System Dynamics-Multiple Objective Optimization Model for Water Resource Management: A Case Study in Jiaxing City, China
Round 1
Reviewer 1 Report
This study uses a combination of system dynamics simulation and multi-objective optimization to model future water supply, demand, and shortage in an area containing the city of Jiaxing China and nearby agricultural land. Key to the approach is “dynamic adjustment”, that is, computation of the effects on simulated study area water shortage of alternative levels of key policy variables. Results of the simulations and adjustments are used in the subsequent optimizations. The case study is apparently modeled at the annual time step and with just one point of water inflow and use (that is, without flow routing among multiple inflow and outflow nodes), which may or may not be a realistic characterization.
General comments
The paper presents an interesting approach to modeling water shortage. The principal contribution of the paper, in my opinion, is to introduce the three-part approach (simulation, dynamic adjustment, optimization) to water resource specialists unfamiliar with pairing system dynamics with optimization. I fall in that group. In that light, I unfortunately found the presentation and application of the method rather confusing, in part because many key details were left unspecified, and also because the practical value of the optimization exercise was not clearly explained.
In addition, while the method is interesting and potentially very useful, the case study/application is useful principally to demonstrate elements of the method. I found the results of the of the application to Jiaxing City only moderately useful, because the application lacked realism and thus was not of much use in showing what policy changes are actually sensible for the Jiaxing region (see comments below for examples of methodological choices and data limitations that limit the practical import of the application). As written the paper leaves the impression that the authors intend for the results to have practical application, but I think the authors should try to avoid that impression.
1. I suggest that the paper say upfront that use of the method is demonstrated with data from the Jiaxing region, but that due to data limitations and practical simplifications the case study is not intended to be a realistic characterization of future conditions or to recommend policy solutions. In line with this suggestion, I urge the authors to consider changing the “Results” section head to something like “Case example,” and moving the Study area and Data sources subsections to section 3. Further, this change would necessitate changes to the Discussion and Conclusions, where the text is overly concerned with its findings about recommended policy changes in the Jiaxing region. More detailed comments follow.
Simulation
2. “Water demand” (e.g., line 92) is typically used in the paper to indicate desired water use, although water “consumed” is also used (line 119). “Demand” could refer to withdrawal or consumption (i.e., withdrawal minus return flow), and thus must be defined in the paper.
3. Description of data sources for the case study is inadequate. Regarding inflow, the paper does not really say how the three levels of inflow (line 111) were determined. All we are told is that local statistics, historical data, and assumed values were used (about line 225) and that the "typical hydrological year method" was used (line 110). More information is needed about both the data and that method. Later on regarding the data sources given for demand and supply (line 229 and thereafter), the information is vague and the cited documents are in Chinese. This is not a problem if the intent if the case study is only to demonstrate how the method might be applied, but much more is needed if the case study is intended to provide practical suggestions about policy or to demonstrate a viable and comprehensive application.
4. Furthermore, and more importantly, regarding water supply, as seen in Figure 3 the three inflow levels are constant over the time horizon (2018 to 2035) except for changes in water reuse. This is of course unrealistic, since it ignores the stochastic nature of precipitation and streamflow. It is common in modeling water supply to take into account the variability in inflow over time. Because flow variability is ignored, the three supply scenarios are completely unrealistic. Unless the intent of the Jiaxing application in the paper is limited to showing how elements the SD-MOO method could be implemented, this issue must be addressed in the paper.
5. On calibration of the SD model the paper describes three measures (equations 2-4) that each use the difference between “simulated” and “observed” values and the means of those values (line 138). However, the description of the methods fails to reveal what the simulated and observed values are, or what data were used to obtain the observed values. That is, what do the Vs represent in equations 2-4? In section 3.1 we find that eight variables were used (Table 2) but there is no explanation of why those specific variables were chosen to calibrate the model; an explanation is needed.
6. In the case study, it is projected that domestic demand will soar, that green space will also increase, and that agricultural demand will decrease slightly due to gains in efficiency. This suggests that the land for expanding domestic housing and green space will not come from existing agricultural land (or any other water-using area). Is this realistic? Again, unless the Jiaxing application is used only to show how elements the SD-MOO method could be implemented, the paper should explain where the land for expanding domestic water use would come from. Note that in many areas of the world, cities expand into what had been agricultural land.
“Dynamic adjustment”
The paper reports that in former applications of SD in water resource management the "parameters of the SD model remain constant" (line 53), and it asserts that "dynamic management with feedback mechanisms is more suitable" to SD-based simulation of water supply and demand. The paper goes on (section 2.1.2) to describe a “dynamic adjustment” approach, whereby the effects on WSR (simulated study area water shortage) of systematic adjustment of 11 different policy-adjustable variables (Table 1) are estimated.
