Spatiotemporal Variation in Compound Dry and Hot Events and Its Effects on NDVI in Inner Mongolia, China

Round 1

Reviewer 1 Report

Compound dry and hot events have been extensively investigated both regionally and globally in recent decades. The authors select Inner Mongolia to investigate changes in CDHE and its effects on NDVI based on SDHI and multiple statistical methods. In general, the manuscript is well-written, logically organized and the figures are appropriate. The manuscript can be accepted for publication with minor revisions. I provide some comments that may help improve the manuscript.

1. Line 80-114, there is confusion here, and I suggest that the compound dry and hot events be described first, and then the effect of the compound dry and hot events on the vegetation.

2. Why select Inner Mongolia to investigate changes in CDHE and its effects on NDVI?

3. Line 195 and 198-201, Mann-Kendall test, Mutation analysis, Mann-Kendall mutation detection, which is very confusing for readers.

4. 2.3.5. Relative importance analysis, what is the meaning of the "LMG", "First"….? Are they different estimation methods?

5. Figure 3, Whether there is a mutation of SDHI is not clear in this manuscript.

6. Figure 4 and Figure 8, I recommend add trend value map in the manuscript.

7. SDHI and NDVI trend in different vegetation types is still unclear, please add some statistical results, such as boxplot, histogram, etc. and please highlight the differences in different vegetation types.

8. Figure 10. What’s the mean of the yellow section?

9. why are you select four methods: LMG, First, Genizi, and CAR to assess the relative importance of the three climatic conditions to the interannual variation of NDVI in Inner Mongolia?

10. Line 381, What’s the mean of the sand-blocking "edge-locked" shelterbelts?

11. Authors must indicate clearer the main results obtained in the research.

Author Response

Dear Reviewer,

Thanks for your comments on our manuscript; your comments are of great assistance to us for improving and revising our manuscript so as to be acceptable for publication in ''Remote Sensing''. We had tried to revise and improve the manuscript in line with the suggestions made by you, other reviewer and editor.

We accepted your suggestion and had a careful check for the English writing and the formation throughout the manuscript, meanwhile, the language of our manuscript have been refined and polished by a professional editing company.

The followings are responses to your comments and remarks:

Comments:

Compound dry and hot events have been extensively investigated both regionally and globally in recent decades. The authors select Inner Mongolia to investigate changes in CDHE and its effects on NDVI based on SDHI and multiple statistical methods. In general, the manuscript is well-written, logically organized and the figures are appropriate. The manuscript can be accepted for publication with minor revisions. I provide some comments that may help improve the manuscript.

1. Line 80-114, there is confusion here, and I suggest that the compound dry and hot events be described first, and then the effect of the compound dry and hot events on the vegetation.

Thanks for your suggestion. We have reorganized the introduction section, in new lines 60-128.

2. Why select Inner Mongolia to investigate changes in CDHE and its effects on NDVI?

Thanks for your suggestion. During the past two decades, the strength of land-atmosphere coupling has been profoundly enhanced over a large part of Asia, including southeastern Asia, China, and Mongolia (Figure1, A and B). The most extensive change occurred over Mongolia and northern China. The implication of these findings is that compound extremes heatwaves and heatwave-droughts of summer may occur more frequently and potentially become more severe in inner East Asia [1]. It has been noted that in the global semi-arid zone, the interpretation of vegetation varies greatly between climatic conditions, and local and regional trends reveal large differences in the direction and magnitude of change [2].

Inner Mongolia (China) has a unique geographical location in an arid and semi-arid region with low annual precipitation and special climatic conditions where precipitation is usually concentrated in summer. As one of the most sensitive regions to external environmental changes, studies related to its climatic conditions and vegetation have important implications for North China and even for the whole China. However, studies of climate events and climate on vegetation in Inner Mongolia are still focused on independent events of drought or high temperature, and studies of compound events are mostly focused on large-scale areas with geographical divisions. At present, the analysis of vegetation in this region of Inner Mongolia has not been carried out for the compound dry and hot events on it.

