Spatial and Temporal Characteristics of Water Use Efficiency in Typical Ecosystems on the Loess Plateau in the Last 20 Years, with Drivers and Implications for Ecological Restoration
, Dacheng Wang 2,*, Shudong Wang 2, Hongsen Wang 1, Xiaojing Yao 2 and Shenshen Li 2
Round 1
Reviewer 1 Report
The authors present the trend of WUE changes for different vegetation types in the Loess Plateau from 2001 to 2020 and the correlations with NDVI, precipitation, and temperature values using a trend analysis method and correlation coefficients. Although their research has important reasoning for increasing understanding over the carbon and water coupling mechanisms of different vegetation types and allocation of regional water resources, the results aren’t insightful or novel in anyway. Moreover, the writing, organization, and presentation of results are severely lacking. I provide more detailed comments below, however I think considerable revision is needed before a proper evaluation can be completed.
A few major points include:
It is ok to use vegetation indices to study carbon-water links, but NDVI has so many limitations (which should be named in the article), that I would recommend adding other vegetation indices in the analysis as well. Maybe add NIRv and SIF to make a stronger argument. Especially, “The NDVI was the dominant factor affecting the spatial and temporal variations in WUE in the Loess Plateau”. Why is that? NDVI isn’t an intrinsic property of the vegetation, it is rather just a vegetation index measuring greening. This isn’t insightful or novel. A better analysis would deepen the reasoning for why NDVI controls WUE. What is the physiological reason for that? Does the vegetation have higher Leaf area index? Higher Vcmax? Higher leaf Nitrogen content? Higher LMA? All these variables should be brought up to the manuscript. I wonder if this would become a stronger paper if these observations were used.
The title does not really come across the content of the paper apart from being too long. There’s nothing about implications for ecological restoration. This is misleading.
I acknowledge the authors have tried to present some other metrics for drought, but they are not really tight together with the remaining results.
The figures are not good enough, or clear enough.
Minor comments:
The title is too long. Try to be concise, conveying the same message.
Line 18: WUE was already defined. Use the acronym after defining it. What do you mean by driving factors of the Loess Plateau? Photosynthesis? Please clarify.
Line 23: The yellow Plateau? Be consistent.
Line 29: the size of the WUE response rate? What does that mean?
Line 31: a-1? You mean yr-1? Please correct this throughout.
Line 35: precipitation temperature? Are you tracking the temperature of precipitation? Do you mean something else? What is a “facilitating effect”?
Line 37: What is threshold effect?
Line 47: WUE was already defined. Use the acronym after defining it.
Line 56: This should be the beginning of the paragraph below.
Line 61: Scouring action? I believe this is poor translation.
Line 89: WUE was already defined. Use the acronym after defining it.
Line 96: Numerous studies? Which ones?
Figure 1: What is DEM?
Figure 2: It is really hard to see the changes. Maybe show only the differences.
Figure 3: The color palette of Fig3b should be equally distributed with 0 in white. It is impossible to see changes. Why is NDVI going up?
Figure 4: Show the names on the figures as well (i.e., summer etc)
Figure 5: same as figure 4.
Figure 6: WUE growth rate should be indicated in the figure. Use a better divergent color palette. It is hard to see what is negative, what is zero, and what is positive. Also, use a more inclusive color palette, consider colorblind friendly.
Figure 7: Growth rate of what? Explicitly say on axis and name the plots.
Figure 8: This figure needs lots of work. Name the columns. You want your figures to convey the message clearly without the need to read the legend.
Figure 9: Same as figure 8. Everything is very small and hard to see. This figure isn’t very informative either. Try to come up with a better way to show this.
Conclusion is just repetition of the paper. Discussion is shallow.
Line 432: “The NDVI was the dominant factor affecting the spatial and temporal variations in WUE in the Loess Plateau”. Why is that? NDVI isn’t an intrinsic property of the vegetation, it is rather just a vegetation index measuring greening. This isn’t insightful or novel. A better analysis would deepen the reasoning for why NDVI controls WUE? What is the physiological reason? Does the vegetation have higher Leaf area index? Higher Vcmax? Higher Nitrogen content?
Author Response
Response to reviewer 1
Dear anonymous reviewer,
Thank you very much for your enthusiastic help and valuable comments.
- The title is too long. Try to be concise, conveying the same message.
Response/action: Thank you for your scrupulous reminding. I have changed the title of my dissertation to “Evolution and Influencing Factors of Water Use Efficiency in Loess Plateau ecosystems driven by drought”
2.Line 18: WUE was already defined. Use the acronym after defining it. What do you mean by driving factors of the Loess Plateau? Photosynthesis? Please clarify.
Response/action: Thank you for your scrupulous reminding. I have changed the water use efficiency to WUE throughout the text except for the first occurrence. In addition, the drivers are factors that can affect WUE. in this paper, we only study the effects of NDVI, temperature, and precipitation on WUE.
3.Line 23: The yellow Plateau? Be consistent.
