Impacts of Human Activities on Hydrodynamic Structures during the Dry Season in the Modaomen Estuary

Round 1

Reviewer 1 Report

The present study investigates the impact of human made changes in the coastline on the tidal hydrodynamics in Modaomen Estuary. It is important to understand this as the PRD coastline changes quickly and long-term environmental consequences are not well-studied. However, I have following comments/suggestions to authors before their manuscript can be accepted for publication.

- There is so much previous work done on tidal-river hydrodynamics. Authors should expand the introduction to have a reasonable and a thorough literature review. Currently they only cite the ones that focus on PRD. One seminal example of which is by Hoiting and Jay (2016) and references therein (https://doi.org/10.1002/2015RG000507). Then the authors should more explicitly explain the novelty of their work and the advantages of their model compared with those methods reported elsewhere.

- Next, they need to provide more details about their “quasi-three-dimensional river-tide-salt dynamic model”. It would be helpful to learn more about the theoretical background (e.g., equations solved). Also, they need to discuss why they chose this model. Also, it is also not clear if model is validated at all. So, authors must present results from a validation exercise here as well. Similarly, they mention model parameter calibration in Section 2.3, however, this part does not mean much without prior description of these parameters. For instance, what is NSE and if RMSE of 0.92 is low or high? Finally, they can summarize the calibration results in a table.

- They need to expand into why they choose Mellor and Yamada turbulence model. What makes this turbulence model appropriate for this study?

- Tide data is fundamental to this study. Instead throwing a single RMSE number on Line 190, please also provide the comparison of modeled vs. measured tide time series at various locations along the ME.

- In addition to Figures 4 and 5, authors can include figures where they compare results quantitatively like relative difference of the flow velocities. Their discussion in Section 3.1 based on these figures are very qualitative and hard to convince the readers. On line 244 they say, “tide velocities in ME increased by 25%–30%, reaching 0.4–0.6 m/s.”, however this is not apparent from the figures. Where did this somewhat dramatic increase occur specifically?

- When comparing flow velocities time series would be more useful. In addition to Figures 5 to 8, please show velocity comparisons from different locations along the ME and at various depths.

- Figure 6-7-8-9: Please indicate the location of the transect shown in these figures, where does it start and end?

- Section 3.2: Authors use only density stratification to explain the changes in hydrodynamics, but what about the changes in bathymetry. Pre-EROP and transition period bathymetry is very different from the present case. What is the impact of this on hydrodynamics in ME?

- Section 4.2: Pollutant modeling is not mentioned neither in the abstract nor in the introduction but suddenly appears on page 15. It is completely out-of-place and should be taken out.

- There is so much overlap between Section 4.1 and Chapter 5. I think Section 4.1 can benefit from more discussion about the physics behind the changes in the hydrodynamics due to human activity. Right now it just consists of merely the explanation of the model results.

Minor

- L55-L58: these two statements contradict with each other.

- L141 and Figure 3: What is the source of the bathymetry data? Is it available freely to other researchers?

- L145-L146: These stations are not identified in Figure 1.

- L177: These stations are not identified in Figure 1.

- L203: Why wind and wave affects are neglected? Provide an explanation.

- L251-L252: “Moreover, saltwater intrusion is enhanced due to the combined influence of human activities.”. This statement is not supported by any means in this section. How did you come up with this conclusion?

- Figures 4-5-6-7: The directions that the arrows indicate are not visible at all. Additionally, do these figures show the maximum current speeds or snapshot from a specific time step?

- Figures 4-5: Please add the coordinates to x and y axes of these.

- Figures 6-7: Sub-labels are missing.

- Figure 10: Figure is missing.

Comments for author File: Comments.pdf

Author Response

Response letter

We thank the two Reviewers for the careful consideration of our work. Their constructive and thoughtful comments and suggestions led to a much improved and complete revision of the manuscript. In the revised paper, we have addressed all the comments formulated by the Reviewers by replying (in black) to their remarks (in blue). The line numbers in this rebuttal refer to the revised version of the manuscript.

Responses to comments by Reviewer #1

Point 1:

There is so much previous work done on tidal-river hydrodynamics. Authors should expand the introduction to have a reasonable and a thorough literature review. Currently they only cite the ones that focus on PRD. One seminal example of which is by Hoiting and Jay (2016) and references therein (https://doi.org/10.1002/2015RG000507). Then the authors should more explicitly explain the novelty of their work and the advantages of their model compared with those methods reported elsewhere. 