Reference:

[1] Zhang P, Jee-Hoon Jeong, Jin-Ho Yoon, Hyungjun Kim, S.-Y. Simon Wang, Hans W. Linderholm, Keyan Fang, Xiuchen Wu, Deliang Chen., 2020. Abrupt shift to hotter and drier climate over inner East Asia beyond the tipping point. Science. 370(6520):1095-1099.

[2] Wei Z, Xiu, B.Y., Jiao, C.C, Xu C.D., L i u, Y., Wu, G.A. Increased association between climate change and vegetation index variation promotes the coupling of dominant factors and vegetation growth. Science of total environment. 2021, 767.

3. Line 195 and 198-201, Mann-Kendall test, Mutation analysis, Mann-Kendall mutation detection, which is very confusing for readers.

Thanks for your suggestion. Sequential Mann-Kendall is conducted to determine

trend fluctuation and existence of changing point in the data series [1]. Based on your comments, we have modified the confusing statement in this manuscript.

The non-parametric Mann-Kendall test is widely used in detecting trends of variables in meteorology and hydrology fields [2–5]. Statistic S can be obtained by Eq.(1).

where n is the length of the sample, ?_k and ?_j are from ?=_{1, 2, …, n-1} and ?= _{?+1, …, n}. If n is bigger than 8, statistic S approximates to normal distribution. The mean of S is 0 and the variance of S can be acquired as follows:

Then the test statistic Z is denoted by Eq.(4).

If Z>0, it indicates an increasing trend, and vice versa. Given a confidence level α, the sequential data would be supposed to experience statistically significant trend if |Z|>Z(1-α/2), where Z(1-α/2) is the corresponding value of P=α/2 following the standard normal distribution. In this study, 0.05 confidence levels were used.

When the absolute value of Z is larger than 1.96, the input data can be considered as having a significant trend at a 0.05 confidence level.

Mann-Kendall test can also be used to detect the abrupt changes of climate and hydrological data [6–9]. First, building an order serial ?_k:

where ?_ij = 1 when ?_i > ?_j; ?_ij = 0 when ?_i ≤ ?_j. Test statistic can be expressed as:

where ?(?_k) = ?(? − 1/4) ; ???(?_k) = ?(? − 1)(2? + 5)/72 . ??_k is the forward sequence and follows the normal distribution. ??_k can then be denoted by reversing the series of data based on the same equation. The null hypothesis (no abrupt change point) will be rejected if the ??_k values are greater than the confidence interval, and the approximate time of occurrence of the change point can be located according to the intersection between ??_k and ??_k within the confidence interval. If the intersection is outside the confidence interval, we need to employ another method (a moving t-test technique was used for this study) to analyze the stationarity of hydrometeorological data again.

Reference:

[1] Bari, S.H.; Rahman, T.U.; Hoque, M.A.; Hussain, M. Analysis of seasonal and annual rainfall trends in the northern region of Bangladesh. Atmos Res. 2016, 1–28.

[2] Ahn, K.H.; Merwade, V. Quantifying the relative impact of climate and human activities on streamflow. J. Hydrol. 2014, 515, 257-266.

[3] Liang, L.; Li, L.; Liu, Q. Temporal variation of reference evapotranspiration during 1961–2005 in the Taoer River basin of Northeast China. Agr. Forest Meteorol. 2010, 150, 298-306.

[4] Wang, X.; He, K.; Dong, Z. Effects of climate change and human activities on runoff in the Beichuan River Basin in the northeastern Tibetan Plateau, China. Catena 2019, 176, 81-93.

[5] Sen, P.K.J.J.o.t.A.s.a. Estimates of the regression coefficient based on Kendall's tau. J. Am. Stat. Assoc. 1968, 63, 1379-1389.

[6] Li, L.-J.; Zhang, L.; Wang, H.; Wang, J.; Yang, J.-W.; Jiang, D.-J.; Li, J.-Y.; Qin, D.-Y. Assessing the impact of climate variability and human activities on streamflow from the Wuding River basin in China. Hydrol. Process. 2007, 21, 3485-3491.

