Sensitivity of Potential Groundwater Recharge to Projected Climate Change Scenarios: A Site-Specific Study in the Nebraska Sand Hills, USA

Round 1

Reviewer 1 Report

The manuscript is to investigate the response of potential groundwater recharge to the climate change scenarios using a case study of Nebraska Sand Hills. The overall approach and structure are reasonable as a scenario-based study, and enough prediction models are adopted in the analysis. Hydrus 1-D model was adopted for the groundwater simulation, but its modeling details are minimal. The interception and overland flow components are ignored in the water balance. This assumption may be acceptable as the sum of actual ET and the potential GR is close to P. Although there is not much new or significance of the study (as a good number of estimated were compared to the past studies), it would be a good one to check for water resource managers especially in the mid-west of US. There are some comments that need to be addressed for publication as below.

Pg. 4-5: Table 1A is listed as a table in appendix, which does not exist. I got the table separately, which is essential to understand the overall values in the analysis. Table 1 only lists the names of models. Probably you can just put Table 1A as Table 1 and define the acronyms in the same table, in the context, or even in a caption. Showing Table 1A is more important than Table 1. 

Pg. 6 Ln 235-242: According to Figure 2, GR from RCP 6.0 looks skewed. Medians decreases more linearly. Analyzing medians might be more reasonable than means in this case?

Pg. 7 Ln 255-261: This paragraph discusses normality, but I am not sure what it tells. Maybe authors can discuss bit more about what normality tells in the analysis. 

Pg. 9 Ln 307: Drop the numbers in the model acronyms. Table 1 only used CSIRO, GFDL_M1 , and MPI_MR. Better be consistent.   

Pg. 10: GR distribution in Figure 5 shows more normal shape while GR in Figure 3 shows more skewness. In the context, GR also used log-transformed analysis. Is there any reason or any thoughts why both cases show a different shape of distribution?

Pg. 10 Ln 322.: It says the three P series appears to be close to each other, but to me the dry P and the wet P are quite different. Hard to understand why they are considered to be close each other. 

Author Response

Reply to Reviewer #1

The manuscript is to investigate the response of potential groundwater recharge to the climate change scenarios using a case study of Nebraska Sand Hills. The overall approach and structure are reasonable as a scenario-based study, and enough prediction models are adopted in the analysis. Hydrus 1-D model was adopted for the groundwater simulation, but its modeling details are minimal. The interception and overland flow components are ignored in the water balance. This assumption may be acceptable as the sum of actual ET and the potential GR is close to P. Although there is not much new or significance of the study (as a good number of estimated were compared to the past studies), it would be a good one to check for water resource managers especially in the mid-west of US. There are some comments that need to be addressed for publication as below.

REPLY- We thank Rev.#1 for the comments to our work. Below we reply to the concerns raised by the reviewer and incorporate her/his edits and suggestions where appropriate.

Pg. 4-5: Table 1A is listed as a table in appendix, which does not exist. I got the table separately, which is essential to understand the overall values in the analysis. Table 1 only lists the names of models. Probably you can just put Table 1A as Table 1 and define the acronyms in the same table, in the context, or even in a caption. Showing Table 1A is more important than Table 1. 

REPLY- We apologize for failing to upload Table 1A during the manuscript submission. Table 1A was uploaded as supporting information later during the manuscript review process.

Pg. 6 Ln 235-242: According to Figure 2, GR from RCP 6.0 looks skewed. Medians decreases more linearly. Analyzing medians might be more reasonable than means in this case?

REPLY – Please note that we are using the t-distribution (as an alternative to the normal distribution) to estimate the expected value (m) of a continuously distributed population by using Eq. [1] (line 180) when sample size is small (16 projections in RCP2.6, 18 projections in RCP4.5, 12 projections in RCP6.0, 19 projections in RCP8.5). We concur that when sample sizes are relatively small, the ideal distribution can be distorted (the 12-GR values in RCP6.0 indeed are skewed). In order to clarify this part, we modified lines 176-181, 193-198 and we added reference to Eq.[1] in line 215,229 and 243.

 

Pg. 7 Ln 255-261: This paragraph discusses normality, but I am not sure what it tells. Maybe authors can discuss bit more about what normality tells in the analysis. 

REPLY – We removed the paragraph to avoid confusion.

