Assessment of Climate Change Impact on Future Groundwater-Level Behavior Using SWAT Groundwater-Consumption Function in Geum River Basin of South Korea
Round 1
Reviewer 1 Report
1. The originality, accuracy, and completeness of the work are satisfactory.
2. The authors are suggested to revise some typing errors in the manuscript. Besides, the arrangements of references should be consistent throughout the references list.
Author Response
Reviewer 1.
The originality, accuracy, and completeness of the work are satisfactory.
1) The authors are suggested to revise some typing errors in the manuscript. Besides, the arrangements of references should be consistent throughout the references list.
As the reviewer pointed out, the typing and references have been corrected. Thank you.
Author Response File: 
Author Response.docx
Reviewer 2 Report
The manuscript requires major revisions of the applied methods. In particular:
1) Equation (not numbered) at line 128 does not respect the dimensional analysis;
2) The estimation of the groundwater levels, by using the literature method was not fully described;
3) MODFLOW groundwater flow simulation, which can be very useful to address groundwater level estimations, was not taken in consideration by authors;
4) Calculations of the uncertainty of the forecasts were not carried out by authors;
5) Hydrogeologic descriptions of the subsoil and groundwaters were missing.
Author Response
Reviewer 2.
1) Equation (not numbered) at line 128 does not respect the dimensional analysis;
The data in the SWAT manual did not fit as you pointed out. Therefore, the unit was modified by referring to the literature of the hoghøt equation (1940).
Line 142
where Qgw is the groundwater or base runoff (mm/hr) introduced to the river each day, Ksat is the hydraulic conductivity of the aquifer (mm/day)
2) The estimation of the groundwater levels, by using the literature method was not fully described;
As the reviewers pointed out, the following text describes the process of converting SWAT simulation results to groundwater levels in line 148-153.
Line 148-153
SWAT does not currently print groundwater height in output file, the water table height is update daily by the model [35]. In this study, the result of SA_ST (shallow water storage, mm) were used, which is calculated as the initial value of the GWQMN (threshold water level in shallow aquifer for base flow, mm) parameter. Therefore, the groundwater level was taken from SA_ST (mm) minus GWQMN (mm) as the amount of groundwater level variation (mm), and the initial value of the monitoring groundwater level was added.
3) MODFLOW groundwater flow simulation, which can be very useful to address groundwater
As the reviewer points out, MODFLOW is very useful for ground water level analysis. Research using SWAT-MODFLOW has been conducted in Korea [10, 14, 15, 16]. But there is no case application to large basin as this study. This study was the third largest basin in South Korea, and not enough input data for the MODFLOW link gathering, so it was analyzed using SWAT only in this study.
Line 94-97
In this study, the SWAT used only, as there was not enough groundwater related DBs (i.e. stream bed thickness, hydraulic conductivity, riverbed filtration) that could be applied to large basins such as Geum river basin.
4) Calculations of the uncertainty of the forecasts were not carried out by authors;
The forecast data used in this study was provided and applied by national agencies (i.e. KMA, MOLIT), and the uncertainty of the data was not studied in this study. However, we have further described the uncertainty of the climate change scenario and the uncertainty of the groundwater use data you have pointed out.
Line 185-190
Groundwater use estimation methods have been dealt with by a number of researchers, and these methods have inherent uncertainties [36, 37, 38, 39]. Given the groundwater use estimating system and realistic conditions, it is necessary to estimate the groundwater use by raising the reliability of prediction of the utilization rate by making a reasonable method. However, as the purpose of this study was not to compare the accuracy of the data, we used the published national reports in South Korea.
Line 218-222
Climate change scenarios are subject to uncertainty due to limitations such as the selection of future virtual scenarios, incomplete physical understanding of various self-connections, and computational capabilities [44, 45, 46, 47]. Convenience corrections need to be made to minimize uncertainty in these climate change scenarios.
Line 225-226
and the precipitation and temperature were corrected conveniently using the quantile mapping and change factor techniques, respectively.
5) Hydrogeologic descriptions of the subsoil and groundwaters were missing.
As the reviewer pointed out, I have written about subsoil and groundwater descriptions. And Added subsoil distribution map in Figure 2.
Line 108-111
Geologically, Korea is comprised mainly of precambriam gneiss and Jurassic granite in the middle area. Quaternary alluvial deposits with high permeability and high groundwater yield (up to 800 m3/day) are mostly found along large rivers [7]. Alluvial aquifers are composed of unconsolidated sediments and are widely distributed within Geum river basins.