[7] Tian, F.; Yang, Y.H.; Han, S.M. Using runoff slope-break to determine dominate factors of runoff decline in Hutuo River Basin, North China. Water Sci. Technol. 2009, 60, 2135-2144.

[8] Ye, X.; Zhang, Q.; Liu, J.; Li, X.; Xu, C.Y. Distinguishing the relative impacts of climate change and human activities on variation of streamflow in the Poyang Lake catchment, China. J. Hydrol. 2013, 494, 83-95.

[9] Zhang, X.; Li, P.; Li, D. Spatiotemporal Variations of Precipitation in the Southern Part of the Heihe River Basin (China), 1984–2014. Water 2018, 10, 410.

4. 2.3.5. Relative importance analysis, what is the meaning of the "LMG", "First"….? Are they different estimation methods?

Thank you for your suggestions. In this manuscript, we analyze the relative importance of three climatic conditions, namely SPI, STI and SDHI, on the annual NDVI values in Inner Mongolia based on a multiple regression framework.This method can be implemented using the R package "relaimpo" [1]，the main calculation principle is in the formula (1)-(2) :

Stepwise multiple regression models were used to assess the relative importance of the three climatic conditions on the inter-annual variability of the NDVI in the Inner Mongolia region. The determined regression and predictor are represented by Equation:

The response of object i is modelled as a linear function of regressor values x_i1 . . .  x_ip, with unknown coefficients β₁ . . .  β_p, and e_i represents the unexplained part.

In linear regression, the coefficients β_k, k=0…p, are estimated by minimizing the sum of squared unexplained parts. If we denote the estimated coefficients as  and the fitted response values as , the coefficient of determination R² can be written as

R² measures the proportion of variation in y that is explained by the p regressors in the model.

Where i=1. . . n, y_i represents the response variable (the annual NDVI values in Inner Mongolia). x_i1 . . .  x_ip represents the control variables (the annual intensity of SPI, STI, and SDHI). β₁ . . .  β_p represent coefficients of different control variables, here representing the relative importance of each climate condition with respect to the response variable NDVI. β₀ is a constant variable and e_i represents the unexplained part.

The R package "relaimpo" provides a variety of models to calculate the relative importance. The difference between these models is mainly in the processing of R² in formula (2). For the same independent variable and dependent variable, multiple models are selected for correlation importance analysis, which can reduce the uncertainty of other factors or the interaction between factors. In this manuscript, four models of LMG [2], First [3], Genizi and CAR [4] are selected. These four models have been proven in past studies to be effective in reducing uncertainty due to different factors [5,6].

Reference：

[1] Groemping, U., 2006. Relative Importance for Linear Regression in R: The Package relaimpo. Journal of Statistical Software. 1, 925-933.

[2] Lindeman, R. H. M. P., 1980. Introduction to Bivariate and Multivariate Analysis. Foresman and Company, Glenview.

[3] Genizi, A., 1993. Decomposition of R2 in multiple-regression with correlated regressors. Statistica sinica. 3, 407-420.

[4] Zuber, V., and K. Strimmer., 2011. High-Dimensional Regression and Variable Selection Using CAR Scores. Statistical Applications in Genetics and Molecular Biology. 10, 34.

[5] Yin L., Feng X., and Fu B., 2020. Irrigation water consumption of irrigated cropland and its dominant factor in China from 1982-2015. Advances in Water Resources. 143, 103661.

[6] Guo, E., Wang, Y., Wang, C., Sun, Z., Bao, Y., Mandula, Naren., Buren, Jirigala., Bao, Y., and Li, H., 2021. NDVI Indicates Long-Term Dynamics of Vegetation and Its Driving Forces from Climatic and Anthropogenic Factors in Mongolian Plateau. Remote sensing. 4, 688.

