Lake Volume Data Analyses: A Deep Look into the Shrinking and Expansion Patterns of Lakes Azuei and Enriquillo, Hispaniola
Round 1
Reviewer 1 Report
These are very interesting research areas for geophysical analysis, for hydrodynamical flow behaviour, reason for climate and their technical solutions. Although, you have done lots of statistical analysis by using hydro-meteorological data, but there is still missing scientific knowledge.
What will learn scientific community from this paper?
Lake analysis needs deep hydrodynamic, Geo-chemical, surface flow and specially groundwater transportation knowledge, why these are missing within your paper, explain clearly?
Explain, which contribution is more important for future analysis or how can scientific community will feel interest to use your results within their research?
Please explain clearly within your paper.
Line 37: missing references;
Line 44: Add reference;
Line 86-100: Explain more details of your results;
Line 99: If possible, add mass balance solution in here:
Line 113-114: check again;
Line 117-127: add mass balance formula in before, and elaborate the section;
Line 173-174: add figure number;
Line 178-191: add water mass balance numerical simulations with statistical analysis;
Line 196-205: Explain clearly by numerical simulation;
Line220-221: add reference;
Line 273: A change point (PC), the geophysical and groundwater analysis is very important for that, why there was no information about that?
Line 334-347: check again;
Line 354-361: check again;
Line 380-382: check again;
Line 389-408: check again;
Line 410-427: check again;
Line 439-443: Missing these in our analysis.
Line 448: Can you make any simulation by using the figure information?
Line 451-471: check again;
Line 474: check again;
Line 520: Conclusion: Write, very clearly the scientific contribution of this paper.
Author Response
Reviewer 1
These are very interesting research areas for geophysical analysis, for hydrodynamical flow behaviour, reason for climate and their technical solutions. Although, you have done lots of statistical analysis by using hydro-meteorological data, but there is still missing scientific knowledge.
What will learn scientific community from this paper?
There are numerous insights that we hope are able to convey to the scientific community. Firstly, the expansion of the lakes has not been an easy-to-explain phenomenon producing numerous attempts of explaining as to the “why”. In this case it turns out that weather patterns at a much larger scale have an important impact on the lakes, i.e. the North Atlantic Oscillation (or NAO) patterns influencing both the occurrence of extreme events and long term shifts in climate in the Caribbean which in turn leads to local changes which are then felt by the lakes. There are processes at play that are not obvious at first look. Secondly, we are introducing a set of statistical or signal processing analyses steps that can aid in understanding what forces and events were at play at certain points in time and their impact on the lakes’ response patterns. This should help or at least be of interest to researchers who are interested in conducting lake research elsewhere in the context of changing climate and weather patterns. Thirdly, we hope that the results obtained are of interest to the general research community and also governmental research laboratories that conduct research around the lakes, which could also be biologists and ecologists. Lastly, the research effort had to deal with a lot of adversity in terms of accessibility in both countries in addition to a significant dearth of missing information on almost all physical aspects that are related to the lakes be it hydro geological, hydro climatological, or land surface in nature. We provided a path forward using information and data that mostly (except the bathymetric data, but this is data we collected and which was not available in either country) does not originate in either of the countries but is available as public information in the United States. In other words, we tried to develop defendable scientific insights from extremely limited data sources to answer the questions posted.
We have adjusted the Conclusion section to better and more clearly emphasize these points.
Lake analysis needs deep hydrodynamic, Geo-chemical, surface flow and specially groundwater transportation knowledge, why these are missing within your paper, explain clearly?
Line 99: If possible, add mass balance solution in here:
Line 117-127: add mass balance formula in before, and elaborate the section;
Line 178-191: add water mass balance numerical simulations with statistical analysis;
Line 196-205: Explain clearly by numerical simulation;
Line 439-443: Missing these in our analysis.
We have lumped the above references from reviewer 1 into one single paragraph as we believe that the items raised are all linked together and address the central question: Why is there no detailed mass balance equation approach?
As mentioned under 1) the absence of data on geology, landcover/use, climatological mid to long term data as well as lake level measurements and the absence of geographical control points all limited the approaches one could in a defensible way to conduct research into the shrinkage and expansion patterns of the two lakes. While we would have liked to build a detailed numerical model of the watershed, the lake, and the subsurface (ground water model), the lack of information to inform these models simply renders any such attempt un-defensible because of the endless assumptions the models would have needed without being to calibrate and validate them. The reasons for this situation are manifold, mostly of course because the lack of funds and culture to support fundamental research with data collection campaigns, especially in Haiti.