Response/action: Thank you for your scrupulous reminding. I have corrected the yellow plateau in line 23 to Loess Plateau
4.Line 29: the size of the WUE response rate? What does that mean?
Response/action: Thank you for your scrupulous reminding. Line 29 of this paper is to express the growth rate of WUE, which has been corrected in the original text.
5.Line 31: a-1? You mean yr-1? Please correct this throughout.
Response/action: Thank you for your scrupulous reminding. I have corrected all units representing annual changes in this paper to yr-1
6.Line 35: precipitation temperature? Are you tracking the temperature of precipitation? Do you mean something else? What is a “facilitating effect”?
Response/action: Thank you for your scrupulous reminding. In line 35 refers to precipitation and temperature, which has been corrected in the text. In addition, I have corrected "facilitating effect" to "positive effect". It refers to the positive correlation, when the temperature increases, the WUE will also increase
7.Line 37: What is threshold effect?
Response/action: Thank you for your scrupulous reminding. I have changed the original text to read that there is a threshold effect. Here the threshold effect means that WUE is more sensitive in the CWSI of 0.3-0.4 interval.
8.Line 47: WUE was already defined. Use the acronym after defining it.
Response/action: Thank you for your scrupulous reminding. I have changed the water use efficiency to WUE throughout the text except for the first occurrence.
9.Line 56: This should be the beginning of the paragraph below.
Response/action: Thank you for your scrupulous reminding. I have changed the place where you said it to the opening sentence of the next paragraph.
10.Line 61: Scouring action? I believe this is poor translation.
Response/action: Thank you for your scrupulous reminding. I have removed the incorrect translation to ensure the accuracy of the expression.
11.Line 89: WUE was already defined. Use the acronym after defining it.
Response/action: Thank you for your scrupulous reminding. I have changed the water use efficiency to WUE throughout the text except for the first occurrence.
12.Line 96: Numerous studies? Which ones?
Response/action: Thank you for your scrupulous reminding. I have reordered the citations to ensure that they follow the statement
13.Figure 1: What is DEM?
Response/action: Thank you for your scrupulous reminding. DEM stands for Digital Elevation Model, which represents the elevation information of the ground.
14.Figure 2: It is really hard to see the changes. Maybe show only the differences.
Response/action: Thank you for your scrupulous reminding. Because the change is not very obvious from the figure, I have Table 2 after Figure 2 for area statistics, in order to better see the difference
15.Figure 3: The color palette of Fig3b should be equally distributed with 0 in white. It is impossible to see changes. Why is NDVI going up?
Response/action: Thank you for your scrupulous reminding. The four graphs in Figure 3 are designed to reflect the changes in NDVI of the Loess Plateau from different dimensions, the change in Figure 3b represents an interval, not a zero value, the color scheme is the official color scheme of ARCGIS, and Figure 3c is a statistical statistical chart drawn from statistical data
16.Figure 4: Show the names on the figures as well (i.e., summer etc)
Response/action: Thank you for your scrupulous reminding. I have added the appropriate description to the image.
17.Figure 5: same as figure 4.
Response/action: Thank you for your scrupulous reminding. I have added the appropriate description to the image.
18.Figure 6: WUE growth rate should be indicated in the figure. Use a better divergent color palette. It is hard to see what is negative, what is zero, and what is positive. Also, use a more inclusive color palette, consider colorblind friendly.
Response/action: Thank you for your scrupulous reminding. I have reviewed a large amount of literature and many scholars are using this set of color scheme to indicate increase and decrease, with the legend you can see that decrease is indicated by red, orange yellow, and increase is indicated by green, blue. This is the color scheme that I can see and reflect the problem after using many sets of color schemes
19.Figure 7: Growth rate of what? Explicitly say on axis and name the plots.
Response/action: Thank you for your scrupulous reminding. I have modified the text in the diagram to clarify the meaning
20.Figure 8: This figure needs lots of work. Name the columns. You want your figures to convey the message clearly without the need to read the legend.
Response/action: Thank you for your scrupulous reminding. I have added labels next to the pictures for the reader's convenience.
21.Figure 9: Same as figure 8. Everything is very small and hard to see. This figure isn’t very informative either. Try to come up with a better way to show this.
Response/action: Thank you for your scrupulous reminding. This graph is to show the changes of different vegetation types in different seasons. I have shown each category in the same range in this graph, so as to clearly compare the differences in each season and see the differences at a glance.
22.Conclusion is just repetition of the paper. Discussion is shallow.
Response/action: Thank you for your scrupulous reminding. My discussion is divided into two parts for elaboration. The first part is to explain the reasons for the differences in WUE in the Loess Plateau in different seasons, and the reasons for the differences in different vegetation in the same season. The second part elaborates the changes in the sensitivity of the influencing factors of WUE with the increase of the drought index and the reasons, and I think I elaborated in more detail
23.Line 432: “The NDVI was the dominant factor affecting the spatial and temporal variations in WUE in the Loess Plateau”. Why is that? NDVI isn’t an intrinsic property of the vegetation, it is rather just a vegetation index measuring greening. This isn’t insightful or novel. A better analysis would deepen the reasoning for why NDVI controls WUE? What is the physiological reason? Does the vegetation have higher Leaf area index? Higher Vcmax? Higher Nitrogen content?