5. Figure 3, Whether there is a mutation of SDHI is not clear in this manuscript.

Thanks for your suggestion. From Figure 3 we can see that the UF curve of SDHI values for the last 39 years did not exceed the confidence line, indicating that no mutation in SDHI occurred during this time period, and based on your comments, we have described the absence of mutation in the manuscript, as detailed in line 269-270.

6. Figure 4 and Figure 8, I recommend add trend value map in the manuscript.

Thanks for your suggestion. We have included maps on SDHI and NDVI trend values in the manuscript. as detailed in Figure 4a and Figure 8a.

Figure 4. Spatial distribution of SDHI trends in Inner Mongolia from 1982–2020. (a) SDHI trend value, (b) Multi-year average SDHI trend levels.

Figure 8. Spatial distribution of NDVI trends in Inner Mongolia from 1982–2020. (a) NDVI trend value, (b) Multi-year average NDVI trend levels.

7. SDHI and NDVI trend in different vegetation types is still unclear, please add some statistical results, such as boxplot, histogram, etc. and please highlight the differences in different vegetation types.

Based on your comments, we have presented the changes in SDHI and NDVI trends for different vegetation types through histograms, as detailed in Figure 4b and Figure 8b.

Figure 4. Spatial distribution of SDHI trends in Inner Mongolia from 1982–2020. (a) SDHI trend value, (b) Multi-year average SDHI trend levels.

Figure 8. Spatial distribution of NDVI trends in Inner Mongolia from 1982–2020. (a) NDVI trend value, (b) Multi-year average NDVI trend levels.

8. Figure 10. What’s the mean of the yellow section?

Thanks for your suggestion. The yellow section in Figure 10 refers to the outliers that appear in the calculation of NDVI for each grassland type in relation to the three climatic conditions.

Outlier is a value that lies in a data series on its extremes, which is either very small or large and thus can affect the overall observation made from the data series. Outliers are also termed as extremes because they lie on the either end of a data series. Outliers are usually treated as abnormal values that can affect the overall observation due to its very high or low extreme values and hence should be discarded from the data series.
    An outlier can also be stated as a value that lies outside the overall pattern of a distribution and thus can affect the overall data series. Outliers is often regarded as the cause of an error in measurement due to presence of extreme values which may underestimate or overestimate a study because it lies at an abnormal distance from other values in a random sample from a population.

As per the basic standards followed by all statisticians a convenient definition of an outlier is a point which falls more than 1.5 times the interquartile range above the third quartile or below the first quartile. When performing least squares fitting to data, it is often best to discard outliers before computing the line of best fit since these points may greatly influence the result.

9. why are you select four methods: LMG, First, Genizi, and CAR to assess the relative importance of the three climatic conditions to the interannual variation of NDVI in Inner Mongolia?

Thanks for your suggestion. In order to further verify the accuracy of the importance ranking results of the three climate conditions on the NDVI changes in Inner Mongolia, this manuscript introduces four models to calculate the relative importance scores and then obtain the final importance ranking to determine the dominant climate factors affecting the NDVI changes in Inner Mongolia among STI, SPI and SDHI.

The main calculation process is as follows:

The response of object i is modelled as a linear function of regressor values x_i1 . . .  x_ip, with unknown coefficients β₁ . . . β_p, and ei represents the unexplained part.

In linear regression, the coefficients β_k, k=0…p, are estimated by minimizing the sum of squared unexplained parts. If we denote the estimated coefficients as  and the fitted response values as , the coefficient of determination R² can be written as

R² measures the proportion of variation in y that is explained by the p regressors in the model.

Where i=1. . . n, y_i represents the response variable (the annual NDVI values in Inner Mongolia). x_i1 . . .  x_ip represents the control variables (the annual intensity of SPI, STI, and SDHI). β₁ . . .  β_p represent coefficients of different control variables, here representing the relative importance of each climate condition with respect to the response variable NDVI. β₀ is a constant variable and e_i represents the unexplained part.

