the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 12 Jul 2022
Surface circulation in the Gulf of Thailand from remotely sensed observations: seasonal and interannual timescales
Abstract. The Gulf of Thailand (GoT), a shallow basin located in the western equatorial Pacific, undergoes highly-variable wind influences on both seasonal and interannual timescales. On seasonal timescale, the Asian monsoon prevails, dominating the circulation pattern. The most dominant complex empirical orthogonal function of the Gulf of Thailand (GoT) current explains 28 % of the total variance with the phase relationship following the monsoon seasons. High current variability is observed over the upper GoT (uGoT), along the western boundary which is highly correlated with that at the southeastern entrance, and in the GoT interior. Satellite altimetry suggests that approximately 50 % of the surface current variability is geostrophic set up by wind stress curl, particularly at basin-scale. Also, the study hints at the importance, to sea surface height near the coastline, of coastal trapped Kelvin waves forced by wind stress curl. Wind-driven current accounts for a significant fraction of the ageostrophic flow along the western boundary and the GoT interior; however, its effect is unclear over the uGoT. At interannual timescale, the GoT circulation is directly impacted by both El Niño Southern Oscillation (ENSO) and the Indian Ocean Dipole (IOD). Interestingly, the climate modes affect the circulation over each region of the GoT differently. The IOD highly correlates with interannual variability of current along the GoT western boundary and the southern boundary of the observing domain (8° N), while the ENSO correlates with that in the interior. Over the uGoT, both climate modes explain similar percentage of the interannual current variability.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(13138 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(13138 KB) - Metadata XML
- BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-495', Anonymous Referee #1, 15 Aug 2022
The manuscript describes the seasonal and interannual surface circulation variability in the Gulf of Thailand (GoT) through the analysis of remote sensed data (i.e., SST, SSH, etc.). Complex EOF and complex correlation and other spatio-temporal analysis were taken into consideration. Since this is, I think the first paper that tries to explain the dynamics of surface circulation in the GoT using observation (remote sensing) data, more detail explanation is necessary. I think not only oceanographers want to know the surface circulation in the GoT, but also other scientists (marine biologists, chemical oceanographers, fishery scientists, etc.). So, my comments will be pointed out in the following.
Major comments:
- Line 16: author defined the GoT domain (6-14 N and 98-106 E), but when author begins to analyze the data, why author defined the analyzed area as 8-14 N (Line 73) not 6-14 N. During SW monsoon, northward flow from East Malaysian Peninsula could also play significant role for the circulation in the GoT as is showed the inflow from South China Sea (SCS) during NE monsoon. Please clarify this.
- Line 78-80: when author mentioned that the largest difference between OSCAR and HF Radar is found at UGoT where there are only 6 OSCAR data points. Does author try to conclude that largest difference found at UGoT is because of there are only 6 OSCAR data points? What about other areas in the GoT?
- Line 83-87: Please clarify sea surface level (is it monthly data? And what is the source of ADT data?). Do you think 3 stations situated in the UGoT and western GoT are enough to validate satellite-derived ADT? What about the stations in the eastern GoT? I think there are available data from local authorities that you can request. When author mentioned western boundary, does it include the western side of UGoT (no data for OSCAR).
- Line 99: How to estimate A?
- Line 111-113: Why high variance of ADT indicating the influence of geostatic flow and low variance indicating ageostrophic flow? Could you give a reference for that statement?
- Line 116: Any references for reader to follow, regarding to CEOF that the author used?
- Line 117: The first 2 modes represent just only 48%. What about other 52%?
- Line 117-119: Without the knowledge background about the GoT dynamics, from fig.3a-c, how do you know that during southwest monsoon, anticyclonic circulation exists at the center of the GoT? Please clarify or give any references.
- Line 125-129: For CEOF2 “Negative phase means southward flow along the western boundary and vice versa for positive phase” How do you know that? What figures tell you that?
- Line 164: Why strong mismatch occurs at UGoT and western boundary?
- Line 239: What is the source of SST? The author didn’t mention it in Datasets Section.
- Line 240: Please give the reference of complex correlation that the author mentioned.
Minor comments:
- Line 21 and others: “Buranatheprat” -> “Buranapratheprat”
- Line 26 and others: I’m not sure how to order the citation, first name or year
- Line 160: Better to start new paragraph after (Kubryakov …..)
- Lines 299 & 304: “farther”
- Line 357: “Copernicus Marine Service Information”
- For the Reference, the journal name is full or short name? I found both full and short name (Be consistency).
- Figure 2: unit of Fig.2a is m or m/s?
- Figure 3: Could not see Marron boxes in (a) and (d).
- Figure 4: better to have a title for colorbar in the figure and have colorbar for every row.
- Figure 7: intext, author use r instead of R as shown in the figure. Just curios! When the author shows the comparison between sea surface height at different areas and negative wind stress curl, which clearly show perfect match but why r in the figure is negative not positive?
- Figure 8: Since color scale in the first row and second row are different, how color contour in (a) & (b) differ from the rest? Are they correlation coefficient? Better to have the title for the colorbar and have colorbar for all rows.
- Figure 9: Better to have no. of data legend equal to the no. of variables. For example, 9(a) has 3 lines, so it should have 3 data legend, velocity, Nino3.4 and DMI.
Citation: https://doi.org/10.5194/egusphere-2022-495-RC1 -
AC1: 'Reply on RC1', Arachaporn Anutaliya, 05 Oct 2022
Major comments:
- Line 16: author defined the GoT domain (6-14 N and 98-106 E), but when author begins to analyze the data, why author defined the analyzed area as 8-14 N (Line 73) not 6-14 N. During SW monsoon, northward flow from East Malaysian Peninsula could also play significant role for the circulation in the GoT as is showed the inflow from South China Sea (SCS) during NE monsoon. Please clarify this.
- Thank the reviewer for pointing it out. The GoT domain is redefined to be between 8-14° N (L15).
- Discussion relating to the influence of the South China Sea and that along the East Malaysian Peninsular is added.
- Line 78-80: when author mentioned that the largest difference between OSCAR and HF Radar is found at UGoT where there are only 6 OSCAR data points. Does author try to conclude that largest difference found at UGoT is because of there are only 6 OSCAR data points? What about other areas in the GoT?
- The large difference between the OSCAR data and the coastal-radar found at the uGoT is essentially due to the spatial resolution of the OSCAR data that is too coarse. As a result, there are only a few measurements in the uGoT regions (only 6). The author understands that the original text is quite misleading and has revised to better explain the point. The available OSCAR data in each defined region is presented in Figure 3; there are 14 data points in the western boundary box and 72 data points in the interior box.
- Line 83-87: Please clarify sea surface level (is it monthly data? And what is the source of ADT data?). Do you think 3 stations situated in the UGoT and western GoT are enough to validate satellite-derived ADT? What about the stations in the eastern GoT? I think there are available data from local authorities that you can request. When author mentioned western boundary, does it include the western side of UGoT (no data for OSCAR).
- The ADT is the gridded all-satellite merged product provided at daily resolution. See Line 97.
- Thanks to the reviewer for the suggestion to also compare the ADT data to the available tide gauge along the eastern boundary. Further ADT validation against four additional tide gauge stations is carried out with the two out of four stations located along the eastern side of the GoT (Laem Sing and Ko Sichang). The calculated correlation coefficients between the ADT and tide gauge data are presented in Figure 1.
- Line 99: How to estimate A?
- The estimation of A is added.
- Line 111-113: Why high variance of ADT indicating the influence of geostatic flow and low variance indicating ageostrophic flow? Could you give a reference for that statement?
- The geostrophic current can be directly estimated from the satellite-derived ADT. The relationship is added to the datasets section and the reference to these equations is inserted in the text. Therefore, the change (hence variance) of ADT directly associates with the geostrophic current. On the contrary, geostrophic current is suggest to not be important when the ADT barely fluctuates (low variance), and thus the ageostrophic component dominates.
- Line 116: Any references for reader to follow, regarding to CEOF that the author used?
- The author thanks the reviewer for the suggestion. A methodology section describing the complex empirical orthogonal function (CEOF) as well as providing the associated literature is added to the manuscript.
- Line 117: The first 2 modes represent just only 48%. What about other 52%?
- The CEOF technique decomposes the velocity vector fields into 173 orthogonal modes (i.e. 173 independent CEOF maps). By the mathematical setup, the first mode explains the biggest portion of the variance while the subsequent modes explain subsequently smaller portion. Therefore, the rest of the variance (58%) is distributed over 171 modes not being considered in the manuscript.
- Line 117-119: Without the knowledge background about the GoT dynamics, from fig.3a-c, how do you know that during southwest monsoon, anticyclonic circulation exists at the center of the GoT? Please clarify or give any references.