Response/action: Thank you for your scrupulous reminding. As you can see from Figure 8 in the paper, the correlation between NDVI and WUE is the strongest (look at the legend, the areas that are the darkest and pass the significance test are also the most numerous), so NDVI is the main driver of WUE, which is the conclusion I see in most references. The innovation of my paper is not finding NDVI as the main driver of WUE, but finding in which interval can make the link between NDVI, precipitation, temperature and WUE stronger under the control of drought.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The mechanism of how drought impact WUE?
This paper investigates:
1) Spatial distribution of WUE and the growth rate of WUE;
2) WUE, growth rate of WUE classified by CWSI for different vegetation types;
3) Spatial distribution of the correlation between WUE and environmental factors (precipitation, temperature and NDVI);
4) Correlation of WUE and environmental factors classified by CWSI for four different vegetation types,
and tries to explain the different response of WUE and the correlation between environmental factors and WUE to CWSI of these four vegetation types. Overall, tons of cool and interesting results but little effort to dig into the mechanism of how the environmental factors impact WUE. The scientific question is not sharp enough. Please see the specific questions in the attached word doc.
Comments for author File: Comments.pdf
Author Response
Response to reviewer 2
Dear anonymous reviewer,
Thank you very much for your enthusiastic help and valuable comments.
- l) Line 90: Do you have any insight why the drought events can marginally raise WUE? And
could you add a sentence to explain the mechanism of it in the same line?
Response/action: Thank you for your scrupulous reminding. Because for scrub and forest, a slight drought for the vegetation can make them work to absorb groundwater and stimulate the vegetation WUE to increase
2) Table 1.a) Add another column to show the time span of the data; b) Add another row to
show the information of PET as for GPP, ET, NDVl and LUCC; c) Add another row to show the information of meteorological data as for PET.
Response/action: Thank you for your scrupulous reminding. I have explained the time span of the data used in the text, the table only shows the basic information of the data source, in addition the meteorological data used is not remotely sensed data, so I did not put it together, but describe it in the next paragraph on a separate line
3) Equation (4).Could you please explain what j, i, Xj, Xi are?
Response/action: Thank you for your scrupulous reminding. I have made the addition
4) Line 197.The spatial difference is subtle rather than obvious.
Response/action: Thank you for your scrupulous reminding. I have changed the obvious to subtle
5) Line 207.Please correct "scrub" in the whole manuscript. lt is "shrub".
Response/action: Thank you for your scrupulous reminding. I have corrected the errors in the whole text
6) l doubt the land cover data used. Table 2 shows the percentage of each land cover type in
LP, and the four land cover types cover more than 93% of the total area of LP, which means the bare ground including the desert accounts for less than 7% in LP. Do you believe the numbers if you have ever been to the LP? lf the numbers are true, it means to me that the LP is even more vegetated compared with Alaska where the area is only 74% vegetated. If you don't have any sense of it, you could look at the LP and Alaska in google maps in the satellite mode. Do you know how the four vegetation (forest, shrub, grassland, and crop)are defined in the land cover product used? Especially grassland which covers ~50% of the total area. It could be that even though there is only one stand of grass in a 30m by 30mpixel, the pixel is classified as grassland which doesn't make sense at least to me. If the grassland accounts for more than 50% for example in the pixel, I have no doubt it is a grassland pixel. l suggest you review the definition of the land cover types in the land cover product.
Response/action: Thank you for your scrupulous reminding. This data is the data used in my reference "The 30 m annual land cover datasets and its dynamics in China from 1990 to 2020", this data makes a land classification data that is recognized to be more accurate than MODIS. Professors Jie Yang and Xin Huang produced the first Landsat-derived annual landcover dataset (CLDC) of China on the Google Earth Engine (GEE) platform using 335,709 Landsat images.) The data includes 9 land types, namely: agricultural land, forest, shrub, grassland, water, snow and ice, bare land, impervious surface, and wetland. I intercepted the vector range of the Loess Plateau and used only forest, scrub, grassland and farmland for the study.
7) Could you provide the spatial distribution of CWSI Figure as those of WUE rates shown in Figure 4?
Response/action: Thank you for your scrupulous reminding. I certainly did the CWSI distribution map for the study, but CWSI was a limiting factor for me, and I didn't think about introducing CWSI distribution in the article
8) Lines 244-245.The order is true only when the CWSI is between 0.1 and 0.3, When the CWSl is larger than 0.3, the order is shrub > woodland > cropland > grassland, right?
Response/action: Thank you for your scrupulous reminding. The conclusion I draw in the article is not a comparison of a point or an interval, but a comparison of the average of all intervals, regardless of the use of an interval will first have some one-sided
9) Line 245.The value of 1.72 of WUE corresponds to which range of CWSl in Figure 5a?