These four methods (Genizi, CAR, LMG, and First) are fundamentally different in terms of conditional and marginal perspectives in order to provide robust analysis outcomes while accounting for correlations among input variables[1,2]. The main reason for the differences in the results of these models is the treatment of R² in Eq. (4).

LMG is the R² contribution averaged over orderings among regressors [3].

First is each variables contribution when included first, which is just the squared covariance between y and the variable [3].

Genizi is the R² decomposition according to Genizi 1993 [4].

CAR decomposes the proportion of the variance explained, and it is an intermediate between the marginal correlation and standardized regression coefficient [5].

This is the reason for the difference in the results they presented in the end. The Genizi measure and the CAR model have the advantage of being fast to compute. Genizi and CAR are considered to be computationally efficient when dealing with a large number of variables, while still being applicable to correlated variables. The difference between the CAR score and the Genizi score is that the Genizi score tends to favor marginal perspectives, while the CAR score can well balance conditional and marginal perspectives [5]. The other two methods (LMG and First) can address the difficulty that the order of the input variables in the model has a strong effect on the relative importance ranking [6].

These four methods consider the calculation speed and calculation accuracy at the same time, and obtain the relative importance ranking of the three climate condituons to NDVI. Considering a variety of calculation methods, although the results are somewhat different, it can be seen that the four models get the ranking results are very consistent (STI>SDHI>SPI).

Reference:

[1] Ghani, I., Ahmad S., 2010. Stepwise Multiple Regression Method to Forecast Fish Landing. Procedia - Social and Behavioral Sciences. 8, 549-554.

[2] Groemping, U., 2006. Relative Importance for Linear Regression in R: The Package relaimpo. Journal of Statistical Software. 1, 925-933.

[3] Lindeman, R. H. M. P., 1980. Introduction to Bivariate and Multivariate Analysis. Foresman and Company, Glenview.

[4] Genizi, A., 1993. Decomposition of R2 in multiple-regression with correlated regressors. Statistica sinica. 3, 407-420.

[5] Zuber, V., and K. Strimmer., 2011. High-Dimensional Regression and Variable Selection Using CAR Scores. Statistical Applications in Genetics and Molecular Biology. 10, 34.

[6] Yin L., Feng X., and Fu B., 2020. Irrigation water consumption of irrigated cropland and its dominant factor in China from 1982-2015. Advances in Water Resources. 143, 103661.

10. Line 381, What’s the mean of the sand-blocking "edge-locked" shelterbelts?

Thanks for your suggestion. It refers to a technical model proposed by the Chinese government for desert management, and the full name of the model is "Dust Storm Blocking Border Protection Forest and Grass Belt". In order not to cause ambiguity, we have modified the full name in the manuscript.

11. Authors must indicate clearer the main results obtained in the research.

Thanks for your suggestion. We have further sorted and summarized the concluding sections of the manuscript, as detailed in 511-537.

I am very appreciating all your suggestions, comments and favorable consideration again.

Sincerely yours,

Enliang Guo

Author Response File: Author Response.docx

Reviewer 2 Report

Based on the temperature and precipitation data of 115 stations in Inner Mongolia, the author constructed a compound dry and hot index and studied the characteristic changes of compound dry and hot events in Inner Mongolia. They considered various approaches to find temporal, spatial, and mutational signatures of compound events. At the same time, the NDVI data was used to analyze the characteristics of vegetation, and on the basis of the above two indicators, the driving mechanism of the compound dry and hot event for vegetation in Inner Mongolia was studied. In general, the manuscript is well organized and is of great significance for studying the characteristics of compound dry and hot events in arid and semi-arid regions and for further research on climate effects. However, I have several concerns which need to be addressed before I can recommend acceptance. I recommend a minor revision.

1.     Why did you choose annual time scale data in the manuscript? Please explain the reason for your choice.

2.     The SDHI has been widely used in previous studies to investigate the spatiotemporal variations of CDHE across the globe (e.g., Hao et al., 2018 in the reference list) and China (e.g., Wu et al., 2020 in the reference list), changes in CDHE in Inner Mongolia must be covered. When the authors select this unique area, unique results are expected rather than similar analysis at a smaller region.