- As shown in Figure 3a, there is an anticyclonic circulation in the middle of the GoT basin. To understand when this pattern occurs, the PC has to be considered. Magnitude of the PC indicates the intensity of the pattern and the phase of the PC indicates the rotation that CEOF1 has to rotate. During the southwest monsoon and fall monsoon transition, the phase of PC is 0 indicating no rotation needed. During the northeast monsoon and spring monsoon transition, the phase of PC is +/- pi indicating that the pattern shown in Figure 3a has to be reversed; this yields a cyclonic circulation in the middle of the GoT. The author has added a methodology section to describe the CEOF technique as well as how to interpret the results.
- Line 125-129: For CEOF2 “Negative phase means southward flow along the western boundary and vice versa for positive phase” How do you know that? What figures tell you that?
- The original text is unclear regarding how to interpret results from the CEOF analysis, the author thanks the reviewer for the suggestion. The author has included text describing how the CEOF and PC are interpreted in the methodology section.
- Line 164: Why strong mismatch occurs at UGoT and western boundary?
- The difference between the total and geostrophic flow is simply the ageostrophic flow, i.e., flow driven by wind of affected by friction, etc. In this case, the author speculates that the large difference is likely due to strong total and geostrophic currents at these locations. Therefore, the author has rephrased the sentences to clarify this point.
- Line 239: What is the source of SST? The author didn’t mention it in Datasets Section.
- The ENSO and IOD indices are added in the dataset section. The data sources are provided in the data availability section.
- Line 240: Please give the reference of complex correlation that the author mentioned.
- Complex correlation is introduced, and reference for the calculation is provided.
Minor comments:
- Line 21 and others: “Buranatheprat” -> “Buranapratheprat”
- The author apologizes for the wrong information provided. The text has been updated.
- Line 26 and others: I’m not sure how to order the citation, first name or year
- The citation style in this manuscript follows the Copernicus Publications LaTex Package, version 6.8, 28 March 2022.
- Line 160: Better to start new paragraph after (Kubryakov …..)
- The text has been updated accordingly.
- Lines 299 & 304: “farther”
- The text has been updated.
- Line 357: “Copernicus Marine Service Information”
- The text has been updated.
- For the Reference, the journal name is full or short name? I found both full and short name (Be consistency).
- The reference has been updated to be consistent.
- Figure 2: unit of Fig.2a is m or m/s?
- The author thanks the reviewer for pointing out the mistake. The unit has to be m for (a) and m2 for (b) and the correction has been made.
- Figure 3: Could not see Marron boxes in (a) and (d).
- The maroon boxes are missing from the figure and have now been inserted.
- Figure 4: better to have a title for colorbar in the figure and have colorbar for every row.
- The title has been added to the figure. Since all subplots share the same colorbar, only one colorbar is provided on the lower right corner of the figure. The figure caption has been updated to emphasize this point.
- Figure 7: intext, author use r instead of R as shown in the figure. Just curios! When the author shows the comparison between sea surface height at different areas and negative wind stress curl, which clearly show perfect match but why r in the figure is negative not positive?
- The author has updated the text to use R to represent the correlation coefficient.
- The author thanks the reviewer for pointing the inconsistency. The correlation coefficient shown on the figure is from the correlation between wind stress curl and ADT; therefore, the number is negative. However, the reviewer is correct that presenting the figure and correlation coefficient this way is inconsistent so the figure has been updated.
- Figure 8: Since color scale in the first row and second row are different, how color contour in (a) & (b) differ from the rest? Are they correlation coefficient? Better to have the title for the colorbar and have colorbar for all rows.
- All subplots show correlation maps; therefore, the color contours are all showing correlation coefficients but between different variables. Since the top row ((a) and (b)) is showing correlation with the velocity, complex correlation is applied here. The author understands that the figure caption might not be clear, so the text has been updated to emphasize this point.
- The title is added at the top of the colorbar (upper right corner of the figure).
- Colorbars are added for all the rows.
- Figure 9: Better to have no. of data legend equal to the no. of variables. For example, 9(a) has 3 lines, so it should have 3 data legend, velocity, Nino3.4 and DMI.
The accordant change has been made.
Citation: https://doi.org/10.5194/egusphere-2022-495-AC1
-
RC2: 'Comment on egusphere-2022-495', Anonymous Referee #2, 23 Aug 2022
The paper introduces a discussion about the GoT surface circulation using remotely sensed data. The impacts of monsoon wind as well as those of ENSO and IOD on GoT surface circulation are discussed. Complex EOF and correlation analyses were used to investigate the drivers of surface circulation patterns. Seasonally, Complex EOF showed that about 28% of changes in surface circulation were attributed to the monsoon wind reversal. On interannual timescales, ENSO and IOD have spatially varying impacts on surface circulation with ENSO influence being more pronounced in the GoT interior whereas the GoT western boundary responds more to IOD conditions as evidenced by correlation analysis.
The paper is fairly well written. It introduces an important topic that can benefit a larger community to understand the impacts of climate on surface circulation and the extension to biogeochemical processes. However, there are a number of improvements that can help build a stronger case.
- The significance of the study should be emphasised. Besides the use of field observations and numerical modelling to study the surface circulation of GoT, previous studies also used remotely sensed data but findings did not converge. So, what makes the application of remotely sensed data in this study relevant. Also, most of the remotely sensed data used in this study were available from the 90s but the analysis only focused on the period starting from 2014. Why?
- In addition to datasets section, please also add a methods section. Most of the statistical analysis used were neither thoroughly explained nor proper references were provided. This is a weakness of the current manuscript. Complex correlation, rms correlation, etc. are not common correlation analysis methods and should be well introduced with proper citation prior to their use. Without such information it is difficult for the reader to make an easy interpretation of results presented. As for the datasets, please include the data source URLs in the datasets section and explanation about the ENSO and IOD data, and their provenance.
- SCAR and ADT data with spatial resolution of 27–37 km were used to discuss surface circulation patterns in the uGoT. This is a small (~ 100 km horizontally) and shallow area (Figure 1). The author recognises the limitations (L340) but still places a lot of emphasis on the variability of OSCAR current data in the uGoT (e.g., L120, L135, L184, etc.). I would recommend introducing the L340 text earlier and limit the discussion of uGoT surface circulation based on OSCAR data. Please note that the circulation patterns in the uGoT have been discussed using local wind data derived from meteorological stations as a way to overcome the limitations of coarse resolution remotely sensed wind data (Buranapratheprat et al., 2006).
- The sequence of some figures should be revised to make the flow easier. Complex EOF (Figure 3) can be understood with easy after Figure 4 is introduced, and after the method has also been introduced. Similarly, the correlation maps in Figure 8 are better after Figure 9 is introduced which will be in line with the text in L250.
Details
L30: please revise the text for clarity. Numerical simulations were done for uGoT but the spatially uniform wind does not represent that of GoT?
L38: “The study suggests an overall cyclonic circulation…” Which study?
L55: better open a new paragraph.
L77: It is helpful to add the locations of the high frequency radar system in Figure 1.
L80: please explain how this comparison was done. How the spatial and temporal resolution differences were addressed? As admitted in L340 the OSCAR data may contain large error there. How to distinguish between the two, large difference or large error?
L90: 20 km is below the ADT spatial resolution. How to know this is not noisy data?
L99-100: references for A and wind stress curl estimation?
L115: why use complex EOF? Does it improve the results over the classical EOF. What (additional) information is gained from the use of complex EOF? Again, these issues can be addressed by a proper methods section.
L120: much of this discussion should be removed from the text. 6 grid points are too rough to have a meaningful discussion. Any derived parameter will even include less grid pixels which further limits the interpretations.
L125: briefly explain the reader how to look at the results of complex EOF. For classical EOF the mode often indicates temporal variability when multiplied by the amplitude. What is the relationship between Figure 3a-c?
L151: reference(s) for the correlation coefficient from non-parametric method?
L165: what is rms correlation? reference(s)?
L171: what is total current? Most of these details should be addressed by a proper methods section.
L224: what is the difference between this correlation coefficient and the other so far introduced?
L227: “…much of the correlation is attributed…”
L232: “Still, the result suggests the importance of coastal trapped Kelvin waves…” add reference(s) as this is speculative?
L239: add the description (and/or sources) of the sea surface temperature and the Niño 3.4 box in the text.
L245: as mentioned above, the discussion in this paragraph suggested that the sequence of Figure 8 and 9 is reversed.
L250: “…correlations between low-frequency Niño3.4 and selected forcings…”. What parameter is the forcing, Niño3.4 or ADT, etc.?
L300: “…but might relate to the winter warm pool” where?
L310: the last sentence is long, break it into smaller parts.
