia Giardino 1
Round 1
Reviewer 1 Report
The manuscript titled, “The use of multisource optical sensors to study phytoplankton spatio-temporal variations in a shallow turbid lake” by Bresciani et al. presents a detailed approach to investigate spatiotemporal phytoplankton dynamics in Lake Trasimeno (Italy) for the year 2018. This study utilizes the advantages of current satellite sensors (e.g., moderate to fine spatial resolution) and a suite of field sensors at a point location (e.g., fine temporal resolution) to study spatial and temporal (inter-, intra-daily, and seasonal) variability of Chlorophyll-a concentration in the lake. Furthermore, authors have also provided the annual and summer maps of spatial variations in Chl-a over the period in Lake Trasimeno in addition to the policy level WFD class map. In my opinion, the key results include 1) good correlations between the WISP-, Sentinel-derived, and in-situ Chl-a concentration, and 2) the western part of Lake Trasimeno had generally poor water quality (based on the spatiotemporal distribution of Chl-a concentration) than the eastern part in 2018. The approach and results are scientifically sound and could be of a great interest to many researchers working in lacustrine regions/on such a multi-layer approach. Manuscript is well-written with minor errors. I would recommend this manuscript for publication in Water.
Minor comments/suggestions:
Line 52: “…such as light intensity, water temperature, wind, food, sex and size [15,16].” I didn’t get how “sex” will affect diurnal rhythms of phytoplankton activity. First of all, can we determine a gender of phytoplankton at all? Perhaps “species” or a similar word would be more appropriate here.
Line 64: Chlorophyll-a fluorescence band is also important in addition to the bands with Chl-a absorption feature.
Line 160: It is not clear that what “standard water quality algorithms” have been used to estimate Chl-a from WISP Rrs measurements. WISP-derived Chl-a is used to show its association with Sentinel-derived Chl-a and to study inter- and intra-daily Chl-a variability in the lake. Thus, please clarify how Chl-a is derived.
Line 217: Please add a reference for NPMR.
Section 2.6 Statistical Analysis: A suite of statistical variables has been used in this study such as CV, MAE, RMSE, and R2 but none is defined in the methodology section. I would suggest adding a table for them under section 2.6.
Figure 2: What does the black line represent? Figure 2a shows hourly data in a black color, whereas Figure 2b shows hourly data in a grey color. Figure 2c and 2d have both colors in the figure. Please include a clear description of lines and points in a figure caption.
Line 279: RMSE should have a unit of “mg m-3”.
Figure 5: What is “Rrs(-)”? Rrs just below the surface? If true then how it has been measured because the WISP is an above water spectroradiometer. Also, Rrs has a unit of sr-1.
Figure 7: What Chl-a values are used to generate the WFD map? An average of 40 chl-a images?
Author Response
REVIEW 1
The manuscript titled, “The use of multisource optical sensors to study phytoplankton spatio-temporal variations in a shallow turbid lake” by Bresciani et al. presents a detailed approach to investigate spatiotemporal phytoplankton dynamics in Lake Trasimeno (Italy) for the year 2018. This study utilizes the advantages of current satellite sensors (e.g., moderate to fine spatial resolution) and a suite of field sensors at a point location (e.g., fine temporal resolution) to study spatial and temporal (inter-, intra-daily, and seasonal) variability of Chlorophyll-a concentration in the lake. Furthermore, authors have also provided the annual and summer maps of spatial variations in Chl-a over the period in Lake Trasimeno in addition to the policy level WFD class map. In my opinion, the key results include 1) good correlations between the WISP-, Sentinel-derived, and in-situ Chl-a concentration, and 2) the western part of Lake Trasimeno had generally poor water quality (based on the spatiotemporal distribution of Chl-a concentration) than the eastern part in 2018. The approach and results are scientifically sound and could be of a great interest to many researchers working in lacustrine regions/on such a multi-layer approach. Manuscript is well-written with minor errors. I would recommend this manuscript for publication in Water.
We would like to thank you for the effort and time spent in providing your expertise to improve the manuscript. We have endeavored to address all of the corrections suggested as best as we can.
Minor comments/suggestions:
Line 52: “…such as light intensity, water temperature, wind, food, sex and size [15,16].” I didn’t get how “sex” will affect diurnal rhythms of phytoplankton activity. First of all, can we determine a gender of phytoplankton at all? Perhaps “species” or a similar word would be more appropriate here.