Our reply: Thanks for the reviewer's suggestion. We have cited the literature of Hoiting and Jay(https://doi.org/10.1002/2015RG000507). We have also expanded the introduction to the goal of being a reasonable and a thorough literature review. Specific modifications are as follows.

Estuaries are the primary transitional zones between land and sea [1,2]. There are many researches on tidal river all over the world, such as the Amazon in Brazil, the Columbia, Fraser and Saint Lawrence in North America, the Yangtze and Pearl in China [3]. Among all these estuaries, the hydrodynamic process is controlled by both fluvial and tidal dynamics, leading to complex annual, monthly, and daily variation characteristics [4,5]. Moreover, large-scale anthropogenic activities (e.g., dam construction, land reclamation, sand excavation, and waterway dredging) significantly change the riverbed topography and estuarine coastlines [7,8]. These changes, which are different from those of natural evolution, will also cause corresponding adjustments in the hydrodynamic characteristics of the estuary [9]. For example, a study of Chesapeake Bay, eastern United States (U.S.), showed an increase of the residence time of freshwater and the penetration of up-stream salt water due to the sea-level rise [10]. The historic dredging and deepening of the navigation channel in the Caloosahatchee Estuary, Florida, U.S., has caused a significant system-wide increase in salt transport [11]. In china, the Pearl River Delta (PRD) is exactly a highly-urbanized delta that is affected by high-intensity human intervention [12]. (Line34-48)

 

Point 2:

Next, they need to provide more details about their “quasi-three-dimensional river-tide-salt dynamic model”. It would be helpful to learn more about the theoretical background (e.g., equations solved). Also, they need to discuss why they chose this model. Also, it is also not clear if model is validated at all. So, authors must present results from a validation exercise here as well. Similarly, they mention model parameter calibration in Section 2.3, however, this part does not mean much without prior description of these parameters. For instance, what is NSE and if RMSE of 0.92 is low or high? Finally, they can summarize the calibration results in a table.

Our reply: We thank the Reviewer for this comment. The solution equations of the model in this paper are relatively complex, which will affect the focus in this paper, including the continuity equation in sigma coordinate system, three-dimensional momentum equation and salinity transport equation. There is a detailed description of the equation in the reference (Hamrick, J.M., Wu, T.S., 1997. Computational design and optimization of the EFDC/HEM3D surface water hydrodynamic and eutrophication models. In: Delich, G.,Wheeler, M.F. (Eds.), Next Generation Environmental Models and Computational Methods. Society for Industrial and Applied Mathematics, Pennsylvania, pp. 143-161). The Environmental Fluid Dynamic Code (EFDC) model used in this paper was created by Hamrick, it has been successfully applied in environmental studies of the PRD and other estuarine systems worldwide.

We added tables of calibration of water level, current and salinity at stations in ME. We have published a paper on the study of the Modaomen Estuary by using the same model mentioned in the manuscript. The model has been fully calibrated and verified, and the calculation results are also very reliable. For the detailed process of calibration verification, please refer to the reference (Li, C.; Yu, M.; Cai, H.; Chen, X. Recent Changes in Hydrodynamic Characteristics of the Pearl River Delta during the Flood Peri-od and Associated Underlying Causes. Ocean & Coastal Management 2019, 179, 104814, https://doi.org/10.1016/j.ecss.2019.106345).

 

Point 3:

They need to expand into why they choose Mellor and Yamada turbulence model. What makes this turbulence model appropriate for this study?

Our reply: The Mellor and Yamada turbulence model was chosen by Hamrick  in the EFDC model. (Hamrick, J.M., Wu, T.S., 1997. Computational design and optimization of the EFDC/HEM3D surface water hydrodynamic and eutrophication models. In: Delich, G.,Wheeler, M.F. (Eds.), Next Generation Environmental Models and Computational Methods. Society for Industrial and Applied Mathematics, Pennsylvania, pp. 143-161). It can well simulate the hydrodynamic characteristics of three-dimensional stratified flow in an acceptable time range. It means that EFDC well balance the calculation time and calculation error, and has a good calculation effect for Estuarine gravity circulation. As mentioned in the article, this model has been successfully applied to other estuaries in the world, including PRD.