Please consider revising the style of figure captions. I think starting with the label followed by the explanation is a commonly used caption style and is easy to read. E.g., Figure 1: Map of the Gulf of Thailand. (a) Triangles indicate the locations of tide gauges: FP denotes Fort Phrachula Chomklao station, KL denotes Ko Lak station, and KM denotes Ko Mattaphon station. (b) Shows the location of the Gulf of Thailand. Colour contours in (a) and (b) represent bathymetry. The black contour in (a) represents the zero-depth level.
Figure 3. I could not locate boxes in Figure 2.
Figure 5. Please use the positive half of the (b) Figure colour palette. On (a) blue corresponds to extreme positive but on (b) the same blue is extreme negative. This can be very confusing.
Figure 6. Same as in Figure 5. Add zero contours in (b) and (d).
Figure 7. sea surface height (black) just needs to be mentioned once. How can a negative wind stress curl have both negative and positive values? I think the point here is that the displayed data was multiplied by (-1) to be in phase with the other parameter? Winds act on a much larger scale, so I am not clear about the point of the correlation between the green cross and triangle.
Figure 8. Non-significant correlation could be masked?
Reference
Buranapratheprat A, Yanagi T, Sojisuporn P, Booncherm C (2006) Influence of local wind field on seasonal circulation in the Upper Gulf of Thailand. Coast Mar Sci 30(1):19–26
Citation: https://doi.org/10.5194/egusphere-2022-495-RC2 -
AC2: 'Reply on RC2', Arachaporn Anutaliya, 05 Oct 2022
The paper introduces a discussion about the GoT surface circulation using remotely sensed data. The impacts of monsoon wind as well as those of ENSO and IOD on GoT surface circulation are discussed. Complex EOF and correlation analyses were used to investigate the drivers of surface circulation patterns. Seasonally, Complex EOF showed that about 28% of changes in surface circulation were attributed to the monsoon wind reversal. On interannual timescales, ENSO and IOD have spatially varying impacts on surface circulation with ENSO influence being more pronounced in the GoT interior whereas the GoT western boundary responds more to IOD conditions as evidenced by correlation analysis.
The paper is fairly well written. It introduces an important topic that can benefit a larger community to understand the impacts of climate on surface circulation and the extension to biogeochemical processes. However, there are a number of improvements that can help build a stronger case.
- The significance of the study should be emphasised. Besides the use of field observations and numerical modelling to study the surface circulation of GoT, previous studies also used remotely sensed data but findings did not converge. So, what makes the application of remotely sensed data in this study relevant. Also, most of the remotely sensed data used in this study were available from the 90s but the analysis only focused on the period starting from 2014. Why?
- This study improves the understanding on the dynamics associated with the GoT circulation that has not been well-studied. It shows the role of wind on modifying both geostrophic and ageostrophic current. In addition, the study provides the variability of the circulation at the interannual timescale. The author tried to emphasize the significance of this study better in abstract, results, and conclusion sections.
- Thanks to the reviewer for the helpful suggestion, discussions of the interaction between GoT circulation and (1) the upper GoT and (2) the SCS are also added to the manuscripts.
- The reviewer is correct that most of the datasets are available since the 90s but under a time constrain of the provided grant, the author can only analyze these few years. The author picked 2014 to 2019 as the studied period because it includes a few major interannual events, such as the 2015/2016 El Nino, the 2016 negative IOD, 2017/2018 La Nina, and the 2019 positive IOD.
- In addition to datasets section, please also add a methods section. Most of the statistical analysis used were neither thoroughly explained nor proper references were provided. This is a weakness of the current manuscript. Complex correlation, rms correlation, etc. are not common correlation analysis methods and should be well introduced with proper citation prior to their use. Without such information it is difficult for the reader to make an easy interpretation of results presented. As for the datasets, please include the data source URLs in the datasets section and explanation about the ENSO and IOD data, and their provenance.
- The author thanks the reviewer for the suggestion. The methodology section is added to describe the complex empirical orthogonal function.
- Complex correlation is introduced, and reference for the calculation is provided.
- The ENSO and IOD indices are added in the dataset section. The data sources are provided in the data availability section.
- OSCAR and ADT data with spatial resolution of 27–37 km were used to discuss surface circulation patterns in the uGoT. This is a small (~ 100 km horizontally) and shallow area (Figure 1). The author recognises the limitations (L340) but still places a lot of emphasis on the variability of OSCAR current data in the uGoT (e.g., L120, L135, L184, etc.). I would recommend introducing the L340 text earlier and limit the discussion of uGoT surface circulation based on OSCAR data. Please note that the circulation patterns in the uGoT have been discussed using local wind data derived from meteorological stations as a way to overcome the limitations of coarse resolution remotely sensed wind data (Buranapratheprat et al., 2006).
- The author thanks the reviewer for the suggestion. Discussion regarding the uGoT circulation, particularly that derived from the OSCAR products, has been limited and the author introduced the limitation regarding the observations over the uGoT in the dataset section (L87) and again at the beginning of the results (Circulation in the Gulf of Thailand) section (L146).
- The introduction has been updated and the reference has been added.
- The sequence of some figures should be revised to make the flow easier. Complex EOF (Figure 3) can be understood with easy after Figure 4 is introduced, and after the method has also been introduced. Similarly, the correlation maps in Figure 8 are better after Figure 9 is introduced which will be in line with the text in L250.
- The author used the CEOF to understand the dominant pattern associated with circulation over the GoT. As a result of the CEOF analysis, the most distinct circulation pattern associates with the seasonal monsoon winds, thus, the monthly circulation is examined in detail. Therefore, the author believes that the original figure sequence provides the more coherent story.
- The methodology is added immediately after the dataset section.
- Similarly, the author used the correlation map to understand which regions are heavily impact by each climate mode. Then, the regions with high impact, i.e., that along the western boundary and that in the GoT interior with 2 sub-regions, are further analyzed to understand how low-frequency variability over these regions change with the climate modes. The author believes that presenting figure 9 before showing figure 8 would cause confusions to the readers.
Details
L30: please revise the text for clarity. Numerical simulations were done for uGoT but the spatially uniform wind does not represent that of GoT?
- The original text pointed out that the study by Yanagi and Takao (1998) indicates that the wind field over the entire GoT (including the uGoT), and thus the use of uniform wind field over the uGoT could yield inaccurate results. The author understands that the original text might provide a misleading detail, and thanks the reviewer for pointing it out. Thus, the text has been updated.
L38: “The study suggests an overall cyclonic circulation…” Which study?
- The author refers to the study by Saramul (2017) which is based on the high spatial-resolution coastal radar mentioned in the previous sentence. The text has been updated to clarify this point.
L55: better open a new paragraph.
- The text has been updated.
L77: It is helpful to add the locations of the high frequency radar system in Figure 1.
- Figure 1 has been updated to include locations of the high frequency coastal radar stations.
L80: please explain how this comparison was done. How the spatial and temporal resolution differences were addressed? As admitted in L340 the OSCAR data may contain large error there. How to distinguish between the two, large difference or large error?
- The monthly mean data as provided by Saramul et al. (2017) is calculated from the OSCAR products for the comparison; the text has been updated to clarify this point. As of the author’s understanding, it is not possible to pinpoint whether the difference is truly a difference or an error. However, given the process associated with estimating the OSCAR products, the derived current along coastal areas could contain high error and a validation is necessary. As there has been no validation of OSCAR products against an observation over the GoT, the comparison to an observation which is the high frequency coastal radar is needed. The results show that the OSCAR products is quite reliable over the GoT, particularly to the south of the uGoT.
L90: 20 km is below the ADT spatial resolution. How to know this is not noisy data?
- The reviewer is correct, noisy data can contribute to a lower correlation coefficient. However, it is difficult to specify which factor contributes to the low correlation at the specific location. The author has updated the text to address this issue.
L99-100: references for A and wind stress curl estimation?
- The estimations of A and wind stress curl are added.
L115: why use complex EOF? Does it improve the results over the classical EOF. What (additional) information is gained from the use of complex EOF? Again, these issues can be addressed by a proper methods section.
- In many studies, complex EOF is applied when the dataset being considered is in vector form. The statement is included in the methodology section added to this manuscript.
L120: much of this discussion should be removed from the text. 6 grid points are too rough to have a meaningful discussion. Any derived parameter will even include less grid pixels which further limits the interpretations.
- The author thanks the reviewer for the suggestion. The discussion has been removed.
L125: briefly explain the reader how to look at the results of complex EOF. For classical EOF the mode often indicates temporal variability when multiplied by the amplitude. What is the relationship between Figure 3a-c?