Answer: Done, we have replaced sex with species (Line 56).
Line 64: Chlorophyll-a fluorescence band is also important in addition to the bands with Chl-a absorption feature.
Answer: Done – we have adjusted the sentence to include a reference related to Chl-a fluorescence band in order to cover this point (Lines 65-68).
“For the last few decades, Chl-a maps retrieved by satellite remote sensing data have been widely used for water quality monitoring [21–24], following different approaches such as bio-optical modeling (e.g. [25,26]), and semi-analytical methods based on band ratios at wavelengths with specific Chl-a absorption features (e.g. [27]) and in some conditions with the use of Chl-a fluorescence band (e.g. 28)”
Line 160: It is not clear that what “standard water quality algorithms” have been used to estimate Chl-a from WISP Rrs measurements. WISP-derived Chl-a is used to show its association with Sentinel-derived Chl-a and to study inter- and intra-daily Chl-a variability in the lake. Thus, please clarify how Chl-a is derived.
Answer: Thanks for this point. We specified the Gons (1999) algorithm (Lines 165-168). The new sentence is: “For water quality monitoring purposes a first estimate of Chl-a was derived through a standard water quality algorithm [47], which make use of a reflectance band ratio at 704 and 672 nm with backscattering derived from the reflectance at 776 nm.”
Line 217: Please add a reference for NPMR.
Answer: Done, this reference has been added to the sentence and also several references covering examples of its use.
Section 2.6 Statistical Analysis: A suite of statistical variables has been used in this study such as CV, MAE, RMSE, and R2 but none is defined in the methodology section. I would suggest adding a table for them under section 2.6.
Answer: We have put in how the CV was calculated, its usefulness and included three references on its calculation and recent use in a similar context for Lake Erie. We have also defined the MAE, RMSE and R2 in the methods section and provided a reference.
Figure 2: What does the black line represent? Figure 2a shows hourly data in a black color, whereas Figure 2b shows hourly data in a grey color. Figure 2c and 2d have both colors in the figure. Please include a clear description of lines and points in a figure caption.
Answer: we added in the text a more clear explanation of the black dashed lines. The rectangle highlights the time of WISPstation acquisition.
Line 279: RMSE should have a unit of “mg m-3”.
Answer: Done
Figure 5: What is “Rrs(-)”? Rrs just below the surface? If true then how it has been measured because the WISP is an above water spectroradiometer. Also, Rrs has a unit of sr-1.
Answer: Thanks for this point. There was a mistake in the figure. Rrs is the standard Remote Sensing Reflectance above the water. We changed the Y axis title, adding the unit of measure (sr-1).
Figure 7: What Chl-a values are used to generate the WFD map? An average of 40 chl-a images?
Answer: We have clarified the approach to calculating the WFD status from Chl-a (Lines 201-213) – explicitly separating it into three steps and detailing the seasons and number of images used:
“The ecological status of the Lake Trasimeno (category: lake mean depth 3-15 m) was classified, in three steps following the national protocol approach applied in [51]. Firstly, average seasonal concentrations were calculated for spring (1 April–15 May; n=7 images), spring–summer transition (15 May–15 June; n=7 images), summer (1 July–31 August; n=11 images), summer–autumn transition (1 September–1 October; n=6 images), autumn (1 October–31 November; n=9 images) and winter (1 January–20 March; n=2 images). Secondly, these seasonal averages were combined to give an annual average. Thirdly the annual average was classified using the published boundaries [52]: high/good = 4.4 mg m-3; good/moderate = 8.0 mg m-3; moderate/poor = 14.6 mg m-3; poor/bad = 26.7 mg m-3. The advantage of using satellite data with this approach is the possibility to have more than one Chl-a value for each season, to give the average of n images grouped per season, instead of one in situ sampling data per season as required by national regulations (D. Lgs 152/2006). This approach can utilize a more robust dataset to classify water quality on an annual basis, in addition to the obviously greater spatial coverage of the classification. ”
Reviewer 2 Report
Review of The use of multisource optical sensors to study phytoplankton spatio-temporal variation in a shallow lake.