Line 160-162
As for the surface soil types in the Geum River Basin, the sandy loam and loam represented high proportions (58% and 24%, respectively). Sandy loam and loam occupy 42% and 27% respectively, of subsoil textures.
I have come to think once more about what the you have pointed out. Thank you.
Author Response File: Author Response.docx
Reviewer 3 Report
The authors present an approach of using SWAT model in combination with future climate and water consumption forecasts to derive information about the dynamics of groundwater levels in the subbasins of Geum Rier, South Korea. I think the developed approach is useful to improve future water resources management in South Korea making the paper a suitable addition to “Water”. However, major changes in the manuscript are required before publication. Below you can find my list of comments and questions.
Major Comments: I see critical scale issues by using SWAT for groundwater level estimations which need to be discussed by the authors:
1) I assume that SWAT gives one hydraulic head per time per HRU. Considering the size of an HRU, real hydraulic heads may show variations of several meters within a HRU, particularly in mountainous areas. In addition the authors validate their model using hydraulic heads from wells, which are point information only. How this scale discrepancy does impacts the meaningfulness of your results?
2) The authors include the impact of GWU in the SWAT model. Depending on the location of the pumps, groundwater abstraction forms local cones of depression which do not necessarily impact groundwater levels on the scale of an HRU and in particular do not influence measured groundwater levels in wells, if pumps are far away enough. Hence, how do the authors explain that including GWU seems to have an impact of the calibration and verification results for groundwater levels (Fig. 4)?
Minor comments:
L58 to 87: Why do you distinguish between domestic and overseas when talking about methodological approaches for groundwater assessment? I think, for our science, the applicability of approaches do not depend on country borders but may on other factors such as climate zones, etc. Please revise
L105-108: Numbers should be easily comparable. If precipitation is mm/a, a similar unit should be used for water consumption to see how much of total precipitation/recharge is used
L106. What is average Evapotranspiration in the basin?
L131: The parameter Alpha_gw has a different name in table 2, please change
L131: How did you determine the initial hydraulic head (hwtbl,1-i) in the second equation?
L128-131: How do you retrieve these set of equations? Is it hard-coded in SWAT?
L128-131: How to you calculate the parameters KSat and µ?. Why haven’t they been included in the calibration process? In particular, KSat shows large variations between different hydrogeological units, do you use different values for different HRU?
L141: Please add the reference to the bibliography
L148-151: How did you interpolate/extrapolate from the nine weather station to retrieve data for each HRU?
L156-157: It is, at least to me, unclear whether you finally used the “empirical equations” or the “groundwater values measured” for you analysis. Please revise
L193: Which spatial resolution does the HadGEM3-RA dataset has, please add information
L330: You mentioned several times that the aim is to have “reliable” simulations. What does make your approach more reliable than others?
L337ff: Do you include groundwater use changes in the climate change scenarios? How much do future results depend on either GWU changes or climate changes
Table5 and Figure 6 seems to show the same content. Please select one (I prefer the figure)
Best regards
Author Response
Reviewer 3
Major Comments: I see critical scale issues by using SWAT for groundwater level estimations which need to be discussed by the authors:
1) I assume that SWAT gives one hydraulic head per time per HRU. Considering the size of an HRU, real hydraulic heads may show variations of several meters within a HRU, particularly in mountainous areas. In addition the authors validate their model using hydraulic heads from wells, which are point information only. How this scale discrepancy does impacts the meaningfulness of your results?
In this study, the details of the radius influence of groundwater could not be analyzed in detail. However, threshold of the HRU values were designated as 0% so that they could be divided in as much detail as possible. The average area per HRU was 0.3 km2 (the number of HRU is 27,173), it split within the radius influence of South Korea's ground water (0.5 km). In the future, we try to more detailed analysis by SWAT-MODSIM.
We have added this in line 180-182.
2) The authors include the impact of GWU in the SWAT model. Depending on the location of the pumps, groundwater abstraction forms local cones of depression which do not necessarily impact groundwater levels on the scale of an HRU and in particular do not influence measured groundwater levels in wells, if pumps are far away enough. Hence, how do the authors explain that including GWU seems to have an impact of the calibration and verification results for groundwater levels?. (Fig. 4)?
