Assessing the Impact of Artificial Recharge Ponds on Hydrological Fluxes in an Irrigated Stream–Aquifer System

Round 1

Reviewer 1 Report

This manuscript shows interesting study about influence of recharge ponds to the assessing of the fluxes and states in the irrigated stream-aquifer system. I have no serious complaints to analysis because it presents relatively comprehensive study related to the different aspects of the recharge ponds influence to the such systems. However, before publication I think that authors need to answer to the few following serious questions:

1. Methodology uses MODFLOW for stream-aquifer irrigated system which is quite naive for such complex systems. Flux interaction between ponds, canals and streams with aquifers cannot be properly described only by recharge module of MODFLOW. Rather, it is more appropriate to use fully physically based models such us Parflow, Hydrogeosphere or Cathy which simulate coupling fluxes in the more "physical" way defining surface and subsurface flow and its interaction. Basically, flow solution in surface channels depends on flow subsurface solution and vice-versa and its interaction cannot be obtained without appropriate "dynamic" coupling between two domains, while recharge MODLOW module considers surface flow only via imposed boundary conditions to the subsurface domain.
2. It is quite important question in this study is this only some case study or original scientific paper. Please, define novelties of this manuscript?
3. Coupling between DeyCent and MODFLOW strongly depends on irrigation efficiency parameter. Please, describe this calibration parameter in more detailed way. Is it only non-physical calibration parameter prone to calibration error or something else? 

Author Response

Reviewer #1

This manuscript shows interesting study about influence of recharge ponds to the assessing of the fluxes and states in the irrigated stream-aquifer system. I have no serious complaints to analysis because it presents relatively comprehensive study related to the different aspects of the recharge ponds influence to the such systems. However, before publication I think that authors need to answer to the few following serious questions:

Response: Thank you for the comments. We have responded to each comment below.

1. Methodology uses MODFLOW for stream-aquifer irrigated system which is quite naive for such complex systems. Flux interaction between ponds, canals and streams with aquifers cannot be properly described only by recharge module of MODFLOW. Rather, it is more appropriate to use fully physically based models such us Parflow, Hydrogeosphere or Cathy which simulate coupling fluxes in the more "physical" way defining surface and subsurface flow and its interaction. Basically, flow solution in surface channels depends on flow subsurface solution and vice-versa and its interaction cannot be obtained without appropriate "dynamic" coupling between two domains, while recharge MODLOW module considers surface flow only via imposed boundary conditions to the subsurface domain.

Response: We agree that fully coupled models are ideal for many hydrologic modeling situations. However, MODFLOW and its related packages have also been used extensively (and been tested and corroborated) to satisfy groundwater head and flow conditions in coupled stream-aquifer systems. Each model solves the water balance for respective systems (soil system, groundwater system, stream/channel system), and fully coupled models like those listed in this comment also use groundwater flux values and groundwater head values as boundary conditions for streams, and vice versa. We neglected to state that we used the Stream Flow Routing (SFR) package of MODFLOW to simulate stream flow and stream-aquifer water exchange along the South Platte River and Beebe Draw, as was described in the original paper that describes the MODFLOW model construction and testing (Deng and Bailey, 2020, reference #10 in this paper). We have now included a description of this in the manuscript:

Lines 197-201:

“Streamflow and stream-aquifer water exchange for the South Platte River (western boundary of the model domain) and Beebe Draw (eastern side of domain) are simulated using the Stream Flow Routing (SFR) package, using a stream water budget and Manning’s equation to compute the stream flow rate and depth. The width of the South Platte River was set to 45 m, and the width of Beebe Draw was set to 6 m.”

2. It is quite important question in this study is this only some case study or original scientific paper. Please, define novelties of this manuscript?

Response: Lines 67-69 outlined the novelty of the study, i.e. the gaps in knowledge from previous studies: “While these studies demonstrate the influence of artificial recharge on a groundwater system and associated fluxes, none quantify the influence of existing recharge ponds using historical recharge pond volumes and pumping rates.”. To make this novelty clear, we have modified the first sentence of the last paragraph of the Introduction to read:

Lines 70-72:

“To satisfy this gap in recharge pond understanding, the objective of this study is to quantify the influence of seepage from existing recharge ponds on groundwater system-response variables and fluxes in a highly managed irrigated stream-aquifer system.”

3. Coupling between DeyCent and MODFLOW strongly depends on irrigation efficiency parameter. Please, describe this calibration parameter in more detailed way. Is it only non-physical calibration parameter prone to calibration error or something else? 

Response: We have added the following sentence to the description:

Lines 272-273:

“Simulations were run successively until a best fit, in terms of Root Mean Square Error (RMSE), was found.”

Results are shown in Figure 6A (to determine the optimal irrigation efficiency), with remaining testing results (1:1 plot, time series plots of observed and simulated head) also shown in Figure 6.

Reviewer 2 Report

This paper can be considered as an interesting scientific work presenting a quite well documented research regarding the assessment of the impact of artificial recharge ponds on the groundwater system of the Gilcrest/La Salle area in the South Platte River Basin, Colorado, by introducing a new coupled model that links the MODFLOW groundwater flow model and the DayCent agronomic hydrologic model. The subject is within the topics of the Hydrology Journal. The manuscript is clearly written following a structure that contains analysis and elaboration results documented and presented in a quite informative, reliable and explanatory way. My recommendation is that the manuscript should be accepted for publication in its present form.