The authors evaluated remote sensing methods for measuring chlorophyll a in a large Italian, shallow lake from April 24 to October 3, 2018. They obtained high frequency observations using in situ optical sensors from a WISP station in the lake. They also used 42 images obtained from Sentinel 3-OLCI and Sentinel 2-MSI satellites to map spatial and seasonal change in chlorophyll a. On seven occasions during the study field samples for laboratory analyses of chlorophyll a were collected near the in situ WISP station. Ancillary data were also collected on the study site. They found that the field measurements of chlorophyll a validated the results of both the high frequency in situ WISP station and the products of the Sentinel 3-OLCI and Sentinel 2-MSI satellites. Using nonparametric multiplicative regression, they determined that the “day of the year” was the most important variable related to the phytoplankton population, followed by the east-west wind component. They concluded that the combined approach of field sources, fixed station, and multiple satellites will provide an improvement to ecological assessments of lakes.
Overall, I found no problems with the science used in this study. The results should be useful to others, though I would have liked a discussion of what was unique in this study. What did they find that others had not already found? My comments are minor.
There are many acronyms used throughout the study, and in many cases they are never defined. I would suggest that every acronym be defined the first time it is used. This would make the manuscript much easier to understand.
As a help to others that might want to use the satellite data as they did, I would like to see a discussion of the minimum size of lake where it could be applied. I would like to know if inorganic turbidity interferes with the estimation of chlorophyll a in lakes.
Line 238
Insert a comma after Chl-a
Section 3.1 of Results
For several variables a mean is listed with an error term. Indicate what the error term is (standard error or standard deviation?).
Section 3.2 of Results
I would like to see more information on the statistical tests used to compare the WISP station and satellite data with the manual field measurements. What would be the standard deviations of the differences between estimated and measured chlorophyll values?
Figure 4 legend. I cannot find a definition of “Trewness”. Please explain.
Lines 324-325
I would like to see an illustration of the three types of clusters.
Figure 5
What are the units on the vertical axes?
Figure 6
The font size is too small for the dates above each map. They are very hard to read.
Line 382
Insert comma after time.
Line 385
I can’t see this on Figure 7a. More explanation needed. Are we looking at the green areas?
Line 395
I could barely see the effect on the eastern shore area in Figure 7b. The colors are too similar.
Figure 7
The squares do not show up well. The font sizes are too small.
Line 420
Replace change with chance.
Author Response
REVIEW 2
The authors evaluated remote sensing methods for measuring chlorophyll a in a large Italian, shallow lake from April 24 to October 3, 2018. They obtained high frequency observations using in situ optical sensors from a WISP station in the lake. They also used 42 images obtained from Sentinel 3-OLCI and Sentinel 2-MSI satellites to map spatial and seasonal change in chlorophyll a. On seven occasions during the study field samples for laboratory analyses of chlorophyll a were collected near the in situ WISP station. Ancillary data were also collected on the study site. They found that the field measurements of chlorophyll a validated the results of both the high frequency in situ WISP station and the products of the Sentinel 3-OLCI and Sentinel 2-MSI satellites. Using nonparametric multiplicative regression, they determined that the “day of the year” was the most important variable related to the phytoplankton population, followed by the east-west wind component. They concluded that the combined approach of field sources, fixed station, and multiple satellites will provide an improvement to ecological assessments of lakes.
We would like to thank you for the effort and time spent in providing your expertise to improve the manuscript. We have endeavored to address all of the corrections suggested as best as we can.
Overall, I found no problems with the science used in this study. The results should be useful to others, though I would have liked a discussion of what was unique in this study. What did they find that others had not already found? My comments are minor.
Answer: Thanks for this – while the paper is an addition to the local scientific knowledge of the lake the main contribution is in demonstrating the additional scientific understanding that can be leveraged from combining the latest technologies in satellite and in situ monitoring. We have added a final paragraph to the paper to capture this point (Lines 485-490):
“While the separate use of satellite or in situ methods have benefits, this study demonstrated the synergistic dividend from combining approaches. Combining S3-OLCI and S2-MSI data yielded over 40 images that were able to identify potential pollution events entering the lake while coupling this with the continuous Chl-a measurements of the WISP and weather stations allowed the dynamics and manifestation of bloom events to be predicted with clear benefits for water management.”
There are many acronyms used throughout the study, and in many cases they are never defined. I would suggest that every acronym be defined the first time it is used. This would make the manuscript much easier to understand.
Answer: Done – we checked throughout the manuscript and have inserted abbreviations where needed.
As a help to others that might want to use the satellite data as they did, I would like to see a discussion of the minimum size of lake where it could be applied. I would like to know if inorganic turbidity interferes with the estimation of chlorophyll a in lakes.