3.     This manuscript uses the Mann-Kendall test and the Mann-Kendall mutation test to perform a spatial significance test and a time series mutation test, respectively. Please explain the differences between the two methods in detail.

4.     Line 279, What is "the compound dry and hot events in Inner Mongolia are mainly mild"? Does mild refer to the level of compound dry and hot events?

5.     Lines 392-393, the analysis of the relationship between STI and NDVI in meadow steppe seems to show a positive correlation?

6.     Line 477, what does "we studied the compound dry and hot events on a long-term scale, the temporal and spatial evolution characteristics of thermal events and NDVI" mean?

7.     Figure 11, what does R²=25.14 mean, please explain.

8.     The conclusion should be largely shorted.

9.     For all figures with latitudinal and longitudinal labels, the degree format would be enough, minute and second can be removed if they are both zeros.

Author Response

Dear Reviewer

Thanks for your valuable comments on our manuscript; your comments are of great assistance to us for improving and revising our manuscript so as to be acceptable for publication in ''Remote Sensing''. We had tried to revise and improve the manuscript in line with the suggestions made by you. We accepted your valuable suggestion and had a careful check for the English writing and the formation throughout the manuscript, meanwhile, the language of our manuscript have been refined and polished by a professional editing company.

The followings are responses to your comments and remarks:

Comments:

Based on the temperature and precipitation data of 115 stations in Inner Mongolia, the author constructed a compound dry and hot index and studied the characteristic changes of compound dry and hot events in Inner Mongolia. They considered various approaches to find temporal, spatial, and mutational signatures of compound events. At the same time, the NDVI data was used to analyze the characteristics of vegetation, and on the basis of the above two indicators, the driving mechanism of the compound dry and hot event for vegetation in Inner Mongolia was studied. In general, the manuscript is well organized and is of great significance for studying the characteristics of compound dry and hot events in arid and semi-arid regions and for further research on climate effects. However, I have several concerns which need to be addressed before I can recommend acceptance. I recommend a minor revision.

1. Why did you choose annual time scale data in the manuscript? Please explain the reason for your choice.

Thanks for your kind suggestion. we selected the monthly and yearly data in Inner Mongolia for the past 40 years for time series analysis, and found that the yearly data (0.0219/a) has a more obvious change trend than the monthly data (0.0014/a), and it is easier to find the change characteristics of the data. (Figure.1)

Figure 1. The annual data (a) and monthly data (b)in Inner Mongolia for the past 39 years for time series analysis.

2. The SDHI has been widely used in previous studies to investigate the spatiotemporal variations of CDHE across the globe (e.g., Hao et al., 2018 in the reference list) and China (e.g., Wu et al., 2020 in the reference list), changes in CDHE in Inner Mongolia must be covered. When the authors select this unique area, unique results are expected rather than similar analysis at a smaller region.

Thanks for your suggestions. Although SDHI has conducted research on compound dry and hot events in large-scale regions such as the world and China, due to its unique geographical location in arid and semi-arid regions, the Inner Mongolia (China) has less annual precipitation and special climatic conditions that precipitation usually concentrated in summer. One of the regions where are most sensitive to changes in the external environment, its climatic conditions have an important impact on North China and even the whole China.

During the past two decades, the strength of land-atmosphere coupling has been profoundly enhanced over a large part of Asia, including southeastern Asia, China, and Mongolia (Figure 2, A and B). The most extensive change occurred over Mongolia and northern China. The implication of these findings is that compound extremes heatwaves and heatwave-droughts of summer may occur more frequently and potentially become more severe in inner East Asia [1]. However, the current research on climatic events in Inner Mongolia still focuses on independent events of drought or high temperature, and compound events are mostly studied in large-scale regions that focus on the unit of geographical divisions. At present, the analysis of compound dry and hot events in Inner Mongolia has not been carried out, which is a region with unique climatic characteristics.