 

Point 4:

Tide data is fundamental to this study. Instead throwing a single RMSE number on Line 190, please also provide the comparison of modeled vs. measured tide time series at various locations along the ME.

Our reply: We have published a paper on the study of the Modaomen Estuary by using the same model mentioned in the manuscript. The model has been fully calibrated and verified, and the calculation results are also very reliable. For the detailed process of calibration verification, please refer to the reference (Li, C.; Yu, M.; Cai, H.; Chen, X. Recent Changes in Hydrodynamic Characteristics of the Pearl River Delta during the Flood Peri-od and Associated Underlying Causes. Ocean & Coastal Management 2019, 179, 104814, https://doi.org/10.1016/j.ecss.2019.106345). Here is an example in the reference, it provided the comparison of modeled vs. measured tide time series at various locations along the ME.

Fig Comparisons between the modeled and observed water levels at the stations in the Modaomen Estuary. The solid and dotted lines indicate model results and observations, respectively.

 

Point 5:

In addition to Figures 4 and 5, authors can include figures where they compare results quantitatively like relative difference of the flow velocities. Their discussion in Section 3.1 based on these figures are very qualitative and hard to convince the readers. On line 244 they say, “tide velocities in ME increased by 25%–30%, reaching 0.4–0.6 m/s.”, however this is not apparent from the figures. Where did this somewhat dramatic increase occur specifically?

Our reply: The dramatic changes occurred in four typical tidal channels of ME. They are located between Sanzao Island and the mainland, between Sanzao Island and Hengqin Island, and between Hengqin Islandand mainland (Figure. 4 (a) and 4 (b)). It has been clearly described in the article. (Line244-247)

 

Point 6:

When comparing flow velocities time series would be more useful. In addition to Figures 5 to 8, please show velocity comparisons from different locations along the ME and at various depths.

Our reply: Thanks for the reviewer's suggestion. Comparing flow velocities time series are very useful. But, here we are concerned with the changes of hydrodynamic structures in the whole horizontal plane and longitudinal section in this paper. More attention is paid to the gravity circulation pattern and velocity stratification structure. The velocity from different locations along the ME and at various depths very complex and changeable. We will discuss the relationships between divers human activities and the velocity from different locations along the ME and at various depths in the follow-up study.

 

Point 7:

 Figure 6-7-8-9: Please indicate the location of the transect shown in these figures, where does it start and end?

Our reply: The sub-lable of the longitudinal section has been added in Fig.1  lower right corner (as shown below). It starts at Dahengqin and ends at Zhuyin, nearly 40 km.

 

 

Point 8:

Section 3.2: Authors use only density stratification to explain the changes in hydrodynamics, but what about the changes in bathymetry. Pre-EROP and transition period bathymetry is very different from the present case. What is the impact of this on hydrodynamics in ME?

Our reply: According to the conclusion of the article, There is a seriousriverbed down-cutting in the of the present case than in the Pre-EROP and transition period. Thee riverbed down-cutting weakened the morphological resistance of the estuary, thus enhancing tidal dynamics. The velocities of the flood tide and ebb tide increased by 20–30%, and the flow discharge was increased by twice as much as before. Moreover, the length of the saltwater intrusion was increased by 10–20 km, which resulted in the intensification of the velocity stratification in the ME during the flood tide, causing the spatial scale of gravity circulation expanded 1–2 times.

The impact of the enhancement of riverbed down-cutting on tidal dynamics in the ME was greater than the weakening effect of land reclamation. Hence, due to the combined influence of human activities between the 1970s and 2010, the velocities of the flood tide and ebb tide increased by approximately 1.5 times. The hydrodynamic structures in the ME changed from a state of atypical gravity circulation, with imperceptible stratification, to a state of highly stratified and large-scale gravity circulation. The stratification of the neap tide was more significant than that of the spring tide, and the spatial scale of gravity circulation is larger. (Line513-526)

 

Point 9:

 Section 4.2: Pollutant modeling is not mentioned neither in the abstract nor in the introduction but suddenly appears on page 15. It is completely out-of-place and should be taken out.

Our reply: We thank the Reviewer for this comment. The relevant results of pollutant simulation have been added to the abstract and conclusion. And we added relevant descriptions to reveal why we discuss pollutant transport, as follows.