Answer: Regarding the question of lake size we have inserted a sentence and a reference on the use Sentinel 2 data in small lakes. (Lines 459-461)
“Sentinel 2 data has been used on lakes as small as 7 ha, but given the resolution of < 60 m the approach can be applied to much smaller lakes [82], and rivers [83].”
Regarding the question on inorganic turbidity –While some algorithms for chlorophyll may be influenced by suspended sediment, especially owing to spectral signal overlap, the use of BOMBER or Gons (1999) as did in this study should minimizes this effect by considering the contribution due to the backscattering of Non Algal Particles.
Line 238: Insert a comma after Chl-a
Answer: Done
Section 3.1 of Results: For several variables a mean is listed with an error term. Indicate what the error term is (standard error or standard deviation?).
Answer: we added in the text that the error reported is standard deviation (Line 275).
Section 3.2 of Results: I would like to see more information on the statistical tests used to compare the WISP station and satellite data with the manual field measurements. What would be the standard deviations of the differences between estimated and measured chlorophyll values?
Answer: We have added a paragraph on the linear regression used to compare the WISP and water sample chlorophyll a (Lines 259-264). We have detailed the error terms and provided a reference for them:
“WISPstation and water sample determined Chl-a were compared by linear regression. While the number of monitoring occasions was only seven, therefore allowing only a limited comparison, the range from <5 to >45 mg m-3 largely represented the seasonal variation in the lake. Reported values included the R2 (coefficient of determination) as well as both the MAE (Mean Absolute Error) and RMSE (Root Mean Squared Error), the former being more easily interpretable and the latter penalizing larger errors ([65]).”
Figure 4 legend. I cannot find a definition of “Trewness”. Please explain.
Answer: we have now listed it in the methods as a smoothing technique and provided a reference to the software. (Lines 251-252).
Lines 324-325: I would like to see an illustration of the three types of clusters.
Answer: Done – as this was also a request from reviewer 3 we have inserted a new figure (Figure 5) to illustrate the general pattern for the three cluster types. However, we would prefer not to include the cluster diagram as it is very large and probably illegible in a publication and in the end not so informative. We include it below for reference but figure 5 we believe is a concise way to present the different responses evident in these three main cluster groups. ATTACHED FIGURE
Figure 5: What are the units on the vertical axes?
Answer: The variable remote sensing reflectance has sr-1 as unit. We added the unit in the figure.
Figure 6: The font size is too small for the dates above each map. They are very hard to read.
Answer: We re-edited the font size of the dates
Line 382: Insert comma after time.
Answer: Done
Line 385: I can’t see this on Figure 7a. More explanation needed. Are we looking at the green areas?
Answer: We changed the range of the color ramp of the map to better identify the areas with the higher CV. However, the reviewer was correct, the green areas were the object of the sentence.
Line 395: I could barely see the effect on the eastern shore area in Figure 7b. The colors are too similar.
Answer: As for the previous map, we changed the range the color ramp of the map to better identify the different CV values.
Figure 7: The squares do not show up well. The font sizes are too small.
Answer: We re-edited the figure, and now the edges of the squares are in black.
Line 420: Replace change with chance
Answer: Done
Author Response File: 
Author Response.pdf
Reviewer 3 Report
Different sources of chlorophyll-a concentration measurements in water are used here to characterized Lake Trasimeno. High frequency data from the WISP instrument is used together with satellite data and manual field measurements. This allows a study of the phytoplankton dynamics with a high temporal and spatial resolution. The paper is quite well structured and written. But there are a few comments or clarifications I would like to point out.
In general, the methodology section (2.6 Statistical Analysis), could be improved. Some important questions that arise:
Why authors use the coefficient of variation for the analysis and comparison? How is this coefficient calculated? What is the basis of this CV for water dynamic studies? Some references could be useful. The NPMR method needs also to be referenced and some examples of its use should be given. I think the authors use the results from the sensitivity analysis of the model to derive conclusions on the weight of the different variables used as inputs on the results, but it is not clear to me how the model is validated. The clustering procedure is neither very clear in its application or objectives. Is this done on the results of the NPMR model? Why is this necessary? The final approach applied (Kruskal-Wallis One Way Analysis of Variance on Ranks) is neither referenced or explain with clarity.
Section 3.2 refers to a scatter plot (Figure3a) with 7 matchups. Authors should reflect on the significance of the statistics derived with this low number of points.
Lines 312-315 repeat what is said in lines 295-298.