- A brief explanation regarding how to interpret the results of complex EOF is included at the end of the added methodology section. The CEOF mode (Figure 3a) describes the dominant pattern that accounts for 28% of the total variance of the velocity anomaly over the GoT. Generally, the pattern that is rotate according to the phase of the PC (Figure 3C; positive clockwise), i.e., the pattern displayed in Figure 3a is observed when the phase of PC is zero and the reverse is observed when the phase is ±p. The pattern is present with a great intensity when the magnitude of the PC (Figure 3b) is high.
L151: reference(s) for the correlation coefficient from non-parametric method?
- The reference is added to the manuscript.
L165: what is rms correlation? reference(s)?
- The text refers to root-mean-square of the correlation coefficient. Rms stands for root-mean-square as defined earlier. To compute the rms, a square root of the mean of the squared quantity of interest(correlation coefficient over the interested area in the GoT in this case), i.e. √(mean(R^2)). The rms calculation is commonly used in similar way as calculating average; hence the reference is not necessary. Using the rms instead of the average emphasizes the importance of the variance being explained.
L171: what is total current? Most of these details should be addressed by a proper methods section.
- The author understands the confusion since the original text does not clearly address the different type of currents. A statement has been added in the datasets section to clarify this point.
L224: what is the difference between this correlation coefficient and the other so far introduced?
- The correlation here is calculated between the wind stress curl and sea surface height data, while the previous correlations (subsection 4.1.2 Geostrophic and ageostrophic component) indicate how (1) the geostrophic current and the total current covary (Figure 6a, b) and (2) the ageostrophic current and the wind-driven Ekman current (Figure 6c, d).
L227: “…much of the correlation is attributed…”
- The accordant change has been implemented.
L232: “Still, the result suggests the importance of coastal trapped Kelvin waves…” add reference(s) as this is speculative?
- Coastal trapped Kelvin waves are known to travel equatorward along the western boundary of the basin and poleward along the eastern boundary based on the wave equation. Therefore, the statement regarding the Kelvin wave traveling direction is not speculative. Still, the author agree that an addition of references could be beneficial particularly to readers who are not familiar with the wave.
L239: add the description (and/or sources) of the sea surface temperature and the Niño 3.4 box in the text.
- Descriptions for Niño 3.4 and DMI are added at the end of the datasets section and the corresponding sources are added to the data availability statement.
L245: as mentioned above, the discussion in this paragraph suggested that the sequence of Figure 8 and 9 is reversed.
- The author used the correlation map to understand which regions are heavily impact by each climate mode. Then, the regions with high impact, i.e., that along the western boundary and that in the GoT interior with 2 sub-regions, are further analyzed to understand how low-frequency variability over these regions change with the climate modes. The author believes that the original figure sequence provides the more coherent story. Reversing the order of figure 8 and 9 could confuse the reader regarding the region selected to compare with each climate mode.
L250: “…correlations between low-frequency Niño3.4 and selected forcings…”. What parameter is the forcing, Niño3.4 or ADT, etc.?
- The forcings are ADT, zonal wind stress, and wind stress curl. The author investigated the impact of these forcings on the low-frequency variability of the uGoT circulation.
L300: “…but might relate to the winter warm pool” where?
- The author thanks the reviewer for pointing out the missing detail. The accordant change has been made.
L310: the last sentence is long, break it into smaller parts.
- The accordant change has been made.
Please consider revising the style of figure captions. I think starting with the label followed by the explanation is a commonly used caption style and is easy to read. E.g., Figure 1: Map of the Gulf of Thailand. (a) Triangles indicate the locations of tide gauges: FP denotes Fort Phrachula Chomklao station, KL denotes Ko Lak station, and KM denotes Ko Mattaphon station. (b) Shows the location of the Gulf of Thailand. Colour contours in (a) and (b) represent bathymetry. The black contour in (a) represents the zero-depth level.
- The author thanks the reviewer for the suggestion. The author tried the suggested caption style but cannot successfully change all of them without creating awkward sentences. Therefore, the author kindly asks the reviewer to allow leaving the caption format as is.
Figure 3. I could not locate boxes in Figure 2.
- The boxes are missing from the original figure; the figure is corrected.
Figure 5. Please use the positive half of the (b) Figure colour palette. On (a) blue corresponds to extreme positive but on (b) the same blue is extreme negative. This can be very confusing.
- The author has changed the color contour to avoid the mentioned confusion.
Figure 6. Same as in Figure 5. Add zero contours in (b) and (d).
- The author has changed the color contour and the zero contours are added in (b) and (d).
Figure 7. sea surface height (black) just needs to be mentioned once. How can a negative wind stress curl have both negative and positive values? I think the point here is that the displayed data was multiplied by (-1) to be in phase with the other parameter? Winds act on a much larger scale, so I am not clear about the point of the correlation between the green cross and triangle.
- The figure caption has been updated accordingly.
- The figure has been updated to plot wind stress curl on the second (right) y-axis with the reversed y-axis.
- In many scenarios, winds act on large scale, such as the seasonal current reversal that responds to the monsoon wind as being discussed in this manuscript and change in the warm water volume in the western equatorial Pacific due to the westerly wind bursts related to the development of ENSO events (e.g. Chen et al., 2016). However, winds do not necessarily have large-scale impact on the ocean current, for example, upwelling response to the local alongshore wind. This process is both local and small-scale. Dynamics, particularly those associated with planetary waves, associated with the ocean circulation are also sometimes remote and small-scale, for example, the effect of wind stress curl in the western Pacific can propagate in forms of Rossby waves and modify sea level in the Solomon Sea (Anutaliya et al., 2019). The effect of remotely-force waves on sea level or thermocline variation is commonly found (e.g. Dickinson, 1968; Sprintall et al., 2000; Delman et al., 2016). The author understands that impacts of planetary waves on circulation in the GoT has not been hitherto studied and could be unfaimilar to the reader. Therefore, the author has updated the text to provide some literacy regarding the topic.
- Anutaliya, A., Send, U., Sprintall, J., McClean, J. L., Lankhorst, M., & Koelling, J. (2019). Mooring and seafloor pressure end point measurements at the southern entrance of the Solomon Sea: Subseasonal to interannual flow variability. Journal of Geophysical Research: Oceans, 124, 5085– 5104. https://doi.org/10.1029/2019JC015157
Chen, S., Wu, R., Chen, W., Yu, B. and Cao, X. (2016), Genesis of westerly wind bursts over the equatorial western Pacific during the onset of the strong 2015–2016 El Niño. Atmos. Sci. Lett., 17: 384-391. https://doi.org/10.1002/asl.669
Dickinson, R. E. (1968). Planetary Rossby Waves Propagating Vertically Through Weak Westerly Wind Wave Guides, Journal of Atmospheric Sciences, 25(6), 984-1002.
Figure 8. Non-significant correlation could be masked?
- The figure has been updated accordingly.
Citation: https://doi.org/10.5194/egusphere-2022-495-AC2
-
RC3: 'Comment on egusphere-2022-495', Anonymous Referee #3, 23 Aug 2022
The paper presented the surface circulation in the Gulf of Thailand from remotely sensed observations. There are many interesting study results presented in this paper. I propose to accept this paper for publication with minor revision.
The comments are
1. The GoT is influenced by the SCS. More results related to the influence from the SCS should be mentioned or analyzed.
2. The uGoT region used in this paper should be clearly defined because it is mentioned differently in the paper.
3. CEOF is widely used in the atmospheric and marine scientific research and is also used in the paper, but it is rarely found to be used to study for the GoT. Therefore the CEOF method should be briefly described, although it is well known.
4. There is an error in Figure 1, please revise.
5. Please review the references, for examaple: line 370, Atoms?
6. The validations of the results were still unclear. Please describe the results of the main goal clearly.
Citation: https://doi.org/10.5194/egusphere-2022-495-RC3 -
AC3: 'Reply on RC3', Arachaporn Anutaliya, 05 Oct 2022
The paper presented the surface circulation in the Gulf of Thailand from remotely sensed observations. There are many interesting study results presented in this paper. I propose to accept this paper for publication with minor revision.
The comments are
The GoT is influenced by the SCS. More results related to the influence from the SCS should be mentioned or analyzed.
- More discussions regarding the interaction with the SCS is added in both the introduction and results sections.
The uGoT region used in this paper should be clearly defined because it is mentioned differently in the paper.
- The author thanks the reviewer for pointing this out; the text has been updated to be consistent.
CEOF is widely used in the atmospheric and marine scientific research and is also used in the paper, but it is rarely found to be used to study for the GoT. Therefore the CEOF method should be briefly described, although it is well known.
- The author thanks the reviewer for the suggestion. The methodology section is added to describe the complex empirical orthogonal function.
There is an error in Figure 1, please revise.
- The author cannot locate the error and would greatly appreciate it if the reviewer could provide any detail regarding the error.
Please review the references, for examaple: line 370, Atoms?
- The author apologizes for the wrong information provided. The text has been revised and the reference has been reviewed.