7. A first suggestion is that the explanation of the approach could be simplified. Explaining the adjustment approach in terms of multiple dimensions (SV1, SV2, etc.; mu1, mu2, etc.) is not necessary for this paper, since the approach was demonstrated with just one dimension (i.e., one state variable, WSR). Why not just say that the approach could be applied to multiple dimensions, but here we use just one? This would eliminate the m subscript.
8. The role of the feedback ratios in specifying changes in the policy variables is not clear in the explanation of the method (section 2.1.2) and that role is not clarified in the results (section 3.3). I suggest the addition of a simple explanation of the computation of feedback ratios and of effect of a feedback ratio on the various levels of a respective policy variable.
9. Eleven policy variables are listed in section 2.1.2 (Table 1). Those variables are specific to the Jiaxing case study, and should probably be presented in section 3, not in the methods section.
10. The methods section should say something about how one chooses the policy variables to include and how one decides what values of those variables (or at least the value range) can be sensibly used in an analysis of water supply and demand. Changes in the policy variables are critical to the approach, yet in section 3.2 the paper says nothing about the actual levels of the variables. It would be helpful to the reader, for example, to see the values of p(n) for the S0 scenario and for one or more of the other scenarios listed in Table 3. (If the paper were also trying to present realistic results for the Jiaxing region, it should also discuss how realistic the modeled levels of the policy variables are, and how they could be achieved.)
Optimization
11. In section 2.1.3 it was not clear to me how simulation (including adjustment) results were use as constraints in the optimization step. This needs to be clearly explained.
12. Regarding the results of the optimizations (section 3.4), so little detail about the inputs to the different optimization objectives (e.g., water fees, COD levels, minimum and maximum requirements) is presented that the actual results are difficult to interpret. That is not a problem if the paper intends to merely demonstrate an approach (as opposed to present useful results). Here especially, if the intent of the case study is limited, some changes are needed to de-emphasize the practical utility of the case study.
Some further details
When modeling future water supply and demand, it is now common to include the projected effects of climate change. There is no mention of climate change in the paper. Given the importance of climate change for future water supply and demand, the paper should indicate that case study ignores climate change.
Line 130. The figure caption should indicate what the red font means.
Line 271. In the Figure 3 caption, note what the subscripts (1, 2, 3; 1, 2, 3, 4) mean.
Line 38. “To meet increasing demands, humans have extensively interfered with the hydrological system and artificially accelerated the dualistic water cycle by altering surface processes and storage, which has led to severe droughts and water shortages [9].” My take on the Van Loon et al. paper cited at the end of that quote from the paper is that the Van Loon et al. paper raises a lot of issues and questions, and indicates that human-caused changes in storage and surface process have caused both increases and decreases in drought, all else equal. The authors of the current paper, however, focus only the negative impacts and state that human-caused changes in “surface processes and storage” have resulted in “severe droughts and water shortages”. I suggest a reassessment of Van Loon et al. I note that, assuming certain definitions of drought, the main human cause of increasing drought is increasing human consumptive use of water, not changes in surface processes and storage.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Thank you for giving me the opportunity to review this interesting and significant paper. The introduction explains the background of the study very well and the methodology section is vividly explained that will help other researchers to follow the research. The results are critically presented and the only suggestion is to include a little about the policy implications of the study in the discussion. I would highly recommend this paper for publication.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
This paper is based on the development of a framework involving a system dynamics-multiple objective optimization (SD-MOO) model to predict the available water resources, allowing to better plan the allocation of water to different users. Different scenarios were then created to assess the SD-MMO capability to allocate water resources to different users, proving to be an effective tool in achieving sustainable water resource management.
The MS is well structured and its findings provide valuable information on how water resources may/should be managed. The presented results and conclusions are consistent with aiming of the MS. However, some issues arise as discussed below. Overall, the publication of the MS is recommended after minor revisions.
However, what puzzles me is the fact that the authors assumed a static future irrigated area for most land uses despite of a decrease in rural population and an increase of urban population. Also, the authors estimate a significant increase of GPD; but no data is provided on how those estimates were made. The same happens for the assumed water quotas for each sector; no information is provided on how those values were established. I would suggest for a more detailed discussion on how those data were obtained and what the impacts of other types of scenarios would be.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
I have read the authors' response to my comments, which suggests that they did make a credible attempt to amend the paper in line with my comments.