The 1:1 graph on the right of Figure 4 shows the results of before and after GWU input. Black line is not considered for groundwater use, and red line is considered for groundwater use. When considering the GWU, the simulated results of the ground water level were 0.3 cm higher than the observation data. On the other hand, when GWU not entered in SWAT, ground water level were 10.6 cm higher than the observation data. Therefore, the simulated value when considering the amount of groundwater use was more similar to the actual measurement.
We have added this in line 305-308.
Minor comments:
1) L58 to 87: Why do you distinguish between domestic and overseas when talking about methodological approaches for groundwater assessment? I think, for our science, the applicability of approaches do not depend on country borders but may on other factors such as climate zones, etc. Please revise
We modified what the reviewer pointed out.
2) L105-108: Numbers should be easily comparable. If precipitation is mm/a, a similar unit should be used for water consumption to see how much of total precipitation/recharge is used
We quote the units announced in the South Korea national report, it seems to be difficult to convert the units as indicated. please understand.
3) L106. What is average Evapotranspiration in the basin?
As the reviewers pointed out, the following text describes average ET in line 112-113.
Line 112-113
and the average evapotranspiration is 533.2 mm.
4) L131: The parameter Alpha_gw has a different name in table 2, please change
We modified what the reviewer pointed out.
Line 145
αgw is the baseflow recession constant
5) L131: How did you determine the initial hydraulic head (hwtbl,1-i) in the second equation?
As the reviewers pointed out, the following text describes the process of converting SWAT simulation results to groundwater levels in line 148-153.
Line 148-153
SWAT does not currently print groundwater height in output file, the water table height is update daily by the model [35]. In this study, the result of SA_ST (shallow water storage, mm) were used, which is calculated as the initial value of the GWQMN (threshold water level in shallow aquifer for base flow, mm) parameter. Therefore, the groundwater level was taken from SA_ST (mm) minus GWQMN (mm) as the amount of groundwater level variation (mm), and the initial value of the monitoring groundwater level was added.
6) L128-131: How do you retrieve these set of equations? Is it hard-coded in SWAT?
We didn't change the coding. This is a recalculation using SWAT's output. (see line 148-153)
7) L128-131: How to you calculate the parameters KSat and µ?. Why haven’t they been included in the calibration process? In particular, KSat shows large variations between different hydrogeological units, do you use different values for different HRU?
In this study, the calibration parameters were selected through sensitivity analysis. Sensitivity analysis determined that GW_SPYLD(µ) was used in the Geum river basin as it did not affect the calibration, and that the SOL_K (Ksat) parameter was used by entering information on Korean soil into the SWAT DB, so the current state was well reflected without adjustment.
8) L141: Please add the reference to the bibliography
We modified what the reviewer pointed out.
Line 153
9) L148-151: How did you interpolate/extrapolate from the nine weather station to retrieve data for each HRU?
As you know, SWAT includes the wxgen weather generator model to generate climate data or to fill in gaps in measured records. If user prefers a different weather generator, daily input values for the different weather parameters may be generated with an alternative model and formatted for input to SWAT. In this study, data of weather stations in the Geum river basin and adjacent areas were input. SWAT interpolates meteorological data of subbasins and HRU using Thissen method.
10) L156-157: It is, at least to me, unclear whether you finally used the “empirical equations” or the “groundwater values measured” for you analysis. Please revise
We modified what the reviewer pointed out.
Line 173-174
11) L193: Which spatial resolution does the HadGEM3-RA dataset has, please add information
We modified what the reviewer pointed out.
Line 224.
In this study, the HadGEM3-RA RCP (Representative Concentration Pathway) 4.5 and 8.5 scenario data (precipitation, lowest/highest temperatures, wind speed, and relative humidity) with 12.5 km resolution provided by the Korea Meteorological Administration (KMA) were used,
12) L330: You mentioned several times that the aim is to have “reliable” simulations. What does make your approach more reliable than others?
Generally, streamflow data are used for calibration of SWAT models. In this study, the reliability was secured through multi-point calibration of dam storage, dam inflow, weir storage, weir inflow and ground water levels in the basin.
13) L337ff: Do you include groundwater use changes in the climate change scenarios? How much do future results depend on either GWU changes or climate changes
The future groundwater use data were estimated using historical trends and climate change scenarios were not taken into in this data.
In this study, we analyzed only climate change + groundwater use changes in futuer. I will present the results of the futher study, considering only climate change or changes in groundwater use.
However, Figure 4 shows a 1:1 graph when considering the amount of ground water use when the SWAT model is calibrated using observed data.