 

Pg. 9 Ln 307: Drop the numbers in the model acronyms. Table 1 only used CSIRO, GFDL_M1 , and MPI_MR. Better be consistent.   

REPLY – We agree with this comment, thanks for pointing it out. We clarified by mentioning only the acronyms (consistent with Table 1) and added the RCP classes in parenthesis (lines 312-313).

Pg. 10: GR distribution in Figure 5 shows more normal shape while GR in Figure 3 shows more skewness. In the context, GR also used log-transformed analysis. Is there any reason or any thoughts why both cases show a different shape of distribution?

REPLY – Yes, GR-values in Figure 3 are annual values that are lognormally distributed. In Figure 5, we show the cumulative sums of GR rates. According to the central limit theorem, the probability distribution of the sum approaches the normal distribution.

Pg. 10 Ln 322.: It says the three P series appears to be close to each other, but to me the dry P and the wet P are quite different. Hard to understand why they are considered to be close each other. 

REPLY – We rephrased 327-329 and report the new sentence: “While the projected multi-decadal-averages of P are 54.0 cm yr^-1, 59.4 cm yr^-1, 58.9 cm yr^-1, in dry, mean, wet scenarios, respectively, the three ET₀ time series are characterized by time-variant increasing rates as predicted in other semi-arid environments”.

 

 

Author Response File: Author Response.docx

Reviewer 2 Report

The results of the investigations show possible trends in climatic changes in groundwater recharge (GR) in semi-arid regions using the example of Nebraska Sand Hills. Analysis of all forecast climate scenarios shows an unambiguous decreasing of GR, which is associated with an increase in potential evapotranspiration with an almost stable amount of precipitation.Results of root water stress analysis are very interesting because they showed a significant increase in the probability of wilting point soil moisture conditions in all forecast climate scenarios. However, it should be noted that the analysis does not take into account possible climatic changes in air humidity and wind conditions, which significantly affect the processes of evapotranspiration. For example, our studies conducted for the European part of Russia show that in the modern period the increase in air temperatures is accompanied by a decrease in wind speed in the surface layer. This significantly reduces the increase in potential evapotranspiration. However, long-term regional forecasting of these processes is problematic and requires additional research.

There are two minor comments on the figures in the manuscript.

 In Fig. 2, it seems reasonable to draw a line showing historical values of AI and GR. In this case, the explanations in the lines 235-242 will be clearer

Fig.7 shows monthly averages of P, ET and GR under three climate projections, which are compared with historical scenario. In this case, it seems reasonable to show the percentage changes in the figures b,c,d as compared to historical monthly averages

Author Response

Reply to Reviewer #2

The results of the investigations show possible trends in climatic changes in groundwater recharge (GR) in semi-arid regions using the example of Nebraska Sand Hills. Analysis of all forecast climate scenarios shows an unambiguous decreasing of GR, which is associated with an increase in potential evapotranspiration with an almost stable amount of precipitation. Results of root water stress analysis are very interesting because they showed a significant increase in the probability of wilting point soil moisture conditions in all forecast climate scenarios. However, it should be noted that the analysis does not take into account possible climatic changes in air humidity and wind conditions, which significantly affect the processes of evapotranspiration. For example, our studies conducted for the European part of Russia show that in the modern period the increase in air temperatures is accompanied by a decrease in wind speed in the surface layer. This significantly reduces the increase in potential evapotranspiration. However, long-term regional forecasting of these processes is problematic and requires additional research.

REPLY- We greatly appreciate Rev.#2 for the comments and suggested edits. We agree with the reviewer that availability of weather data would have been extremely useful for properly assessing ET₀. Unfortunately, availability of weather data was limited and were compelled to use a temperature-based approach to calculate ET₀.

There are two minor comments on the figures in the manuscript.

 In Fig. 2, it seems reasonable to draw a line showing historical values of AI and GR. In this case, the explanations in the lines 235-242 will be clearer

REPLY- We agree with this comment and thanks for this suggestion. We modified Fig. 2 accordingly.

Fig.7 shows monthly averages of P, ET and GR under three climate projections, which are compared with historical scenario. In this case, it seems reasonable to show the percentage changes in the figures b,c,d as compared to historical monthly averages

REPLY- While we agree that displaying the figure in percentages provides a comparative assessment, it may not depict the P, ET, and GR relationship well enough. Thus, we have chosen to keep the figure that displays the monthly values.

Author Response File: Author Response.docx