Is the cluster analysis made in section 3.3 (lines 323-325) the one specified in section 2.6? The inclusion of a figure showing the results of the clustering approach would increase understanding of the methodology and results.
The word “littoral” could, in some places, be exchange by coastline or coastal.
Author Response
REVIEW 3
Different sources of chlorophyll-a concentration measurements in water are used here to characterized Lake Trasimeno. High frequency data from the WISP instrument is used together with satellite data and manual field measurements. This allows a study of the phytoplankton dynamics with a high temporal and spatial resolution. The paper is quite well structured and written. But there are a few comments or clarifications I would like to point out.
We would like to thank you for the effort and time spent in providing your expertise to improve the manuscript. We have endeavored to address all of the corrections suggested as best as we can.
In general, the methodology section (2.6 Statistical Analysis), could be improved. Some important questions that arise:
Why authors use the coefficient of variation for the analysis and comparison? How is this coefficient calculated? What is the basis of this CV for water dynamic studies? Some references could be useful.
Answer: Thanks for this point – we have put in how the CV was calculated, its usefulness and included three references on its calculation and recent use in a similar context for Lake Erie (Lines 193-196).
The NPMR method needs also to be referenced (1) and some examples of its use should be given (2). I think the authors use the results from the sensitivity analysis of the model to derive conclusions on the weight of the different variables used as inputs on the results, but it is not clear to me how the model is validated (3). The clustering procedure is neither very clear in its application or objectives. Is this done on the results of the NPMR model? Why is this necessary? (4)
Answer:
This reference has been added to the sentence. Thanks for your suggestion we have now given three examples with references on the use of NPMR. We accept the point on the model validation – while we tested the significance of the models using Monte Carlo techniques, detailed in the methods, in the end the model was run on the data available from April until October 2018. A more thorough validation could be possible with additional data. We have inserted a note on validation in the discussion to cover this: “The NPMR model was developed for a limited time-period (April-October 2018). In Lake Trasimeno the timing and slope of increase as well as the date of maximum Chl-a concentration can vary with year, underlining the need for dynamic monitoring that would also provide additional data for more comprehensive model development and validation [43].” Thanks for this, the paragraph on clustering followed the NPMR and there was not a sufficient introductory/linking sentence to explain the rational. We have included this now. “While the environmental factors causing variation in Chl-a over days and months were identified above, it may be more challenging to explain the dynamic changes occurring within a day. In order to examine this diurnal variation of Chl-a we initially examined graphs of the hourly variation.”
(Lines 230-252).
The final approach applied (Kruskal-Wallis One Way Analysis of Variance on Ranks) is neither referenced or explain with clarity.
Answer: Thanks for this point, we have clarified the text and added a reference (Lines 268-269): “In order to test for overall differences among the regions of interest in Chl-a, the non-parametric Kruskal-Wallis One Way Analysis of Variance on Ranks was applied [67,68].”
Section 3.2 refers to a scatter plot (Figure3a) with 7 matchups. Authors should reflect on the significance of the statistics derived with this low number of points.
Answer: We have added a paragraph on the linear regression used in the materials and methods section (Lines 259-264) that reflects this caution in using a low n in a regression:
“WISPstation and water sample determined Chl-a were compared by linear regression. While the number of monitoring occasions was only seven, therefore allowing only a limited comparison, the range from <5 to >45 mg m-3 largely represented the seasonal variation in the lake. Reported values included the R2 (coefficient of determination) as well as both the MAE (Mean Absolute Error) and RMSE (Root Mean Squared Error), the former being more easily interpretable and the latter penalizing larger errors ([65]).”
Lines 312-315 repeat what is said in lines 295-298.
Answer: Done, we have removed the repetition.
Is the cluster analysis made in section 3.3 (lines 323-325) the one specified in section 2.6? The inclusion of a figure showing the results of the clustering approach would increase understanding of the methodology and results.
Answer: Done – as this was also a request from reviewer 2 we have inserted a new figure to illustrate the general pattern for the three cluster types.
However, we would prefer not to include the cluster diagram as it is very large and probably illegible in a publication and in the end not so informative. We include it below for reference but figure 5 we believe is a concise way to present the different responses evident in these three main cluster groups. FIGURE ATTACHED
The word “littoral” could, in some places, be exchange by coastline or coastal
Answer: Done, we have replaced littoral with shoreline
Author Response File: Author Response.pdf