The validations of the results were still unclear. Please describe the results of the main goal clearly.
- This study improves the understanding on the dynamics associated with the GoT circulation that has not been well-studied. It shows the role of wind on modifying both geostrophic and ageostrophic current. In addition, the study provides the variability of the circulation at the interannual timescale. The author tried to emphasize the significance of this study better in both abstract, results, and conclusion sections.
Citation: https://doi.org/10.5194/egusphere-2022-495-AC3 - More discussions regarding the interaction with the SCS is added in both the introduction and results sections.
-
AC3: 'Reply on RC3', Arachaporn Anutaliya, 05 Oct 2022
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-495', Anonymous Referee #1, 15 Aug 2022
The manuscript describes the seasonal and interannual surface circulation variability in the Gulf of Thailand (GoT) through the analysis of remote sensed data (i.e., SST, SSH, etc.). Complex EOF and complex correlation and other spatio-temporal analysis were taken into consideration. Since this is, I think the first paper that tries to explain the dynamics of surface circulation in the GoT using observation (remote sensing) data, more detail explanation is necessary. I think not only oceanographers want to know the surface circulation in the GoT, but also other scientists (marine biologists, chemical oceanographers, fishery scientists, etc.). So, my comments will be pointed out in the following.
Major comments:
- Line 16: author defined the GoT domain (6-14 N and 98-106 E), but when author begins to analyze the data, why author defined the analyzed area as 8-14 N (Line 73) not 6-14 N. During SW monsoon, northward flow from East Malaysian Peninsula could also play significant role for the circulation in the GoT as is showed the inflow from South China Sea (SCS) during NE monsoon. Please clarify this.
- Line 78-80: when author mentioned that the largest difference between OSCAR and HF Radar is found at UGoT where there are only 6 OSCAR data points. Does author try to conclude that largest difference found at UGoT is because of there are only 6 OSCAR data points? What about other areas in the GoT?
- Line 83-87: Please clarify sea surface level (is it monthly data? And what is the source of ADT data?). Do you think 3 stations situated in the UGoT and western GoT are enough to validate satellite-derived ADT? What about the stations in the eastern GoT? I think there are available data from local authorities that you can request. When author mentioned western boundary, does it include the western side of UGoT (no data for OSCAR).
- Line 99: How to estimate A?
- Line 111-113: Why high variance of ADT indicating the influence of geostatic flow and low variance indicating ageostrophic flow? Could you give a reference for that statement?
- Line 116: Any references for reader to follow, regarding to CEOF that the author used?
- Line 117: The first 2 modes represent just only 48%. What about other 52%?
- Line 117-119: Without the knowledge background about the GoT dynamics, from fig.3a-c, how do you know that during southwest monsoon, anticyclonic circulation exists at the center of the GoT? Please clarify or give any references.
- Line 125-129: For CEOF2 “Negative phase means southward flow along the western boundary and vice versa for positive phase” How do you know that? What figures tell you that?
- Line 164: Why strong mismatch occurs at UGoT and western boundary?
- Line 239: What is the source of SST? The author didn’t mention it in Datasets Section.
- Line 240: Please give the reference of complex correlation that the author mentioned.
Minor comments:
- Line 21 and others: “Buranatheprat” -> “Buranapratheprat”
- Line 26 and others: I’m not sure how to order the citation, first name or year
- Line 160: Better to start new paragraph after (Kubryakov …..)
- Lines 299 & 304: “farther”
- Line 357: “Copernicus Marine Service Information”
- For the Reference, the journal name is full or short name? I found both full and short name (Be consistency).
- Figure 2: unit of Fig.2a is m or m/s?
- Figure 3: Could not see Marron boxes in (a) and (d).
- Figure 4: better to have a title for colorbar in the figure and have colorbar for every row.
- Figure 7: intext, author use r instead of R as shown in the figure. Just curios! When the author shows the comparison between sea surface height at different areas and negative wind stress curl, which clearly show perfect match but why r in the figure is negative not positive?
- Figure 8: Since color scale in the first row and second row are different, how color contour in (a) & (b) differ from the rest? Are they correlation coefficient? Better to have the title for the colorbar and have colorbar for all rows.
- Figure 9: Better to have no. of data legend equal to the no. of variables. For example, 9(a) has 3 lines, so it should have 3 data legend, velocity, Nino3.4 and DMI.
Citation: https://doi.org/10.5194/egusphere-2022-495-RC1 -
AC1: 'Reply on RC1', Arachaporn Anutaliya, 05 Oct 2022
Major comments:
- Line 16: author defined the GoT domain (6-14 N and 98-106 E), but when author begins to analyze the data, why author defined the analyzed area as 8-14 N (Line 73) not 6-14 N. During SW monsoon, northward flow from East Malaysian Peninsula could also play significant role for the circulation in the GoT as is showed the inflow from South China Sea (SCS) during NE monsoon. Please clarify this.
- Thank the reviewer for pointing it out. The GoT domain is redefined to be between 8-14° N (L15).
- Discussion relating to the influence of the South China Sea and that along the East Malaysian Peninsular is added.
- Line 78-80: when author mentioned that the largest difference between OSCAR and HF Radar is found at UGoT where there are only 6 OSCAR data points. Does author try to conclude that largest difference found at UGoT is because of there are only 6 OSCAR data points? What about other areas in the GoT?
- The large difference between the OSCAR data and the coastal-radar found at the uGoT is essentially due to the spatial resolution of the OSCAR data that is too coarse. As a result, there are only a few measurements in the uGoT regions (only 6). The author understands that the original text is quite misleading and has revised to better explain the point. The available OSCAR data in each defined region is presented in Figure 3; there are 14 data points in the western boundary box and 72 data points in the interior box.
- Line 83-87: Please clarify sea surface level (is it monthly data? And what is the source of ADT data?). Do you think 3 stations situated in the UGoT and western GoT are enough to validate satellite-derived ADT? What about the stations in the eastern GoT? I think there are available data from local authorities that you can request. When author mentioned western boundary, does it include the western side of UGoT (no data for OSCAR).
- The ADT is the gridded all-satellite merged product provided at daily resolution. See Line 97.
- Thanks to the reviewer for the suggestion to also compare the ADT data to the available tide gauge along the eastern boundary. Further ADT validation against four additional tide gauge stations is carried out with the two out of four stations located along the eastern side of the GoT (Laem Sing and Ko Sichang). The calculated correlation coefficients between the ADT and tide gauge data are presented in Figure 1.
- Line 99: How to estimate A?
- The estimation of A is added.
- Line 111-113: Why high variance of ADT indicating the influence of geostatic flow and low variance indicating ageostrophic flow? Could you give a reference for that statement?
- The geostrophic current can be directly estimated from the satellite-derived ADT. The relationship is added to the datasets section and the reference to these equations is inserted in the text. Therefore, the change (hence variance) of ADT directly associates with the geostrophic current. On the contrary, geostrophic current is suggest to not be important when the ADT barely fluctuates (low variance), and thus the ageostrophic component dominates.
- Line 116: Any references for reader to follow, regarding to CEOF that the author used?
- The author thanks the reviewer for the suggestion. A methodology section describing the complex empirical orthogonal function (CEOF) as well as providing the associated literature is added to the manuscript.
- Line 117: The first 2 modes represent just only 48%. What about other 52%?
- The CEOF technique decomposes the velocity vector fields into 173 orthogonal modes (i.e. 173 independent CEOF maps). By the mathematical setup, the first mode explains the biggest portion of the variance while the subsequent modes explain subsequently smaller portion. Therefore, the rest of the variance (58%) is distributed over 171 modes not being considered in the manuscript.
- Line 117-119: Without the knowledge background about the GoT dynamics, from fig.3a-c, how do you know that during southwest monsoon, anticyclonic circulation exists at the center of the GoT? Please clarify or give any references.
- As shown in Figure 3a, there is an anticyclonic circulation in the middle of the GoT basin. To understand when this pattern occurs, the PC has to be considered. Magnitude of the PC indicates the intensity of the pattern and the phase of the PC indicates the rotation that CEOF1 has to rotate. During the southwest monsoon and fall monsoon transition, the phase of PC is 0 indicating no rotation needed. During the northeast monsoon and spring monsoon transition, the phase of PC is +/- pi indicating that the pattern shown in Figure 3a has to be reversed; this yields a cyclonic circulation in the middle of the GoT. The author has added a methodology section to describe the CEOF technique as well as how to interpret the results.
- Line 125-129: For CEOF2 “Negative phase means southward flow along the western boundary and vice versa for positive phase” How do you know that? What figures tell you that?
- The original text is unclear regarding how to interpret results from the CEOF analysis, the author thanks the reviewer for the suggestion. The author has included text describing how the CEOF and PC are interpreted in the methodology section.