Response/action: Thank you for your scrupulous reminding. 1.72 is obtained by averaging over the entire interval
10)Does Figure 5 suggest that shrub is the one that is more resilient to drought? And may be the best among the all four for the vegetation restoration in LP?
Response/action: Thank you for your scrupulous reminding. The orange color in Figure 5 is representative of the scrub, from the figure can be seen as the yellow line on top of several other colors, especially the party CWSI becomes larger, and the orange line has a certain rise
11) Line 266."northwestern part" or west edge?
Response/action: Thank you for your scrupulous reminding. The text refers to the northwestern part of the Loess Plateau (the study area), not to the northwestern part of China
12)Figure 6. Why is the WUE rate in the area in the south of LP decreasing in summer (Fig.6c)?
We know that the opening and closing of leaf stomata directly influences WUE, and the environmental factors such as atmosphere CO2, leaf humidity, soil moisture can impact the regulation of leaf stomata aperture. A related question is which environmental factor do you think in your study is the dominant one (increase of atmosphere CO2 or decrease of soil moisture or increasing air temperature) that causes the increase of WUE in LP?
Response/action: Thank you for your scrupulous reminding. Because the vegetation in the south is mostly forested and grows more densely, but the groundwater is fixed, when the vegetation is too dense, the WUE will decrease, and another reason is the high summer temperature and uneven precipitation, which leads to an increase in ET after each heavy rain, resulting in a decrease in WUE
13)Line 280."an increasing trend" and also a decreasing trend, too.
Response/action: Thank you for your scrupulous reminding. The upward trend referred to in the text is the average value of the whole region to see whether it is positive or negative, if it is positive, it is up, if it is negative, it is down, because there is a serious spatial heterogeneity in the Loess Plateau, so the analysis of different regions is carried out later.
l4)Lines 282-283."The WUE growth rate of agricultural land 282 decreased with the increase in the drought index" when CWSI is larger than 0.1.
Response/action: Thank you for your scrupulous reminding. I think the 0.1-0.2 interval may not be accurate for CWSI, the decline I am talking about is looking at the overall trend, when I refer to other literature, their pictures are from CWSI of 0.3 as the starting value, I consider the integrity of the information, starting from 0, but the general trend is indeed that as CWSI increases, cropland WUE decreases, before 0.3 should not be considered as doing drought
l5)Lines 286-287.The rate of change of the WUE is the growth rate of WUE? If it is, then the largest growth rate of WUE of farmland is in autumn Figure 7d) instead of Figure 7b).
Response/action: Thank you for your scrupulous reminding. The figures in the text are averaged over the entire region. The maximum spring farmland expressed is the conclusion drawn from a comparison of the four vegetation types in the spring, not from a comparison between seasons.
16)Why are the growth rates of grassland and shrub in Figure 7b) negative when the CWSI
is 0.1-0.2? Any thoughts on this?
Response/action: Thank you for your scrupulous reminding. Because for forests and scrubs, mild drought has a catalytic effect on their growth, so that in areas with sufficient moisture, instead, WUE is suppressed
17) lt is interesting that it seems the author is trying to answer the question of the correlation
between CWSl and WUE. However, there is not a spatial map of CwSl as those of NDVl, Precipitation and Temperature correlations mentioned in section 3.4.
Response/action: Thank you for your scrupulous reminding. In my discussion, I analyze in detail the changes in NDVI, precipitation, and temperature correlation with WUE as CWSI increases and the reasons for them. The focus of this paper is on the relationship between WUE and NDVI, precipitation, and air temperature, and CWSI is not included in the discussion, so I did not put pictures of CWSI.
18)Correct the title of Figure 8. The title of Figure 8 is Distribution of NDVI(b,e,h,k), precipitation (c,f,i,l), temperature, and WUE correlation coefficients for different influences (a,d,g,i)....Subfigures corresponding to NDVl should be the first column (a,d,g,j), precipitation is the second column (b,e,h,k) and temperature is the third column (c,f,I,l),right?
Response/action: Thank you for your scrupulous reminding. I have modified Figure 8 to make it easier for the reader to understand
19)Figure 8. In Figure 8g and Figure 8j, there are areas marked with red color in northwest LP
where the correlation is less than -0.2. Why is the correlation between NDVl and WUEnegative in these areas in summer and autumn?
Response/action: Thank you for your scrupulous reminding. Because the vegetation in the south is mostly forested and grows more densely, but the groundwater is fixed, when the vegetation is too dense, the WUE will decrease, and another reason is the high summer temperature and uneven precipitation, which leads to an increase in ET after each heavy rain, resulting in a decrease in WUE
20)Table 3.The correlation between temperature and WUE in summer is 0.16 listed in Table 3.However, in Figure 8i most of the areas that passed the significance test have negative correlations colored with red and orange which should result in a negative correlatioron average in these areas. How could a positive correlation (0.16) between temperature andwUE come in Table 3?