The use of locally collected static data (bathymetry) and otherwise mostly remote sensed data is, in our opinion, the only way to find answers to the lakes’ behavior. Introduction of a mass balance equation would necessarily require the knowledge of the items that the reviewer rightly points must be present. We decided to avoid an approach for which we did not have the necessary and needed data and instead focus on what we do have: time series information and bathymetric data what would allow us to analyze (signal process) the time series which could then be coupled to the bathymetric/elevation data.
We realize that we erroneously guided the reader/reviewer to the idea of mass balances and numerical model development with the last sentence of our introduction. This was not intended and ended up introducing a lack of clarity and unwanted expectations. As a consequence, we deleted the sentence and thank the reviewer for stressing that when you introduce aspects and goals delivery of the supporting information is needed. We hope that the removal of any reference to mass balancing models now clarifies that this paper is all about time series analyses and not mass balancing models.
Explain, which contribution is more important for future analysis or how can scientific community will feel interest to use your results within their research?
We believe that the time series analyses carried out helps to direct further research into the lakes’ region. We have highlighted the connections that exist to the weather patterns as well as the anthropogenic influences and also worked out connections that either do not exit or are weak in nature. For example, while we do not believe that a detailed watershed scale knowledge of the subsurface (residence, storage, and flow times) is really all that important (both lakes are endorheic in nature) knowledge of the subsurface flow rates between the two lakes is as it clear that the hydraulic gradient between the lakes will cause flow. We also feel that the importance of anthropogenic impacts is a key outcome, observed behaviors of the environment is not entirely linked to climate change and weather patterns but also to what we as humans inflict and change on nature.
PLEASE FIX THE FOLLOWING ITEMS
Line 37: missing references;
Line 113-114: check again;
Line 173-174: add figure number;
Line220-221: add reference;
Line 334-347: check again;
Line 354-361: check again;
Line 380-382: check again;
Line 389-408: check again;
Line 410-427: check again;
Line 451-471: check again;
Line 474: check again;
Above items have been fixed in the text.
Line 44: Add reference;
The following reference has been added accordingly.
*Kushner, J. The Relentless Rise of Two Caribbean Lakes Baffles Scientists. 2016. Available online: https://www.nationalgeographic.com/news/2016/03/160303-haiti-dominican-republic-lakes/ (accessed on 3 March 2016).
Line 86-100: Explain more details of your results;
While we appreciate the reviewer’s suggestion, we would like to follow the approach we have chosen. In our opinion the Introduction is intended to guide the reader into the paper by stating what the problem is, what motivates the research, and how we want to go about it. We would like to reserve the presentation of results to the Discussion section as otherwise the reader is forced to read something about results which would necessarily be brief before reading it again later on. We think that the structure of the narrative, and with that the flow of it, is better served by dedicating a whole section to the results after the reader has been introduced to the problem, the objectives and the methods used.
Line 273: A change point (PC), the geophysical and groundwater analysis is very important for that, why there was no information about that?
As we have pointed out under 1) and 2) the information available on groundwater and geophysical settings is so poor that there is no hope to getting anything useful out of the information that exists here and there, if at all. We needed to confine ourselves to using change point analyses to time series data that we had access to.
Line 448: Can you make any simulation by using the figure information?
This figure (#12) actually just presents a schematic Illustration of closed-lake basin responses in equilibrium to climatic variation. However, the parameters (the equilibrium response time) shown in the figure can be simulated for each case of a step, spike, and sinusoidal change in forcing factor. The detailed explanation of this concept along with associated equations (derived from simplified mass balance) can be found in Manson et al. (1994). Not only did we use the suggested formulas to calculate , but also did we use this schematics illustration to see if any similarities can be observed in the lakes’ volume time series which could provide an indication of the type of external force on the lakes and their basins. More on this presented in the paragraphs following figure #12.
*Mason, I.M.; Guzkowska, M.A.J.; Rapley, C.G.; Street-Perrott, F.A. The response of lake levels and areas to climatic change. Clim. Change 1994, 27, 161–197.
Line 520: Conclusion: Write, very clearly the scientific contribution of this paper.
We modified the Conclusion section to better present the scientific contribution of this paper.
Reviewer 2 Report
Comments
The first paragraph in the introduction should be moved as a section "Study area" in the Materials and Methods The last paragraph in the Introduction should be changed in order to highlight that you are presenting a methodology of various steps that solves a specific problem using as a case study the two lakes. Figure 1-11 are not referred in the text. In line 174, 222 there is a figure indication without number. Tables are not referred in the text Fix equation (1) as one equation Avoid using "We"Author Response
Reviewer 2:
The first paragraph in the introduction should be moved as a section "Study area" in the Materials and Methods
While we can see where the reviewer is coming from we prefer to keep this first paragraph where it is. For one, we do not really have a study area section per se that would introduce the hydrogeological characteristics simply because we do not have this information. For another, we want to guide the reader to the geographical context so he/she knows immediately where the area of interest is and then proceed to describing the problem at hand, as, so we believe, any introduction should do. We should also like to point out that this all about time series analyses, not about the detailed description of a study area beyond what we feel is just setting geographical the context.