- Line 164: Why strong mismatch occurs at UGoT and western boundary?
- The difference between the total and geostrophic flow is simply the ageostrophic flow, i.e., flow driven by wind of affected by friction, etc. In this case, the author speculates that the large difference is likely due to strong total and geostrophic currents at these locations. Therefore, the author has rephrased the sentences to clarify this point.
- Line 239: What is the source of SST? The author didn’t mention it in Datasets Section.
- The ENSO and IOD indices are added in the dataset section. The data sources are provided in the data availability section.
- Line 240: Please give the reference of complex correlation that the author mentioned.
- Complex correlation is introduced, and reference for the calculation is provided.
Minor comments:
- Line 21 and others: “Buranatheprat” -> “Buranapratheprat”
- The author apologizes for the wrong information provided. The text has been updated.
- Line 26 and others: I’m not sure how to order the citation, first name or year
- The citation style in this manuscript follows the Copernicus Publications LaTex Package, version 6.8, 28 March 2022.
- Line 160: Better to start new paragraph after (Kubryakov …..)
- The text has been updated accordingly.
- Lines 299 & 304: “farther”
- The text has been updated.
- Line 357: “Copernicus Marine Service Information”
- The text has been updated.
- For the Reference, the journal name is full or short name? I found both full and short name (Be consistency).
- The reference has been updated to be consistent.
- Figure 2: unit of Fig.2a is m or m/s?
- The author thanks the reviewer for pointing out the mistake. The unit has to be m for (a) and m2 for (b) and the correction has been made.
- Figure 3: Could not see Marron boxes in (a) and (d).
- The maroon boxes are missing from the figure and have now been inserted.
- Figure 4: better to have a title for colorbar in the figure and have colorbar for every row.
- The title has been added to the figure. Since all subplots share the same colorbar, only one colorbar is provided on the lower right corner of the figure. The figure caption has been updated to emphasize this point.
- Figure 7: intext, author use r instead of R as shown in the figure. Just curios! When the author shows the comparison between sea surface height at different areas and negative wind stress curl, which clearly show perfect match but why r in the figure is negative not positive?
- The author has updated the text to use R to represent the correlation coefficient.
- The author thanks the reviewer for pointing the inconsistency. The correlation coefficient shown on the figure is from the correlation between wind stress curl and ADT; therefore, the number is negative. However, the reviewer is correct that presenting the figure and correlation coefficient this way is inconsistent so the figure has been updated.
- Figure 8: Since color scale in the first row and second row are different, how color contour in (a) & (b) differ from the rest? Are they correlation coefficient? Better to have the title for the colorbar and have colorbar for all rows.
- All subplots show correlation maps; therefore, the color contours are all showing correlation coefficients but between different variables. Since the top row ((a) and (b)) is showing correlation with the velocity, complex correlation is applied here. The author understands that the figure caption might not be clear, so the text has been updated to emphasize this point.
- The title is added at the top of the colorbar (upper right corner of the figure).
- Colorbars are added for all the rows.
- Figure 9: Better to have no. of data legend equal to the no. of variables. For example, 9(a) has 3 lines, so it should have 3 data legend, velocity, Nino3.4 and DMI.
The accordant change has been made.
Citation: https://doi.org/10.5194/egusphere-2022-495-AC1
-
RC2: 'Comment on egusphere-2022-495', Anonymous Referee #2, 23 Aug 2022
The paper introduces a discussion about the GoT surface circulation using remotely sensed data. The impacts of monsoon wind as well as those of ENSO and IOD on GoT surface circulation are discussed. Complex EOF and correlation analyses were used to investigate the drivers of surface circulation patterns. Seasonally, Complex EOF showed that about 28% of changes in surface circulation were attributed to the monsoon wind reversal. On interannual timescales, ENSO and IOD have spatially varying impacts on surface circulation with ENSO influence being more pronounced in the GoT interior whereas the GoT western boundary responds more to IOD conditions as evidenced by correlation analysis.
The paper is fairly well written. It introduces an important topic that can benefit a larger community to understand the impacts of climate on surface circulation and the extension to biogeochemical processes. However, there are a number of improvements that can help build a stronger case.
- The significance of the study should be emphasised. Besides the use of field observations and numerical modelling to study the surface circulation of GoT, previous studies also used remotely sensed data but findings did not converge. So, what makes the application of remotely sensed data in this study relevant. Also, most of the remotely sensed data used in this study were available from the 90s but the analysis only focused on the period starting from 2014. Why?
- In addition to datasets section, please also add a methods section. Most of the statistical analysis used were neither thoroughly explained nor proper references were provided. This is a weakness of the current manuscript. Complex correlation, rms correlation, etc. are not common correlation analysis methods and should be well introduced with proper citation prior to their use. Without such information it is difficult for the reader to make an easy interpretation of results presented. As for the datasets, please include the data source URLs in the datasets section and explanation about the ENSO and IOD data, and their provenance.
- SCAR and ADT data with spatial resolution of 27–37 km were used to discuss surface circulation patterns in the uGoT. This is a small (~ 100 km horizontally) and shallow area (Figure 1). The author recognises the limitations (L340) but still places a lot of emphasis on the variability of OSCAR current data in the uGoT (e.g., L120, L135, L184, etc.). I would recommend introducing the L340 text earlier and limit the discussion of uGoT surface circulation based on OSCAR data. Please note that the circulation patterns in the uGoT have been discussed using local wind data derived from meteorological stations as a way to overcome the limitations of coarse resolution remotely sensed wind data (Buranapratheprat et al., 2006).
- The sequence of some figures should be revised to make the flow easier. Complex EOF (Figure 3) can be understood with easy after Figure 4 is introduced, and after the method has also been introduced. Similarly, the correlation maps in Figure 8 are better after Figure 9 is introduced which will be in line with the text in L250.
Details
L30: please revise the text for clarity. Numerical simulations were done for uGoT but the spatially uniform wind does not represent that of GoT?
L38: “The study suggests an overall cyclonic circulation…” Which study?
L55: better open a new paragraph.
L77: It is helpful to add the locations of the high frequency radar system in Figure 1.
L80: please explain how this comparison was done. How the spatial and temporal resolution differences were addressed? As admitted in L340 the OSCAR data may contain large error there. How to distinguish between the two, large difference or large error?
L90: 20 km is below the ADT spatial resolution. How to know this is not noisy data?
L99-100: references for A and wind stress curl estimation?
L115: why use complex EOF? Does it improve the results over the classical EOF. What (additional) information is gained from the use of complex EOF? Again, these issues can be addressed by a proper methods section.
L120: much of this discussion should be removed from the text. 6 grid points are too rough to have a meaningful discussion. Any derived parameter will even include less grid pixels which further limits the interpretations.
L125: briefly explain the reader how to look at the results of complex EOF. For classical EOF the mode often indicates temporal variability when multiplied by the amplitude. What is the relationship between Figure 3a-c?
L151: reference(s) for the correlation coefficient from non-parametric method?
L165: what is rms correlation? reference(s)?
L171: what is total current? Most of these details should be addressed by a proper methods section.
L224: what is the difference between this correlation coefficient and the other so far introduced?
L227: “…much of the correlation is attributed…”
L232: “Still, the result suggests the importance of coastal trapped Kelvin waves…” add reference(s) as this is speculative?
L239: add the description (and/or sources) of the sea surface temperature and the Niño 3.4 box in the text.
L245: as mentioned above, the discussion in this paragraph suggested that the sequence of Figure 8 and 9 is reversed.
L250: “…correlations between low-frequency Niño3.4 and selected forcings…”. What parameter is the forcing, Niño3.4 or ADT, etc.?
L300: “…but might relate to the winter warm pool” where?
L310: the last sentence is long, break it into smaller parts.
Please consider revising the style of figure captions. I think starting with the label followed by the explanation is a commonly used caption style and is easy to read. E.g., Figure 1: Map of the Gulf of Thailand. (a) Triangles indicate the locations of tide gauges: FP denotes Fort Phrachula Chomklao station, KL denotes Ko Lak station, and KM denotes Ko Mattaphon station. (b) Shows the location of the Gulf of Thailand. Colour contours in (a) and (b) represent bathymetry. The black contour in (a) represents the zero-depth level.
Figure 3. I could not locate boxes in Figure 2.
Figure 5. Please use the positive half of the (b) Figure colour palette. On (a) blue corresponds to extreme positive but on (b) the same blue is extreme negative. This can be very confusing.
Figure 6. Same as in Figure 5. Add zero contours in (b) and (d).
Figure 7. sea surface height (black) just needs to be mentioned once. How can a negative wind stress curl have both negative and positive values? I think the point here is that the displayed data was multiplied by (-1) to be in phase with the other parameter? Winds act on a much larger scale, so I am not clear about the point of the correlation between the green cross and triangle.