Response/action: Thank you for your scrupulous reminding. After checking the correlation should be negative, it has been corrected in the text
2I) Lines 358-360.I am not sure whether you're talking about Figure4c.The red color means
low WUE, green and blue colors mean high WUE, right? lsn't the high-value area concentrated in the forest in the south and east of LP?
Response/action: Thank you for your scrupulous reminding. I have corrected the error in lines 359-361.
22)Lines 374-377.Are you talking about Figure5? None of the figures in Figure 5 indicates
farmland is the least sensitive to CWSl compared with other vegetation types. And the grass is the least sensitive to CwSl in spring (Figure 5b).Figure 5b suggests farmland is actually the most vulnerable to CWSl, and Figure 5a and Figure 5b suggest farmland is the second most vulnerable.
Response/action: Thank you for your scrupulous reminding. Whether it is sensitive or not, this paper uses the slope of each line to indicate
23)Line 379.wUE is higher in areas with farmland distribution. Which figure can support this point? In Figure 4, farmland has lower WUE compared with forest.
Response/action: Thank you for your scrupulous reminding. The land use map in Figure 2, combined with the spatial distribution of WUE in Figure 4, reveals the conclusion in line 379 of the text
24) Lines 397-398."the proportion of water 397 directly used for the photosynthesis of
vegetation. "You mean transpiration?
Response/action: Thank you for your scrupulous reminding. What I mean is that part of the water absorbed by vegetation is used for photosynthesis and part for transpiration, and when the water consumed by transpiration is greater than that consumed by photosynthesis, the correlation between WUE and NDVI will decrease.
25)Lines 399-400."so the correlation between ecosystem the WUE and NDVl began to decrease." The result that the correlation of WUE and NDVl decreases to my understanding suggests that with the increase of water stress, photosynthesis might be limited due to factors such as soil water, leaf moisture which affects the opening or closing of leaf stoma while how much or less read visible light is absorbed by the chlorophyll is not that
important.
Response/action: Thank you for your scrupulous reminding. What I mean is that the increase in WUE of the vegetation indicates that the vegetation is absorbing more water, but the absorbed water is used for evapotranspiration rather than the growth of the vegetation itself, so the correlation between WUE and NDVI decreases
26)Line 404."difficulty". What is the difficulty and why does the plant have the difficulty in
spring rather than in summer?
Response/action: Thank you for your scrupulous reminding. The "difficulty" in this article means that water is difficult to use for photosynthesis because it is used for evapotranspiration from the soil. The text does not mention that spring is more difficult than summer
27)Line 406-408."since the temperature affects the stomatal conductance and vegetation photosynthesis mainly through 407 evapotranspiration" This applies to all the four vegetation types. How could it explain the difference of the correlation between WUE and different vegetation types?
Response/action: Thank you for your scrupulous reminding. These variability is reflected from Figure 5, where there are different expressions of different vegetation
28)Lines 411-412."the drought index increases and the tempera-411 true rises this will
accelerate the evapotranspiration of the vegetation" Drought indicates the soil moisture is already very low, and how could the evapotranspiration increase when both the ground and the vegetation are already very dry?
Response/action: Thank you for your scrupulous reminding. It is because the soil is dry, resulting in a large amount of water being evaporated by the soil after the rain
29)Figure 9.The figures in Figure9 to my understanding are meant to investigate whether the
correlations between WUE and the environmental factors change with CWSl or water shortage stress. Take an example here. In Figure 9a when CWSl increases from 0 to 0.5which means the ground is getting drier, WUE varies with NDVl in a more consistent way. Even though the discussion in Section 4.2 tries to explain of why the correlation between WUE and NDVl varies with CWSl, why the correlation between WUE and temperature becomes stronger with increasing of CWSl, and why correlation between WUE and precipitation is negative instead of why the correlation varies with CWSl, the explanation given is not convincing and my impression is that the author is lack of enough ecosystem knowledge related to the WUE. l suggest the author talk with people from the ecology or read more literature on the ecological mechanism of WUE to really understand how environmental factors drive the opening and closing of leaf stomatal which ultimately controls WUE.
Response/action: Thank you for your scrupulous reminding. I will always learn about vegetation growth mechanisms, and the discussion section is an elaboration of what I understand and what I think will be said in detail.
30)Lines 425-427. Is this concluded from Figure5? lf so, farmland is the second instead of the fourth in the rank
Response/action: Thank you for your scrupulous reminding. The size ranking in this paper is based on the overall average, and is not a conclusion on a particular attachment or point.
31)Lines 433-435."the correlation between the NDVI and WUE 433 was the strongest in
summer, and the correlation between the WUE and NDVl for the ecosystem was the strongest in summer. "What is the difference between these two sentences?
Response/action: Thank you for your scrupulous reminding. I have removed the duplicate expressions
32)Lines 439-441."In comparison to forest and farmland regions, the link between the WUE
and temperature 439 in grassland and scrub areas was more sensitive to the values of the drought index, but 440 there was also a threshold effect." Honestly, l have no clue which figure could support this conclusion since it is the first time you mention this threshold effect in the whole paper except the abstract.