14) Table5 and Figure 6 seems to show the same content. Please select one (I prefer the figure)
Figure 6 shows the monthly change patterns, and Table 5 shows the seasonal quantitative changes. Since we wrote the contents in the text by citing the table and the figure, it is a little trouble to delete it. please understand.
Thank you for your valuable review.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
The revised manuscript has been significantly improved. Thanks for the revision!
Author Response
Thank you :)
Reviewer 2 Report
The equation at line 138 does not respect dimensional analysis: The dimension of this equation is 1/time, whereas Q is L/time.
The reason why MODFLOW was not applied have to be reported in the manuscripts
Author Response
1) The equation at line 138 does not respect dimensional analysis: The dimension of this equation is 1/time, whereas Q is L/time.
We changed what the reviewer pointed out. The SWAT manual describes the following, but the details and initial values were checked and the formula was changed.
Line 138-144
where Qgw is the groundwater or base runoff (mm/day) introduced to the river each day, de is initial depth of water in the deep aquifer (mm), Ksat is the hydraulic conductivity of the aquifer (mm/day), Lgw is the distance from the split boundary for the groundwater system and the sub-watershed to the main river (m), hwtbl,i is the groundwater level on day i (m)
SWAT manual SWAT2009 theory p. 173 SWAT2012 in-output documentation p. 324
|
2) The reason why MODFLOW was not applied have to be reported in the manuscripts
We changed what the reviewer pointed out.
Line 94-97
The previous research using SWAT-MODFLOW has been conducted in Korea [10, 14, 15, 16], but there is no case application to large basin as this study. The study area of this study is the third largest basin in South Korea, and not enough input data (i.e. stream bed thickness, hydraulic conductivity, riverbed filtration) for the MODFLOW, so it was analyzed using SWAT only.
Thank you for your valuable review.
Author Response File: Author Response.docx
Reviewer 3 Report
Conducted Changes in manuscript improved the overall quality significantly:
However, being an hydrogeologist, I can not agree with the authors answer on my comment No 7:
Firstly, the authors state that the parameter µ (specific yield) does not impact calibration. How is this possible considering that the parameter appears in a linear form in the equation on line 141? Hence, I would like to see the result of your sensitivity analysis.
Secondly, the authors state that "at the SOL_K (Ksat) parameter was used by entering information on Korean soil into the SWAT DB, so the current state was well reflected without adjustment."
It is well known from the lterature that Ksat shows variations of at least one magnitude even within the same hydrogeological facies. How is this adressed in the korean soil database and in your model?
Lastly, I feel that the newly added lines of the revised manuscript are less well written considering English language and style. I suggest a moderate polishing of the language before publication.
Aside from my comments above, I think the manuscript is ready for publication and will be a good contribution to Water and the hydrological community
Author Response
Reviewer 3
1) However, being an hydrogeologist, I can not agree with the authors answer on my comment No 7: Firstly, the authors state that the parameter µ (specific yield) does not impact calibration. How is this possible considering that the parameter appears in a linear form in the equation on line 141? Hence, I would like to see the result of your sensitivity analysis.
In this study, the Alpha_BF (αgw) parameter reacted more sensitively than the GW_SPYLD(µ) parameter and was used for calibration. This figure shows sensitivity analysis results of Alpha_bf and GW_SPYLD.
2) The authors state that "at the SOL_K (Ksat) parameter was used by entering information on Korean soil into the SWAT DB, so the current state was well reflected without adjustment." It is well known from the lterature that Ksat shows variations of at least one magnitude even within the same hydrogeological facies. How is this adressed in the korean soil database and in your model?
In this study, regional soil information was obtained from the Rural Development Administration (RDA) (1:25,000) and based surface and subsoil layer. And 405 soil properties information was entered into the usersoil sheet in the SWAT2012.mdb file.
Details of soil properties information received can be found on the link.
http://www.fftc.agnet.org/library.php?func=view&style=type&id=20190417095342
The soil in this study was entered as a grid file with a resolution of 30 m×30 m, and the area per pixel is 0.0009 km2. In this study, the SOL_K parameter was not specially calibrated in this study because the HRU was divided into areas of about 0.3 km2 per unit.
3) Lastly, I feel that the newly added lines of the revised manuscript are less well written considering English language and style. I suggest a moderate polishing of the language before publication.
As the reviewer pointed out, we commissioned an expert to modify the English style.
Thank you for your valuable review.
Author Response File: Author Response.docx