Author Response

Reviewer #2

This paper can be considered as an interesting scientific work presenting a quite well documented research regarding the assessment of the impact of artificial recharge ponds on the groundwater system of the Gilcrest/La Salle area in the South Platte River Basin, Colorado, by introducing a new coupled model that links the MODFLOW groundwater flow model and the DayCent agronomic hydrologic model. The subject is within the topics of the Hydrology Journal. The manuscript is clearly written following a structure that contains analysis and elaboration results documented and presented in a quite informative, reliable and explanatory way. My recommendation is that the manuscript should be accepted for publication in its present form.

Response: Thank you for the kind comments.

Reviewer 3 Report

The manuscript investigates the impact of the artificial recharge ponds in underline aquifers and how it modifies the baseline groundwater gradients or alters natural hydrologic fluxes. 

The study area is situated in an irrigated region northeast of Denver, Colorado, with an area of 246 km^2. The numerical method used is MODFLOW to simulate groundwater heads, groundwater storage, sources, and sinks, and DayCent to estimate spatio-temporal recharge patterns to the water table.

I have some comments/questions.

How the heterogeneous geology mentioned in lines 155-156 has been incorporated into the numerical simulations.

Need a table with all the parameters used in the numerical simulations and justify the assignment values. Are they experimental?

I do not see how both simulation models are related to the data. It is seen that they are independent one from the other. 

The area of the cells is quite large (see lines 145-146). How has the grid dimension been fixed?

How was the vertical discretization assigned? 

How are the results validated? How was the vertical discretization assigned? How are the results validated? Specify the validation of your assignment to emphasize the results are precise. Add any sort of comparison with known results. 

Mention the assumption/limitation of the model if you have not considered any factor.

Some of the characters of the figures are difficult to read. Please increase the numbers in the axes.

Validate results of Fig. 6.

Author Response

Reviewer #3

The manuscript investigates the impact of the artificial recharge ponds in underline aquifers and how it modifies the baseline groundwater gradients or alters natural hydrologic fluxes. 

The study area is situated in an irrigated region northeast of Denver, Colorado, with an area of 246 km^2. The numerical method used is MODFLOW to simulate groundwater heads, groundwater storage, sources, and sinks, and DayCent to estimate spatio-temporal recharge patterns to the water table.

I have some comments/questions.

How the heterogeneous geology mentioned in lines 155-156 has been incorporated into the numerical simulations.

Response: Thank you. We have modified the sentences to read:

Lines 164-169:

“In the previous modeling study for this region [10], a 3D aquifer material map was developed by interpolating material data from 450 boreholes, using the Kriging method. The values of hydraulic conductivity and specific yield for each material type (clay, clay & silt, silt, silt & sand, sand, sand & gravel, gravel) were estimated via model calibration in the previous study [10], with values assigned to cells within each of the 10 layers of the MODFLOW grid. Figure 3 shows the estimated material map for layers near the top, middle, and bottom portions of the aquifer. The aquifer material…”

Need a table with all the parameters used in the numerical simulations and justify the assignment values. Are they experimental?

Response: This table is provided in the previous paper (Deng and Bailey, 2020, reference #10 in this paper). We have referred readers to this table in the text.

I do not see how both simulation models are related to the data. It is seen that they are independent one from the other. 

Response: We are not sure what the reviewer intends with this comment. The DayCent model provides recharge to the MODFLOW model, with groundwater head being very sensitive to these recharge rates (see Figure 6A for the impact of irrigation efficiency). Therefore, there is a strong connection between these two models, and their impact on hydrologic fluxes and state variables (i.e. groundwater head) that can be tested by comparison to measured data.

The area of the cells is quite large (see lines 145-146). How has the grid dimension been fixed?

Response: We attempted to make the cell size approximately the size of a typical farm in the area, since each DayCent model would be simulating soil water balance and crop yield for a single farm area.

How was the vertical discretization assigned? 

Response: The thickness of the layers ranges from 1 to 3 m, with the thinner layers near the top of the model where recharge occurs. This is explained in Deng and Bailey (2020), but now we have included this description also in this manuscript:

Lines 151-153:

“Layer thickness ranges from 1 to 10 m, with the thinnest layers near the top of the model, for more accuracy in recharge exchange between DayCent and MODFLOW.”

How are the results validated? How was the vertical discretization assigned? How are the results validated? Specify the validation of your assignment to emphasize the results are precise. Add any sort of comparison with known results. 

Response: Thank you. We have include corroboration results in Figure 6, for both general results (1:1 plot in Figure 6B) and time series results in Figures 6C and 6D for selected monitoring well locations.

Mention the assumption/limitation of the model if you have not considered any factor.

Response: Thank you. We have included the major assumption in Section 4:

Lines 396-399:

“In this model application, we assume that water exchange between the soil profile and the water table is uni-directional (downward). If exchange in both directions is desired, the reader is referred to a modeling study [8] that tightly couples DayCent and MODFLOW.”

Some of the characters of the figures are difficult to read. Please increase the numbers in the axes.

Response: Thank you. We have increased the size, for easier reading.

Validate results of Fig. 6.

Response: We have provided the testing results for groundwater head in the text preceding the figure.

Round 2

Reviewer 3 Report

The paper has improved, and I am suggesting acceptance after the authors provide a table with all the parameters used in the numerical simulations requested in a previous report and justify the assignment values. Are they experimental? 
In this way, the reader does not need to find them in another paper, and the paper itself is complete.

Author Response

Thank you. We have now included a table that shows the values of hydraulic conductivity, specific yield, and specific storage used in the model simulations.