Response/action: Thank you for your scrupulous reminding. It can be seen from Figure 7 that the highest correlation is in the interval of CWSI of 0.3-0.4. Therefore, I said that there is a threshold effect, but if you think it is not appropriate, I have deleted the relevant statement
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
Please see the comment in the purple lines in the uploaded doc.
Comments for author File: Comments.pdf
Author Response
Response to reviewer 2
Dear anonymous reviewer,
Thank you very much for your enthusiastic help and valuable comments.
- l) Line 90: Do you have any insight why the drought events can marginally raise WUE? And could you add a sentence to explain the mechanism of it in the same line?
Response/action: Thank you for your scrupulous reminding. Because for scrub and forest, a slight drought for the vegetation can make them work to absorb groundwater and stimulate the vegetation WUE to increase
You can say no if you don’t know the mechanism. The answer you provide here doesn’t explain why the drought increases WUE.
Response/action: Thank you for your scrupulous reminding. I believe that drought stress affects the physiological functions of cereal crops in the following order: cell expansion, stomata movement, transpiration, photosynthesis and transport and distribution of photosynthetic products. Stomata closure is the first response of leaves to water stress, water stress leads to the closure of a large number of stomata in leaves, stomata closure will reduce the transpiration rate, when ET increases, GPP remains unchanged, the response WUE will increase
2) Table 1.a) Add another column to show the time span of the data; b) Add another row to show the information of PET as for GPP, ET, NDVl and LUCC; c) Add another row to show the information of meteorological data as for PET.
Response/action: Thank you for your scrupulous reminding. I have explained the time span of the data used in the text, the table only shows the basic information of the data source, in addition the meteorological data used is not remotely sensed data, so I did not put it together, but describe it in the next paragraph on a separate line
“I have explained the time span of the data used in the text”
Could you point me to the text where the time span of the four data sets in Table is described? I can’t find it in the 2.2 Data Sources. What I suggested is that if you add the time span of the data sets in Table 1 it will be easier for the readers to find the information. The readers are not as smart as you and know the paper as well as you.
Response/action:
You don’t answer my second question b) Where does the PET used to calculate CWSI come from?
Response/action: The formula for CWSI is shown in equation (2) in the text, and both ET and PET are downloaded from PML data, with different bands representing ET and PET “the meteorological data used is not remotely sensed data”
Interesting. I saw the meteorological data from 52 stations. However, Figure 8 shows the spatial instead of multi-site coefficient between WUE and precipitation. To my understanding you need a gridded data set of precipitation to calculate the spatial coefficient.
Response/action: I interpolated the downloaded meteorological data in ArcGIS software to obtain a precipitation image, and then used the correlation coefficient obtained from the correlation calculation of the image after checking
6) l doubt the land cover data used. Table 2 shows the percentage of each land cover type in LP, and the four land cover types cover more than 93% of the total area of LP, which means the bare ground including the desert accounts for less than 7% in LP. Do you believe the numbers if you have ever been to the LP? lf the numbers are true, it means to me that the LP is even more vegetated compared with Alaska where the area is only 74% vegetated. If you don't have any sense of it, you could look at the LP and Alaska in google maps in the satellite mode. Do you know how the four vegetation (forest, shrub, grassland, and crop) are defined in the land cover product used? Especially grassland which covers ~50% of the total area. It could be that even though there is only one stand of grass in a 30m by 30m pixel, the pixel is classified as grassland which doesn't make sense at least to me. If the grassland accounts for more than 50% for example in the pixel, I have no doubt it is a grassland pixel. l suggest you review the definition of the land cover types in the land cover product.
Response/action: Thank you for your scrupulous reminding. This data is the data used in my reference "The 30 m annual land cover datasets and its dynamics in China from 1990 to 2020", this data makes a land classification data that is recognized to be more accurate than MODIS. Professors Jie Yang and Xin Huang produced the first Landsat-derived annual landcover dataset (CLDC) of China on the Google Earth Engine (GEE) platform using 335,709 Landsat images.) The data includes 9 land types, namely: agricultural land, forest, shrub, grassland, water, snow and ice, bare land, impervious surface, and wetland. I intercepted the vector range of the Loess Plateau and used only forest, scrub, grassland and farmland for the study.
Thanks for providing the detailed background. Is there any paper from Yang and Huang that describes their CLDC data product? Have you ever checked whether the data you interpolate based on CLDC agrees with the original CLDC on the total percentage of forest, shrub (instead of scrub), grassland and farmland? Or Yang and Huang’s CLDC product suggests that four land cover types cover ~93% LP?