The water quality of estuaries is largely affected by the process of salt transport, which controls the intrusion of salt water, the longitudinal and transverse density gradi-ent, and the baroclinic circulation, and also influences the transport of sediment and nu-trients. The above analysis results show that the hydrodynamic structures in the ME changed from a state of atypical gravity circulation, with nonobvi-ous stratification, to a state of highly stratified and large-scale gravity circulation, due to the comprehensive in-fluence of human activities between the 1970s and 2010. Therefore, the pollutant transport process in the ME may also have undergone great changes. (Line457-464)

Point 10:

There is so much overlap between Section 4.1 and Chapter 5. I think Section 4.1 can benefit from more discussion about the physics behind the changes in the hydrodynamics due to human activity. Right now it just consists of merely the explanation of the model results.

Our reply: Thanks for the reviewer's suggestion. We reanalyzed Section 4.1, and the discussion on tidal range change and its causes is added as follows.

Tidal range is an important index to measure tidal dynamic intensity. In this study, the tidal range of three stations were calculated, representing the tidal dynamic in upstream (GZ), midstream (DA), and downstream (GDJ), respectively. Figure 10 shows the changes in the monthly average tidal range at each station. It indicates that the due to land reclamation, tidal range in the transition case is 0.04-0.18 m lower than that in the Pre-EROP Case. But, the riverbed down-cutting led that the tidal range in the present case increased by 0.08-0.23 m, compared with the transition case, it was also greater than the tidal range in the Pre-EROP Case. Besides, the negative effect of land reclamation on the tidal range was significant in the downstream, but not obvious in the midstream and upstream. However, the positive effect of riverbed down-cutting on the tidal range is prominent both in the middle and lower reaches. In summary, it found that whether from the aspect of tidal range value or the spatial influence range, the effect of terrain cutting was always stronger than that of reclamation, which is consistent with the analysis in section 3.1.

Cai et al.[5] pointed out that in the momentum equation of one-dimensional tidal cycle average, the residual water level gradient(RWLG) is balanced with the friction term( ). Here, the residual water level refers to the average water level within 25 hours of a lunar day. It means that RWLG can reflect the resistance of tidal wave propagation. Through calculation, the RWGL in the ME increased from1.0×10^-5 under Pre-EROP Case to 1.1×10^-5 under transition case, whereas it decreased to 0.8×10^-5 in the Present case. Hence, the reclamation increased the resistance of upstream propagation of tidal waves, which is conducive to the reduction of tidal range. Whereas, the riverbed down-cutting decreased the resistance of tidal waves propagation, which is conducive to the increase of tidal range. The result of the comprehensive action of human activities is net reduction of frictional resistance but net increase in the tidal range. (Line428-452)

 

Minor point:

 

Point 1: L55-L58: these two statements contradict with each other.

Our reply: Thanks for the reviewer's suggestion. In the statement “However, reclamation in the estuary has caused the mouth to continuously extend, thus weakening the tidal dynamics entering the inland and increasing saltwater intrusion”, we changed   “increasing saltwater intrusion” to   “decreasing saltwater intrusion”.

 

Point 2: L141 and Figure 3: What is the source of the bathymetry data? Is it available freely to other researchers? .

Our reply:  The terrain data was measured by Guangdong Waterway Bureau. It's confidential and can't be made public. So that's a pity for researchers who need this data.

 

Point 3: L145-L146: These stations are not identified in Figure 1.

Point 4: L177: These stations are not identified in Figure 1.

Our reply:  These stations have been identified in Figure 1.

 

 

Point 5: L203: Why wind and wave affects are neglected? Provide an explanation.

Our reply: We thank the Reviewer for this comment. Originally, what we want to express is that influence of human activities on hydrodynamic structure in a long time scale (several tens of years) in the paper. On one hand, wind and waves do have an impact on hydrodynamic processes, but their influence is short-lived. On the other hand, in order to control the variables to better distinguish the effects of reclamation and riverbed down-cutting, wind and wave affects are neglected.

 

Point 6: L251-L252: “Moreover, saltwater intrusion is enhanced due to the combined influence of human activities.”. This statement is not supported by any means in this section. How did you come up with this conclusion?

Our reply: We agree with the Reviewer for this comment. This is our negligence, since statement is not supported by any means in this section, it has been deleted in the revised paper.