The last paragraph in the Introduction should be changed in order to highlight that you are presenting a methodology of various steps that solves a specific problem using as a case study the two lakes.
Lakes are indeed very unique in their characteristics and the authors appreciate the suggestion made. We, however, hope that the presented set of steps can act as a general guideline or perhaps method of tackling any time series analyses especially when the scarce and only very specific data sets (remote sensing) are available. We have changed the last paragraph accordingly to stress this.
Figure 1-11 are not referred in the text.
All reference and numbering issues have been corrected.
In line 174, 222 there is a figure indication without number. Tables are not referred in the text
All reference and numbering issues have been corrected.
Fix equation (1) as one equation
We have removed the equation from the text as it did not add substantial information content and also modified the accompanying text accordingly.
Avoid using "We"
We have removed all reference to the use of personal pronouns.
Reviewer 3 Report
The manuscript "Lake Volume Data Analyses: A Deep Look into the Shrinking and Expansion Patterns of Lakes Azuei and Enriquillo, Hispaniola" discusses the temporal variation of lake volume for both lakes and a lot of statistical works have been done. The article is well-written, and the figures and tables are well organized. Some more detailed question are to be answered:
1)Format of manuscript needs to be carefully checked, all the figure and table reference is not correctly presented;
2)In line 125, sudden changes within a few days is said while in line 147, the lakes volume change is described to be slowly (monthly time scale), is the 16 days interval able to catch the change?
3)Results of 15 imputation methods are shown in Table 1, but analysis is not enough, why Stineman interpolation give the best result, is it because the given data matches its assumptions? and what is the difference between these imputation method that causes the difference expect for the randomness you mentioned in line 326?
4)Results from wavelet analysis show LE has statistically significant scale of annual variability while LA has both annual and subannual variability, can you explain why?
5) Event last from 1998 to 2004 in Figure 13 shows a elongated tail for LE, why for LA in figure 5 it is not that clear?
6)The 2007 Trujillo Dam failure seems to be the only human-induced forcing, is there other possible anthropogenic influences?
7)In the discussion part, more content should be provided about how could the analyzed results help to build the lake mass balance model as in line 99.
Author Response
Reviewer 3
The manuscript "Lake Volume Data Analyses: A Deep Look into the Shrinking and Expansion Patterns of Lakes Azuei and Enriquillo, Hispaniola" discusses the temporal variation of lake volume for both lakes and a lot of statistical works have been done. The article is well-written, and the figures and tables are well organized. Some more detailed question are to be answered:
Format of manuscript needs to be carefully checked, all the figure and table reference is not correctly presented;
We have corrected all references to figures and citations in the text.
In line 125, sudden changes within a few days is said while in line 147, the lakes volume change is described to be slowly (monthly time scale), is the 16 days interval able to catch the change?
The general response time scale is slow for both lakes, i.e. in the monthly range. However, occasionally there is a relatively quick response especially after a Hurricane event. For example, Storm Faye passed through the region on 08/16/2008 and an image became available just four days after that on 08/20/2008. If we assume no to only moderate lake level activity until 08/16 than we have a new image within just 4 days that shows a dramatic response, hence there is evidence of a very short scale of just a few days. Other image-after-event intervals were between 4 to 20 days. We have modified the narrative in the first paragraph of section 2.2 to accommodate this better explanation.
Results of 15 imputation methods are shown in Table 1, but analysis is not enough, why Stineman interpolation give the best result, is it because the given data matches its assumptions?
The short answer is yes. We applied the leave-p-out cross-validation approach. In this method p number of observations were removed from the time series and the imputation method was used to estimate the values of the removed data points. Then we calculated the Normalized Root Mean Squared Error (NRMSE) between the removed observed data point and their estimated value using the various imputation methods to estimate the performance of the different methods. This procedure was repeated 1000 times, each time removing a different set of observations and then calculating the NRMSE. At the end the average NRMSE was calculated for each method to identify the method generating the smallest NRMSE, i.e. targeting the best set of removed observed data points. In our case, the Stineman interpolation generated the smallest NRMSE. We have modified the text accordingly to better explain this aspect of the procedure.
What is the difference between these imputation method that causes the difference expect for the randomness you mentioned in line 326?