Response/action: I checked other authors who used this dataset and they all had high ratings for accuracy, and I think my use of this data contributed to the accuracy of the experiment. The following figure shows the land use data I downloaded, values 1-4 represent agricultural land, forest land, scrub, and grassland, respectively. By calculation, agricultural land is 28.83%, forest land is 14.27%, scrub is 0.31%, and grassland is 50.24%, which is indeed calculated to be 93.65%
7) Could you provide the spatial distribution of CWSI Figure as those of WUE rates shown in Figure 4?
Response/action: Thank you for your scrupulous reminding. I certainly did the CWSI distribution map for the study, but CWSI was a limiting factor for me, and I didn't think about introducing CWSI distribution in the article.
Could you provide the CWSI spatial maps in the Supplementary?
Response/action: Thank you for the reminder. The following figure shows the spatial distribution of CWSI
8) Lines 244-245.The order is true only when the CWSI is between 0.1 and 0.3, When the CWSl is larger than 0.3, the order is shrub > woodland > cropland > grassland, right? Response/action: Thank you for your scrupulous reminding. The conclusion I draw in the article is not a comparison of a point or an interval, but a comparison of the average of all intervals, regardless of the use of an interval will first have some one-sided
Could you add what you explained here in the text so that it is easy for the readers like me who are not as smart as you to follow?
Response/action: I have added a description in the text accordingly
9)Line 245.The value of 1.72 of WUE corresponds to which range of CWSl in Figure 5a? Response/action: Thank you for your scrupulous reminding. 1.72 is obtained by averaging over the entire interval
Could you add what you explained here in the text so that it is easy for the readers like me who are not as smart as you to follow?
Response/action: I have added a description in the text accordingly
10) Does Figure 5 suggest that shrub is the one that is more resilient to drought? And may be the best among the all four for the vegetation restoration in LP?
Response/action: Thank you for your scrupulous reminding. The orange color in Figure 5 is representative of the scrub, from the figure can be seen as the yellow line on top of several other colors, especially the party CWSI becomes larger, and the orange line has a certain rise
I saw the lines in Figure 5, and they are very clear to me. The reason why I asked this question is that I am trying to help you to interpret beyond describing the result, i.e., what we can learn from Figure 5.
Response/action: I chose these four plants because they are the most representative plants that occupy the largest area of the Loess Plateau. Figure 5 represents the change of different types of WUE with the increase of CWSI, which is to show the relationship between the degree of water stress and WUE of each vegetation
12)Figure 6. Why is the WUE rate in the area in the south of LP decreasing in summer (Fig.6c)? We know that the opening and closing of leaf stomata directly influences WUE, and the environmental factors such as atmosphere CO2, leaf humidity, soil moisture can impact the regulation of leaf stomata aperture. A related question is which environmental factor do you think in your study is the dominant one (increase of atmosphere CO2 or decrease of soil moisture or increasing air temperature) that causes the increase of WUE in LP?
Response/action: Thank you for your scrupulous reminding. Because the vegetation in the south is mostly forested and grows more densely, but the groundwater is fixed, when the vegetation is too dense, the WUE will decrease, and another reason is the high summer temperature and uneven precipitation, which leads to an increase in ET after each heavy rain, resulting in a decrease in WUE
Oh. Isn’t the red area in the south of LP in Figure 6(c) corresponding to the farmland? I thought you were going to explain in a more convincing way that the irrigation in the crop area in the south in summer results in less WUE and the result suggests an overuse of water in irrigation which could be saved.
Response/action: Thank you for your answer, I think what you said is very reasonable, I should also strengthen the study about the vegetation growth mechanism in the future. Also, I think my explanation is fine, except that I didn't say anything about the reason why the WUE of the farmland became smaller. In my understanding, the southern part of the study area is in the southern part of the farmland there is still a part of forest land present
13)Line 280."an increasing trend" and also a decreasing trend, too.
Response/action: Thank you for your scrupulous reminding. The upward trend referred to in the text is the average value of the whole region to see whether it is positive or negative, if it is positive, it is up, if it is negative, it is down, because there is a serious spatial heterogeneity in the Loess Plateau, so the analysis of different regions is carried out later.
“The upward trend referred to in the text is the average value of the whole region to see whether it is positive or negative”
Could you add a line that represents the average value of the whole region in Figure 7 (a) to support what you assert above? Or could you add what you explained here in the text so that it is easy for the readers like me who are not as smart as you to follow?
Response/action: I have made the corresponding changes in the article, see lines 283-285
l4)Lines 282-283."The WUE growth rate of agricultural land decreased with the increase in the drought index" when CWSI is larger than 0.1.
Response/action: Thank you for your scrupulous reminding. I think the 0.1-0.2 interval may not be accurate for CWSI, the decline I am talking about is looking at the overall trend, when I refer to other literature, their pictures are from CWSI of 0.3 as the starting value, I consider the integrity of the information, starting from 0, but the general trend is indeed that as CWSI increases, cropland WUE decreases, before 0.3 should not be considered as doing drought
“I refer to other literature, their pictures are from CWSI of 0.3 as the starting value, I consider the integrity of the information, starting from 0, but the general trend is indeed that as CWSI increases, cropland WUE decreases, before 0.3 should not be considered as doing drought”
Could you add what you explained here in the text so that it is easy for the readers like me who are not as smart as you to follow?