Figure 8. Non-significant correlation could be masked?
Reference
Buranapratheprat A, Yanagi T, Sojisuporn P, Booncherm C (2006) Influence of local wind field on seasonal circulation in the Upper Gulf of Thailand. Coast Mar Sci 30(1):19–26
Citation: https://doi.org/10.5194/egusphere-2022-495-RC2 -
AC2: 'Reply on RC2', Arachaporn Anutaliya, 05 Oct 2022
The paper introduces a discussion about the GoT surface circulation using remotely sensed data. The impacts of monsoon wind as well as those of ENSO and IOD on GoT surface circulation are discussed. Complex EOF and correlation analyses were used to investigate the drivers of surface circulation patterns. Seasonally, Complex EOF showed that about 28% of changes in surface circulation were attributed to the monsoon wind reversal. On interannual timescales, ENSO and IOD have spatially varying impacts on surface circulation with ENSO influence being more pronounced in the GoT interior whereas the GoT western boundary responds more to IOD conditions as evidenced by correlation analysis.
The paper is fairly well written. It introduces an important topic that can benefit a larger community to understand the impacts of climate on surface circulation and the extension to biogeochemical processes. However, there are a number of improvements that can help build a stronger case.
- The significance of the study should be emphasised. Besides the use of field observations and numerical modelling to study the surface circulation of GoT, previous studies also used remotely sensed data but findings did not converge. So, what makes the application of remotely sensed data in this study relevant. Also, most of the remotely sensed data used in this study were available from the 90s but the analysis only focused on the period starting from 2014. Why?
- This study improves the understanding on the dynamics associated with the GoT circulation that has not been well-studied. It shows the role of wind on modifying both geostrophic and ageostrophic current. In addition, the study provides the variability of the circulation at the interannual timescale. The author tried to emphasize the significance of this study better in abstract, results, and conclusion sections.
- Thanks to the reviewer for the helpful suggestion, discussions of the interaction between GoT circulation and (1) the upper GoT and (2) the SCS are also added to the manuscripts.
- The reviewer is correct that most of the datasets are available since the 90s but under a time constrain of the provided grant, the author can only analyze these few years. The author picked 2014 to 2019 as the studied period because it includes a few major interannual events, such as the 2015/2016 El Nino, the 2016 negative IOD, 2017/2018 La Nina, and the 2019 positive IOD.
- In addition to datasets section, please also add a methods section. Most of the statistical analysis used were neither thoroughly explained nor proper references were provided. This is a weakness of the current manuscript. Complex correlation, rms correlation, etc. are not common correlation analysis methods and should be well introduced with proper citation prior to their use. Without such information it is difficult for the reader to make an easy interpretation of results presented. As for the datasets, please include the data source URLs in the datasets section and explanation about the ENSO and IOD data, and their provenance.
- The author thanks the reviewer for the suggestion. The methodology section is added to describe the complex empirical orthogonal function.
- Complex correlation is introduced, and reference for the calculation is provided.
- The ENSO and IOD indices are added in the dataset section. The data sources are provided in the data availability section.
- OSCAR and ADT data with spatial resolution of 27–37 km were used to discuss surface circulation patterns in the uGoT. This is a small (~ 100 km horizontally) and shallow area (Figure 1). The author recognises the limitations (L340) but still places a lot of emphasis on the variability of OSCAR current data in the uGoT (e.g., L120, L135, L184, etc.). I would recommend introducing the L340 text earlier and limit the discussion of uGoT surface circulation based on OSCAR data. Please note that the circulation patterns in the uGoT have been discussed using local wind data derived from meteorological stations as a way to overcome the limitations of coarse resolution remotely sensed wind data (Buranapratheprat et al., 2006).
- The author thanks the reviewer for the suggestion. Discussion regarding the uGoT circulation, particularly that derived from the OSCAR products, has been limited and the author introduced the limitation regarding the observations over the uGoT in the dataset section (L87) and again at the beginning of the results (Circulation in the Gulf of Thailand) section (L146).
- The introduction has been updated and the reference has been added.
- The sequence of some figures should be revised to make the flow easier. Complex EOF (Figure 3) can be understood with easy after Figure 4 is introduced, and after the method has also been introduced. Similarly, the correlation maps in Figure 8 are better after Figure 9 is introduced which will be in line with the text in L250.
- The author used the CEOF to understand the dominant pattern associated with circulation over the GoT. As a result of the CEOF analysis, the most distinct circulation pattern associates with the seasonal monsoon winds, thus, the monthly circulation is examined in detail. Therefore, the author believes that the original figure sequence provides the more coherent story.
- The methodology is added immediately after the dataset section.
- Similarly, the author used the correlation map to understand which regions are heavily impact by each climate mode. Then, the regions with high impact, i.e., that along the western boundary and that in the GoT interior with 2 sub-regions, are further analyzed to understand how low-frequency variability over these regions change with the climate modes. The author believes that presenting figure 9 before showing figure 8 would cause confusions to the readers.
Details
L30: please revise the text for clarity. Numerical simulations were done for uGoT but the spatially uniform wind does not represent that of GoT?
- The original text pointed out that the study by Yanagi and Takao (1998) indicates that the wind field over the entire GoT (including the uGoT), and thus the use of uniform wind field over the uGoT could yield inaccurate results. The author understands that the original text might provide a misleading detail, and thanks the reviewer for pointing it out. Thus, the text has been updated.
L38: “The study suggests an overall cyclonic circulation…” Which study?
- The author refers to the study by Saramul (2017) which is based on the high spatial-resolution coastal radar mentioned in the previous sentence. The text has been updated to clarify this point.
L55: better open a new paragraph.
- The text has been updated.
L77: It is helpful to add the locations of the high frequency radar system in Figure 1.
- Figure 1 has been updated to include locations of the high frequency coastal radar stations.
L80: please explain how this comparison was done. How the spatial and temporal resolution differences were addressed? As admitted in L340 the OSCAR data may contain large error there. How to distinguish between the two, large difference or large error?
- The monthly mean data as provided by Saramul et al. (2017) is calculated from the OSCAR products for the comparison; the text has been updated to clarify this point. As of the author’s understanding, it is not possible to pinpoint whether the difference is truly a difference or an error. However, given the process associated with estimating the OSCAR products, the derived current along coastal areas could contain high error and a validation is necessary. As there has been no validation of OSCAR products against an observation over the GoT, the comparison to an observation which is the high frequency coastal radar is needed. The results show that the OSCAR products is quite reliable over the GoT, particularly to the south of the uGoT.
L90: 20 km is below the ADT spatial resolution. How to know this is not noisy data?
- The reviewer is correct, noisy data can contribute to a lower correlation coefficient. However, it is difficult to specify which factor contributes to the low correlation at the specific location. The author has updated the text to address this issue.
L99-100: references for A and wind stress curl estimation?
- The estimations of A and wind stress curl are added.
L115: why use complex EOF? Does it improve the results over the classical EOF. What (additional) information is gained from the use of complex EOF? Again, these issues can be addressed by a proper methods section.
- In many studies, complex EOF is applied when the dataset being considered is in vector form. The statement is included in the methodology section added to this manuscript.
L120: much of this discussion should be removed from the text. 6 grid points are too rough to have a meaningful discussion. Any derived parameter will even include less grid pixels which further limits the interpretations.
- The author thanks the reviewer for the suggestion. The discussion has been removed.
L125: briefly explain the reader how to look at the results of complex EOF. For classical EOF the mode often indicates temporal variability when multiplied by the amplitude. What is the relationship between Figure 3a-c?
- A brief explanation regarding how to interpret the results of complex EOF is included at the end of the added methodology section. The CEOF mode (Figure 3a) describes the dominant pattern that accounts for 28% of the total variance of the velocity anomaly over the GoT. Generally, the pattern that is rotate according to the phase of the PC (Figure 3C; positive clockwise), i.e., the pattern displayed in Figure 3a is observed when the phase of PC is zero and the reverse is observed when the phase is ±p. The pattern is present with a great intensity when the magnitude of the PC (Figure 3b) is high.
L151: reference(s) for the correlation coefficient from non-parametric method?
- The reference is added to the manuscript.
L165: what is rms correlation? reference(s)?
- The text refers to root-mean-square of the correlation coefficient. Rms stands for root-mean-square as defined earlier. To compute the rms, a square root of the mean of the squared quantity of interest(correlation coefficient over the interested area in the GoT in this case), i.e. √(mean(R^2)). The rms calculation is commonly used in similar way as calculating average; hence the reference is not necessary. Using the rms instead of the average emphasizes the importance of the variance being explained.
L171: what is total current? Most of these details should be addressed by a proper methods section.