 

Point 7: Figures 4-5-6-7: The directions that the arrows indicate are not visible at all. Additionally, do these figures show the maximum current speeds or snapshot from a specific time step?

Our reply: These figures has been replaced to a clearer vector version. When zoomed in, the arrow can be clearly seen. These figures show the snapshot from a specific time step, such as the time of flood/ebb tide.

 

Point 8: Figures 4-5: Please add the coordinates to x and y axes of these.

Our reply: X and y axes are not applicable to these figures. This is our negligence, since statement is not supported by any means in this section, it has been deleted in the revised paper. We added North pointing labels and scale bars to these pictures to achieve the same effect.

 

Point 9: Figures 6-7: Sub-labels are missing.

Our reply: We checked the picture carefully and there are no Sub-labels missing. Is the incorrect display caused by the problem of software version?

 

Point 10: Figure 10: Figure is missing.

Our reply: Awfully sorry, we missed this picture. Figure 10 has been added.

 

 

 

Author Response File: Author Response.docx

Reviewer 2 Report

• Abstract

The abstract is concise yet provides a good and useful overview of the paper.

• Introduction

The introduction is well written and introduces the reader to the topic. The literature review is adequate. The body part of the paper

The methodology is sound and solid and is clearly explained. The paper uses standard analysis tools.

• Conclusion

The results are well described and are of practical value. Discussions and conclusions are useful and solid.

• Reference

References are relevant.

General comments:

 The paper deals with a numerical modeling approach to understand the impacts of human activities on hydrodynamic structures. A curvilinear grid has been employed. The topic is well within the scope of the journal. The paper is well written and well organized. My suggestions are summarized in the following:

Governing equations need to be provided.

Which turbulence model has been used? Why?

Figure 3. Grid sensitivity analysis needs to be presented.

Explain the numerical method used in the model as well as the order of accuracy

Define all boundary conditions and initial conditions

Section 2.3 Model calibration and verification: summarize all error values in a table Add standard error measures such as R-squared.

Explain the benefits of a curvilinear grid compared to an unstructured grid

Author Response

Response letter

We thank the two Reviewers for the careful consideration of our work. Their constructive and thoughtful comments and suggestions led to a much improved and complete revision of the manuscript. In the revised paper, we have addressed all the comments formulated by the Reviewers by replying (in black) to their remarks (in blue). The line numbers in this rebuttal refer to the revised version of the manuscript.

Responses to comments by Reviewer #2

Point 1:

Governing equations need to be provided

 

Our reply: We thank the Reviewer for this comment. The solution equations of the model in this paper are relatively complex, which will affect the focus in this paper, including the continuity equation in sigma coordinate system, three-dimensional momentum equation and salinity transport equation. There is a detailed description of the equation in the reference (Hamrick, J.M., Wu, T.S., 1997. Computational design and optimization of the EFDC/HEM3D surface water hydrodynamic and eutrophication models. In: Delich, G.,Wheeler, M.F. (Eds.), Next Generation Environmental Models and Computational Methods. Society for Industrial and Applied Mathematics, Pennsylvania, pp. 143-161). The Environmental Fluid Dynamic Code (EFDC) model used in this paper was created by Hamrick, it has been successfully applied in environmental studies of the PRD and other estuarine systems worldwide.

 

Continuity equation:

           

        

Momentum equation:

X:

 

     

Y:

 

    

Z：

              

Point 1:

Which turbulence model has been used? Why?

 

Our reply: The Mellor and Yamada turbulence model mwas used in this pape. The Mellor and Yamada turbulence model was chosen by Hamrick  in the EFDC model. (Hamrick, J.M., Wu, T.S., 1997. Computational design and optimization of the EFDC/HEM3D surface water hydrodynamic and eutrophication models. In: Delich, G.,Wheeler, M.F. (Eds.), Next Generation Environmental Models and Computational Methods. Society for Industrial and Applied Mathematics, Pennsylvania, pp. 143-161). It can well simulate the hydrodynamic characteristics of three-dimensional stratified flow in an acceptable time range. It means that EFDC well balance the calculation time and calculation error, and has a good calculation effect for estuarine gravity circulation. As mentioned in the article, this model has been successfully applied to other estuaries in the world, including PRD.

 

Point 1:

Figure 3. Grid sensitivity analysis needs to be presented.