Response/action: I have made the corresponding changes in the article, see lines288-290.
l5)Lines 286-287.The rate of change of the WUE is the growth rate of WUE? If it is, then the largest growth rate of WUE of farmland is in autumn Figure 7d) instead of Figure 7b). Response/action: Thank you for your scrupulous reminding. The figures in the text are averaged over the entire region. The maximum spring farmland expressed is the conclusion drawn from a comparison of the four vegetation types in the spring, not from a comparison between seasons.
I apologize. My bad. I didn’t notice the unit of WUE on the y-axis in Figure 7 in the revised manuscript.
Response/action: I'll try to make it clearer for the reader in my next writing
16)Why are the growth rates of grassland and shrub in Figure 7b) negative when the CWSI is 0.1-0.2? Any thoughts on this?
Response/action: Thank you for your scrupulous reminding. Because for forests and scrubs, mild drought has a catalytic effect on their growth, so that in areas with sufficient moisture, instead, WUE is suppressed
Is CWSI averaged over several years or just one year?
Response/action: The average here is a 20-year average because I am analyzing with the 20-year WUE average.
17) lt is interesting that it seems the author is trying to answer the question of the correlation between CWSl and WUE. However, there is not a spatial map of CWSl as those of NDVl, Precipitation and Temperature correlations mentioned in section 3.4.
Response/action: Thank you for your scrupulous reminding. In my discussion, I analyze in detail the changes in NDVI, precipitation, and temperature correlation with WUE as CWSI increases and the reasons for them. The focus of this paper is on the relationship between WUE and NDVI, precipitation, and air temperature, and CWSI is not included in the discussion, so I did not put pictures of CWSI.
Could you provide the CWSI spatial maps in the Supplementary?
Response/action: Thank you for the reminder. I will add the spatial distribution map of CWSI in the supplemental file
19)Figure 8. In Figure 8g and Figure 8j, there are areas marked with red color in northwest LP where the correlation is less than -0.2. Why is the correlation between NDVl and WUE negative in these areas in summer and autumn?
Response/action: Thank you for your scrupulous reminding. Because the vegetation in the south is mostly forested and grows more densely, but the groundwater is fixed, when the vegetation is too dense, the WUE will decrease, and another reason is the high summer temperature and uneven precipitation, which leads to an increase in ET after each heavy rain, resulting in a decrease in WUE
I am asking about the red and orange areas in the northwest instead of the south. Are you suggesting that even though there is more rain in summer, high temperature in summer results in more evaporation from the ground since those areas are less vegetated, and the gain of gpp can’t offset the increase of evaporation.
Response/action: The situation you mentioned only applies to the densely vegetated areas in the south of the Loess Plateau, the northwest is basically covered with yellow material, the correlation you mentioned is the red area, the land use type in that part is ice or snow, not the scope studied in this paper, the specific land use classification map is shown in the picture I added in the sixth question
22)Lines 374-377.Are you talking about Figure5? None of the figures in Figure 5 indicates farmland is the least sensitive to CWSl compared with other vegetation types. And the grass is the least sensitive to CwSl in spring (Figure 5b). Figure 5b suggests farmland is actually the most vulnerable to CWSl, and Figure 5a and Figure 5b suggest farmland is the second most vulnerable.
Response/action: Thank you for your scrupulous reminding. Whether it is sensitive or not, this paper uses the slope of each line to indicate
Agree. Let me correct myself here. If we look at Figure 5(b), the most sensitive land cover type is crop. In Figure 5(a) and Figure 5(c), crop is the second most sensitive land cover type. It is not the least sensitive one.
Response/action: This is the change of WUE in spring for different vegetation types with the increase of CWSI, this is my original data, you can see that SLOPE of crop in spring is the largest, that is to say, the most sensitive. Seeing as the grass slope is the smallest and therefore the least sensitive
This is the change of WUE in autumn for different vegetation types with the increase of CWSI, this is my original data, you can see that SLOPE of forest is the largest, that is to say, the most sensitive. Seeing as the grass slope is the smallest and therefore the least sensitive
32)Lines 439-441."In comparison to forest and farmland regions, the link between the WUE and temperature in grassland and scrub areas was more sensitive to the values of the drought index, but there was also a threshold effect." Honestly, l have no clue which figure could support this conclusion since it is the first time you mention this threshold effect in the whole paper except the abstract.
Response/action: Thank you for your scrupulous reminding. It can be seen from Figure 7 that the highest correlation is in the interval of CWSI of 0.3-0.4. Therefore, I said that there is a threshold effect, but if you think it is not appropriate, I have deleted the relevant statement
You can keep it. It is better that you mention the threshold effect and make it clear in the discussion of Figure 7.
Response/action: I have added a description of the threshold effect in lines 295-298 of the text.
Author Response File: Author Response.pdf
