Extracting Remotely Sensed Water Quality Parameters from Shallow Intertidal Estuaries

Round 1

Reviewer 1 Report

Report referring to the manuscript “Extracting remotely-sensed water quality parameters from shallow intertidal estuaries”

1)      As a general point of view, the manuscript is relevant for the field of remote sensing. It is presented in a well-structured manner. In particular the discussion is well documented and honest. It describes the strengths and weaknesses of the method objectively. However, the presentation of the different techniques used sometimes lacks clarity or suffers from gaps.

Major recommendations:

A reference is missing to justify Eq. 2.

In Table 1 the standard deviation of the seabed optical properties over three years would be more relevant than the coefficients a1 and a2 for each band.

Line 302 concerns the temporal method and line 347 the spatial method: How are obtained the standard deviations. For a good understanding of the method, they should be expressed explicitly.

Minor recommendations:

The following writing is clearer:

Eq. 1: ? = ?? (1- ?^−2???) + ???^−2??? = (?? − ??) ?^−2??? + ??

Figure 1 is not clear. Where the dashed line is? The colors of the points are not distinguishable. Where is the transect?

3.2.2. replace Correction with temporal method with Correction with spatial method

Fig. 9, 10: a, b, c, d is missing

Figure 5: what is the meaning of coordinates?

2)      The cited references are mostly from recent publications and relevant. They do not include an excessive number of self-citations.

3)      The manuscript is scientifically sound, and the experimental design is appropriate, allowing the reader to test the hypothesis.

4)      The manuscript’s results are reproducible, based on the details given in the methods section, allowing the reader to reproduce the results. This comment also concerns the annexes.

5)      The figures, tables and schemes are appropriate, and they are easy to interpret and understand. The data is interpreted appropriately and consistently throughout the manuscript.

6)      The conclusions, very synthetic, are consistent with the evidence and arguments presented.

7)      Data availability statements are adequate.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

In this work, the authors apply Lyzenga's methodology to improve (especially for large-scale applications) the optical estimations (from satellite images, for example, Sentinel-2) of water-quality properties in shallow estuarine water, and specifically, the dominant wavelength and diffuse attenuation coefficient, through the dominant wavelength of the water-reflectance (after removing the seabed reflectance).

1) It is mentioned in the methodology that "if Rb is known apriori and multiple observations of R are available from different total water depths, we show that Lyzenga’s model can be used to estimate the likely values of the remaining two parameters Kd and Rw.". Please further explain how different values of the reflectance R can be available from in-field observations (for validation and confirmation purposes). Also, please rephrase the sentence "likely values" (maybe to simply "values"; I do not understand the purpose of the word "likely" there).

2) For the validation of Lyzenga’s model, the authors mention that it is difficult to obtain Kd observations to validate the results, and they use the images from 3 adjacent days. However, it seems that the selection of these 3 days is somehow arbitrary. Please consider further explaining how a selection of different and additional days could affect the validation procedure and change the estimation of the Kd and Rw. For example, is it possible that the seabed properties change during different seasons or in the long-term, and how would this affect the proposed model and parameter-estimation from this study.

3) Regarding the reflected radiance, which is a function of the seabed properties such as vegetation, the authors mention that it may cause biases (i.e., non-water pixels) in remote sensors. However, there are studies in the literature that tackle this issue by adopting the so-called leaf-are index to identify the vegetation in a satellite image (e.g., see a recent review and applications in Liu et al., 2022). Please consider discussing such methods to overcome the aforementioned limitations.

Tian Liu, Huaan Jin, Ainong Li, Hongliang Fang, Dandan Wei, Xinyao Xie, and Xi Nan, Estimation of Vegetation Leaf-Area-Index Dynamics from Multiple Satellite Products through Deep-Learning Method, Remote Sens., 14(19), 4733, https://doi.org/10.3390/rs14194733, 2022.

4) The water-related processes in the atmospheric cycle (including the ones related to water-quality, such as the ones studied here) are known to be impacted by large uncertainty, and specifically by the so-called long-range dependence, which described how spatial and temporal clusters (that are physically being formed in the water-cycle processes; see a recent review in Dimitriadis et al., 2021) increase the variability of the water-processes in both short- and long-term horizon, and thus, increase the uncertainty in any estimation (for example, consider the clusters in the vegetation described in Liu et al., 2022). If observations are not available in a long-term horizon, this variability/uncertainty can be estimated by performing a sensitivity analysis in the models by selecting a range of the input parameters and then by estimating the variance of the output values. Please consider discussing this issue, and how changes/variability/uncertainty in the water-cycle could affect the findings of this study.

P. Dimitriadis, D. Koutsoyiannis, T. Iliopoulou, and P. Papanicolaou, A global-scale investigation of stochastic similarities in marginal distribution and dependence structure of key hydrological-cycle processes, Hydrology, 8 (2), 59, doi:10.3390/hydrology8020059, 2021.

5) In the analysis, it is mentioned that "Pixels were classified as exposed or inundated using normalized deviation water index (NDWI) calculated with Band 3 (Green) and Band 8 (NIR) from the Sentinel-2 data with the threshold of 0.3.", and that "We also show that the proximity restriction for the reflectance correction with the temporal method limits outcomes to monthly or seasonal resolution and the correction with the spatial method performs the best at a spatial resolution of 60 m."

Please further explain how the authors selected these classifications, thresholds, and spatial resolution since it is not very clear in the text.

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 3 Report

This paper presents a method to obtain the watercolor and Kd of shallow intertidal estuaries based on Lyzenga’s model in Tauranga Harbour. It is an interesting work. The influence of seabed reflectance could be reduced to derive more realistic water qualtity feature. Some comments are listed below.

1. For the readers to understand, it is suggested to provide more detailed information about the mothod and result. For example, providing the specific calulation of STD in spatial and tempal methods, typical scatterplot for Rb regression related to Tab.1, the cite source or the process of the median particle size map in Fig 6, the original four R images used in Fig.7.

2. In Tab. 2 and 3, it is suggested to provide the STD and n information for each band.

3. In section 3.3, it is suggested to provide the similar plots for Kd at 3 bands as dominant wavelength for shallow and deep waters.

4. In result validation, it is suggested to provide possible ways of quantitative assessments, directly or indirectly.

5. Text error of the section heading 3.2.2, should be 'Correction with spatial method?

Author Response

Please see the attachment.

Author Response File: Author Response.docx