- The author understands the confusion since the original text does not clearly address the different type of currents. A statement has been added in the datasets section to clarify this point.
L224: what is the difference between this correlation coefficient and the other so far introduced?
- The correlation here is calculated between the wind stress curl and sea surface height data, while the previous correlations (subsection 4.1.2 Geostrophic and ageostrophic component) indicate how (1) the geostrophic current and the total current covary (Figure 6a, b) and (2) the ageostrophic current and the wind-driven Ekman current (Figure 6c, d).
L227: “…much of the correlation is attributed…”
- The accordant change has been implemented.
L232: “Still, the result suggests the importance of coastal trapped Kelvin waves…” add reference(s) as this is speculative?
- Coastal trapped Kelvin waves are known to travel equatorward along the western boundary of the basin and poleward along the eastern boundary based on the wave equation. Therefore, the statement regarding the Kelvin wave traveling direction is not speculative. Still, the author agree that an addition of references could be beneficial particularly to readers who are not familiar with the wave.
L239: add the description (and/or sources) of the sea surface temperature and the Niño 3.4 box in the text.
- Descriptions for Niño 3.4 and DMI are added at the end of the datasets section and the corresponding sources are added to the data availability statement.
L245: as mentioned above, the discussion in this paragraph suggested that the sequence of Figure 8 and 9 is reversed.
- The author used the correlation map to understand which regions are heavily impact by each climate mode. Then, the regions with high impact, i.e., that along the western boundary and that in the GoT interior with 2 sub-regions, are further analyzed to understand how low-frequency variability over these regions change with the climate modes. The author believes that the original figure sequence provides the more coherent story. Reversing the order of figure 8 and 9 could confuse the reader regarding the region selected to compare with each climate mode.
L250: “…correlations between low-frequency Niño3.4 and selected forcings…”. What parameter is the forcing, Niño3.4 or ADT, etc.?
- The forcings are ADT, zonal wind stress, and wind stress curl. The author investigated the impact of these forcings on the low-frequency variability of the uGoT circulation.
L300: “…but might relate to the winter warm pool” where?
- The author thanks the reviewer for pointing out the missing detail. The accordant change has been made.
L310: the last sentence is long, break it into smaller parts.
- The accordant change has been made.
Please consider revising the style of figure captions. I think starting with the label followed by the explanation is a commonly used caption style and is easy to read. E.g., Figure 1: Map of the Gulf of Thailand. (a) Triangles indicate the locations of tide gauges: FP denotes Fort Phrachula Chomklao station, KL denotes Ko Lak station, and KM denotes Ko Mattaphon station. (b) Shows the location of the Gulf of Thailand. Colour contours in (a) and (b) represent bathymetry. The black contour in (a) represents the zero-depth level.
- The author thanks the reviewer for the suggestion. The author tried the suggested caption style but cannot successfully change all of them without creating awkward sentences. Therefore, the author kindly asks the reviewer to allow leaving the caption format as is.
Figure 3. I could not locate boxes in Figure 2.
- The boxes are missing from the original figure; the figure is corrected.
Figure 5. Please use the positive half of the (b) Figure colour palette. On (a) blue corresponds to extreme positive but on (b) the same blue is extreme negative. This can be very confusing.
- The author has changed the color contour to avoid the mentioned confusion.
Figure 6. Same as in Figure 5. Add zero contours in (b) and (d).
- The author has changed the color contour and the zero contours are added in (b) and (d).
Figure 7. sea surface height (black) just needs to be mentioned once. How can a negative wind stress curl have both negative and positive values? I think the point here is that the displayed data was multiplied by (-1) to be in phase with the other parameter? Winds act on a much larger scale, so I am not clear about the point of the correlation between the green cross and triangle.
- The figure caption has been updated accordingly.
- The figure has been updated to plot wind stress curl on the second (right) y-axis with the reversed y-axis.
- In many scenarios, winds act on large scale, such as the seasonal current reversal that responds to the monsoon wind as being discussed in this manuscript and change in the warm water volume in the western equatorial Pacific due to the westerly wind bursts related to the development of ENSO events (e.g. Chen et al., 2016). However, winds do not necessarily have large-scale impact on the ocean current, for example, upwelling response to the local alongshore wind. This process is both local and small-scale. Dynamics, particularly those associated with planetary waves, associated with the ocean circulation are also sometimes remote and small-scale, for example, the effect of wind stress curl in the western Pacific can propagate in forms of Rossby waves and modify sea level in the Solomon Sea (Anutaliya et al., 2019). The effect of remotely-force waves on sea level or thermocline variation is commonly found (e.g. Dickinson, 1968; Sprintall et al., 2000; Delman et al., 2016). The author understands that impacts of planetary waves on circulation in the GoT has not been hitherto studied and could be unfaimilar to the reader. Therefore, the author has updated the text to provide some literacy regarding the topic.
- Anutaliya, A., Send, U., Sprintall, J., McClean, J. L., Lankhorst, M., & Koelling, J. (2019). Mooring and seafloor pressure end point measurements at the southern entrance of the Solomon Sea: Subseasonal to interannual flow variability. Journal of Geophysical Research: Oceans, 124, 5085– 5104. https://doi.org/10.1029/2019JC015157
Chen, S., Wu, R., Chen, W., Yu, B. and Cao, X. (2016), Genesis of westerly wind bursts over the equatorial western Pacific during the onset of the strong 2015–2016 El Niño. Atmos. Sci. Lett., 17: 384-391. https://doi.org/10.1002/asl.669
Dickinson, R. E. (1968). Planetary Rossby Waves Propagating Vertically Through Weak Westerly Wind Wave Guides, Journal of Atmospheric Sciences, 25(6), 984-1002.
Figure 8. Non-significant correlation could be masked?
- The figure has been updated accordingly.
Citation: https://doi.org/10.5194/egusphere-2022-495-AC2
-
RC3: 'Comment on egusphere-2022-495', Anonymous Referee #3, 23 Aug 2022
The paper presented the surface circulation in the Gulf of Thailand from remotely sensed observations. There are many interesting study results presented in this paper. I propose to accept this paper for publication with minor revision.
The comments are
1. The GoT is influenced by the SCS. More results related to the influence from the SCS should be mentioned or analyzed.
2. The uGoT region used in this paper should be clearly defined because it is mentioned differently in the paper.
3. CEOF is widely used in the atmospheric and marine scientific research and is also used in the paper, but it is rarely found to be used to study for the GoT. Therefore the CEOF method should be briefly described, although it is well known.
4. There is an error in Figure 1, please revise.
5. Please review the references, for examaple: line 370, Atoms?
6. The validations of the results were still unclear. Please describe the results of the main goal clearly.
Citation: https://doi.org/10.5194/egusphere-2022-495-RC3 -
AC3: 'Reply on RC3', Arachaporn Anutaliya, 05 Oct 2022
The paper presented the surface circulation in the Gulf of Thailand from remotely sensed observations. There are many interesting study results presented in this paper. I propose to accept this paper for publication with minor revision.
The comments are
The GoT is influenced by the SCS. More results related to the influence from the SCS should be mentioned or analyzed.
- More discussions regarding the interaction with the SCS is added in both the introduction and results sections.
The uGoT region used in this paper should be clearly defined because it is mentioned differently in the paper.
- The author thanks the reviewer for pointing this out; the text has been updated to be consistent.
CEOF is widely used in the atmospheric and marine scientific research and is also used in the paper, but it is rarely found to be used to study for the GoT. Therefore the CEOF method should be briefly described, although it is well known.
- The author thanks the reviewer for the suggestion. The methodology section is added to describe the complex empirical orthogonal function.
There is an error in Figure 1, please revise.
- The author cannot locate the error and would greatly appreciate it if the reviewer could provide any detail regarding the error.
Please review the references, for examaple: line 370, Atoms?
- The author apologizes for the wrong information provided. The text has been revised and the reference has been reviewed.
The validations of the results were still unclear. Please describe the results of the main goal clearly.
- This study improves the understanding on the dynamics associated with the GoT circulation that has not been well-studied. It shows the role of wind on modifying both geostrophic and ageostrophic current. In addition, the study provides the variability of the circulation at the interannual timescale. The author tried to emphasize the significance of this study better in both abstract, results, and conclusion sections.
Citation: https://doi.org/10.5194/egusphere-2022-495-AC3 - More discussions regarding the interaction with the SCS is added in both the introduction and results sections.
-
AC3: 'Reply on RC3', Arachaporn Anutaliya, 05 Oct 2022
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 287 | 136 | 21 | 444 | 5 | 6 |
- HTML: 287
- PDF: 136
- XML: 21
- Total: 444
- BibTeX: 5
- EndNote: 6
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Arachaporn Anutaliya
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(13138 KB) - Metadata XML