 

Our reply: For the plane grid, we initially made three sets of grids. The result error of the sparse grid is very large, and the calculation time of the dense grid is too long. Finally, we compromised and chose the medium-density grid on the premise of ensuring the accuracy of the model. The number of vertical layers of the grid is also determined by this method.

 

Point 1:

Explain the numerical method used in the model as well as the order of accuracy

 

Our reply: The computational scheme utilizes an external-internal mode splitting to solve the horizontal momentum equations and the continuity equation on a staggered grid. The external mode, associated with barotropic long wave motion, is solved using a semi-implicit three time level scheme with a periodic two time level correction. A multi-color successive over relaxation scheme is used to solve the resulting system of equations for the free surface displacement. The internal mode, associated with vertical shear of the horizontal velocity components is solved using a fractional step scheme combining an implicit step for the vertical shear terms with an explicit step for all other terms. The transport equations for the turbulence intensity, turbulence length scale, salinity, temperature, suspended sediment, dissolved and adsorbed contaminants, and dye tracer are also solved using a fractional step scheme with implicit vertical diffusion and explicit advection and horizontal diffusion. A number of alternate advection schemes are implemented in the code.

 

Point 1:

Define all boundary conditions and initial conditions.

Our reply: The upstream inlet boundary of the model adopts the average dry season flow of the Pearl River from 1950 to 2016, i.e. 2900 m3 / s in WZ, 600 m3 / s in SJ and 400 m3 / s in BL; The flow boundary condition is given as constant flow to eliminate the influence of flow change on brine movement; Given the tidal cycle process of two months and the constant salinity of 34 ppt at the downstream open sea boundary.The initial water level and velocity were set to 0, and the initial salinity field was adopted as the salinity distribution after 60 d of model operation under the given boundary conditions.

 

Point 1:

Section 2.3 Model calibration and verification: summarize all error values in a table Add standard error measures such as R-squared.

Our reply:Thank you for your suggestion. And we have added a table containing error values (see Line 206-209)

 

Point 1:

Explain the benefits of a curvilinear grid compared to an unstructured grid.

Our reply: Firstly, the river network accounts for a large proportion in the model area. For the river, the curvilinear grid can better represent the section characteristics. Secondly, the grid type is also related to the numerical method used in the model. For the numerical method of EFDC, the curvilinear grid is more suitable

 

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

Please add your responses for boundary conditions, numerical method, mesh sensitivity analysis to the paper.

Author Response

Please add your responses for boundary conditions, numerical method, mesh sensitivity analysis to the paper.

Response:

Thank you for your suggestions, we have added it to the paper accordingly.

Boudary conditions:

“Daily water discharges were specified as the inflow boundaries at the WZ, SJ, and BL stations. The water levels observed during the same period were specified at other upstream boundaries (i.e., LYG and SZ stations). The water levels at the offshore open boundary were influenced by the tidal elevations provided by the TPXO models. The inverse simulation model of ocean tides proposed by Egbert and Erofeeva [27], which uses eight principal tidal constituents (i.e., Q1, O1, P1, K1, N2, M2, S2, and K2) that account for most of the tidal energy in the South China Sea, was combined with the calibration of measured data for tidal harmonic analysis to obtain the hourly tidal level process needed for the opening boundary of the downstream offshore sea,. The incoming salinities at the offshore open boundary were specified as 34 ppt, which was also provided by WL-Delft Hydraulics [28] and Gong et al. [29]. The initial water level and velocity were set to 0, and the initial salinity field was adopted as the salinity distribution after 60 d of model operation under the given boundary condition”(see line 180-192)

 

Numerical method:

“The numerical scheme employed in EFDC to solve the equations of motion uses second order accurate spatial finite difference on a staggered or C grid. The model's time integration employs a second order accurate three time level, finite difference scheme with an internal-external mode splitting procedure to separate the internal shear or baroclinic mode from the external free surface gravity wave or barotropic mode.”(see line 148-153)

 

Mesh sensitivity analysis:

“ The grid sensitivity was tested. Initially, three sets of grids were prepared for the model, with the number of grids being 42475,75341,132873 respectively. The most sparse grid was discarded due to poor accuracy. The most dense grid took twice as long as the second set of grids, but the accuracy improvement was small. Therefore, the second set of grids was selected as the grid of the model.”(see line 167-171)