Although our results did not show much sensitivity to the use of the different interpolation methods we applied (the table shows how their performances are close to each other), the Stineman interpolation provided us with the best results which is supported by its good performance in the presence of abrupt changes (Stineman, 1980) and when the density of data points changes in time (Leclercq and Oerlemans, 2012), which are characteristics that are both present in the time series. We have modified the text in this section to better highlight this aspect (and also added the two references).
Stineman, R.W. A consistently well behaved method of interpolation. Creat. Comput. 1980, 6, 54–57. Leclercq, P.W.; Oerlemans, J. Global and hemispheric temperature reconstruction from glacier length fluctuations. Clim. Dyn. 2012, 38, 1065–1079 Results from wavelet analysis show LE has statistically significant scale of annual variability while LA has both annual and subannual variability, can you explain why?
Both annual and sub-annual periodicities are present throughout LA and LE Continuous Wavelet Transform (CWT) time series. However, the global wavelet analysis shows the results you mentioned. The reason behind this fact lies in the difference between the responses of the lakes to climatic forces. While LA reacts to quantity of rainfall, LE is more sensitive to intensity of rainfall. Precipitation time series show two rainfall seasons in Spring and Fall with the same amount of monthly rainfall (annual and sub-annual periodicity). However, the intensity of rainfall in the Fall is higher due to hurricane season effects (annual periodicity). That is why the detected periodicities are different for the lakes and this difference is because of their different basins’ characteristics.
Event last from 1998 to 2004 in Figure 13 shows an elongated tail for LE, why for LA in figure 5 it is not that clear?
As mentioned, LE is more sensitive to rainfall intensity which appears in its monthly variation of the lakes volume. It is also sensitive to the North Atlantic cyclone incidents (shown when assessing time series outliers) that effect its interannual trend. Figure 13 shows the interannual response of LE to an extreme event (Hurricane George, 1998) and one can also observe the monthly variations in the picture (smaller dunes). On the other hand, LA is only responding to the amount of rainfall affecting its monthly variations. Its interannual behavior, however, is not sensitive to the North Atlantic cyclones. That is why one cannot observe an elongated tail between 1998 and 2003. The behavior of the lake during that time frame is in response to other factors such as direct rainfall, runoff, and late groundwater flow entering the lake, which are altogether responsible for the bell shape of the time series. In fact, if one plots both time series side by side using the same y-axis scale, one can observe that the changes in LA are very minimal in comparison to LE, showing the stability of its system to internal and external change.
However, additional studies might be required to untangle the details of both lakes response by introducing a hydro-climate model (such RAMS or WRF), taking into account the sub-monthly variations of the lakes, the sensitivity of LE to NAC events, better representation of anthropogenic influences, and the connection of the lakes. The details of these aspects however were beyond of this research scope.
The 2007 Trujillo Dam failure seems to be the only human-induced forcing, is there other possible anthropogenic influences?
The Dam failure is clearly the largest and most significant anthropogenic impact on Lake Enriquillo (and Lake Azuei, albeit to a lesser extent, due to the interconnectedness of the lakes). However, the continued high degree of deforestation on the Haitian side now also spreading onto the DR side of the lakes is leaving its imprint on the lakes. Because rainfall interception and evapotranspiration is reduced more water occurs as run off, rather than as a return to atmospheric water (evapotranspiration) or infiltration (subsurface flows). This has clearly an impact on the time scales of how water moves from store to store as well some impact of how much water reaches the lake. On the other side, total volume of precipitation over the lake watershed (minus the delta in evapotranspiration) largely still reaches the lake. It just happens faster in a deforested state. We have modified the narrative in the discussion section to point out this aspect.
In the discussion part, more content should be provided about how could the analyzed results help to build the lake mass balance model as in line 99.
We recognize the need for additional discussion on the mass balance model if the text would stay as is. In a way, the authors created this need because it was mentioned in the introduction and thus creating perfectly reasonable expectations that the authors should expand on that later on. However, after careful reflection on this issue the authors have come to the realization that this water balance model reference is entirely unnecessary and in fact just causing confusion (reviewer 1 too has had an issue with this). We thus removed any reference to mass balance modeling because this is NOT the focus of this paper; time series analysis is and we should not confuse the scope of the paper. We are grateful for the reviewer stumbling over this aspect and calling for clarification. We hope the removal introduces more clarity.
Author Response File: 
Author Response.pdf
Round 2
Reviewer 1 Report
Dear Author,
Now your manuscript form is much better than previous.
The statistical analysis was very-well, and more informative rather than previous.
Best Regards,
Seema Paul
