Mid-Long-Term Prediction of Surface Seawater Organic Carbon in the Southern South China Sea Based on Multi-Applicability CNN-LSTM Prediction Model
Round 1
Reviewer 1 Report
General Comments：
This manuscript(ms) applied a CNN-LSTM deep learning model to make a 5-year mid-long term rolling prediction of particulate organic carbon (POC) and yellow matter (CDOM) using MODIS Level 2 semimonthly synthetic data from January 2002 to June 2020 in the southern South China Sea (sSCS). The ms’s results showed that the predicted and actual values of sea surface POC in 2015-2020 would show an overall upward trend, but those values of sea surface CDOM in the same period showed an overall downward trend. However, the mid-long term prediction of surface seawater organic carbon will consider not only in the land-ocean interface that originate from the decomposition of living organisms but also global climate change, e.g. ENSO etc. The effects of climate change on the sSCS were mainly through changes in the monsoon winds, and physical-biological oceanography coupling processes. It would be necessary to consider the relative parameters both the monsoon winds, and physical-biological oceanography coupling processes in this ms.
Detailed Comments:
(1) Line 26-28, please add a couple of sentences to explain further deeply mechanisms to drive the different long-term trend (upward trend for POC and downward trend for CDOM).
(2) I don’t know whether you can give the POC and DOC fluxes in this study. The results of POC fluxes are very important in evaluating the efficiency of biological pump.
(3) Line 206. Standard units for pCO2 should be given in µatm.
(4) Line 171-173. Please explain why you choose this tropical oligotrophic marginal reef waters to be stimulated but others. I think the concentrations of DOC/POC in coastal areas is much higher than the oligotrophic basin. It makes more sense to stimulate in coastal areas. Anyway, I'm not sure whether this model can be used in coastal regions or all the nSCS.
(5) Figure 2 shows the correlations of indicators. What does this picture tell validating the ms’s model or need to independence test for them?
(6）Line 421-427, please further express why you conclude these results both of the predicted and actual values of sea surface POC in 2015-2020 in an overall upward trend and sea surface CDOM in the same period in an overall downward trend when these results derived from the no statistically significant (P < 1-α) under M-K test or the least squares test.
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
In the manuscript, the authors report their analyses regarding the possibility of predicting the carbon cycle in nature. In the case of this work, it concerns the surface waters of the South China Sea. The analyzes are based on satellite data for long term. The authors use the method of deep machine learning in an attempt to determine the possibility of predicting a multi-year trend of changes in the carbon content of the sea. The justification for conducting work in the area of the carbon cycle in sea water is based on an extensive literature analysis of this issue. The manuscript is newsworthy. I think that minor adjustments can be made in the final production process of the article. For example, the appearance of formula (1). Or the data look at the bottom of Figure 3. The authors structured their statement well, taking care of relatively easy perception of the information contained in the manuscript. The statements in the summary seem very bold to me (maybe too much), but in my opinion they do not go beyond what the authors are allowed to say about the value of the results obtained.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Reviewer 3 Report
The study represents a very urgent direction of the modern biogeochemistry concerning the quantification and prediction of the carbon cycle in the Ocean. Particularly, the water area of South China Sea is examined. The authors test the model originally developed for mid-long term rolling prediction of Chl behavior (MODIS level 2) in terms of CDOM and POC as major pools of organic carbon in the water column ocean system, dissolved and particulate respectively. The results obtained through numerical modelling, indicate the opposite trends for CDOM and POC, although both of these parameters originally showed a correlation to the Chl a. The predicted and actual values (2015-2020) were shown to be significantly correlated for a shorter time prediction (2-3 years).The minor corrections are indicated in the pdf file attached
Comments for author File: 
Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
No further comments for this revised version.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