Hao et al. (2018) used SDHI to conduct research on global compound dry and hot events, he only assessed the changes in the overall severity of compound dry and hot events on a global scale, and made a brief analysis in space.

In the paper of Wu et al. (2020), the standardised dry and hot index (SDHI) was used to evaluate the compound dry and hot events variation of the warm season from 1961 to 2012 based on monthly precipitation and monthly temperature in China, and to explore its impact on agricultural drought.

In contrast to the raster data used in previous studies, this manuscript uses actual meteorological station data to construct the SDHI for compound dry and hot indicators. On this basis, in addition to the spatial and temporal characterization of compound dry and hot events in Inner Mongolia as a whole, this manuscript also investigates the temporal and frequency characteristics of compound dry and hot events in different grassland types.

Figure 2. land-atmosphere coupling strength over the period 1979–1998 (A) and 2000–2017 (B)

Reference

[1] Zhang P, Jee-Hoon Jeong, Jin-Ho Yoon, Hyungjun Kim, S.-Y. Simon Wang, Hans W. Linderholm, Keyan Fang, Xiuchen Wu, Deliang Chen., 2020. Abrupt shift to hotter and drier climate over inner East Asia beyond the tipping point. Science. 370(6520):1095-1099.

3. This manuscript uses the Mann-Kendall test and the Mann-Kendall mutation test to perform a spatial significance test and a time series mutation test, respectively. Please explain the differences between the two methods in detail.

Thanks for your suggestions. Sequential Mann-Kendall is conducted to determine trend fluctuation and existence of changing point in the data series [1].

 The non-parametric Mann-Kendall test is widely used in detecting trends of variables in meteorology and hydrology fields [2–4]. Statistic S can be obtained by Eq.(1).

where n is the length of the sample, ?_k and ?_j are from ?=_{1, 2, …, n-1} and ?= _{?+1, …, n}. If n is bigger than 8, statistic S approximates to normal distribution. The mean of S is 0 and the variance of S can be acquired as follows:

Then the test statistic Z is denoted by Eq.(4).

If Z>0, it indicates an increasing trend, and vice versa. Given a confidence level α, the sequential data would be supposed to experience statistically significant trend if |Z|>Z(1-α/2), where Z(1-α/2) is the corresponding value of P=α/2 following the standard normal distribution. In this study, 0.05 confidence levels were used.

When the absolute value of Z is larger than 1.96, the input data can be considered as having a significant trend at a 0.05 confidence level.

Mann-Kendall test can also be used to detect the abrupt changes of climate and hydrological data [6–9]. First, building an order serial ?_k:

where ?_ij = 1 when ?_i > ?_j; ?_ij = 0 when ?_i ≤ ?_j. Test statistic can be expressed as:

where ?(?_k) = ?(? − 1/4) ; ???(?_k) = ?(? − 1)(2? + 5)/72 . ??_k is the forward sequence and follows the normal distribution. ??_k can then be denoted by reversing the series of data based on the same equation. The null hypothesis (no abrupt change point) will be rejected if the ??_k values are greater than the confidence interval, and the approximate time of occurrence of the change point can be located according to the intersection between ??_k and ??_k within the confidence interval. If the intersection is outside the confidence interval, we need to employ another method (a moving t-test technique was used for this study) to analyze the stationarity of hydrometeorological data again.

Reference:

[1] Bari, S.H.; Rahman, T.U.; Hoque, M.A.; Hussain, M. Analysis of seasonal and annual rainfall trends in the northern region of Bangladesh. Atmos Res. 2016, 1–28.

[2] Ahn, K.H.; Merwade, V. Quantifying the relative impact of climate and human activities on streamflow. J. Hydrol. 2014, 515, 257-266.

[3] Liang, L.; Li, L.; Liu, Q. Temporal variation of reference evapotranspiration during 1961–2005 in the Taoer River basin of Northeast China. Agr. Forest Meteorol. 2010, 150, 298-306.

[4] Wang, X.; He, K.; Dong, Z. Effects of climate change and human activities on runoff in the Beichuan River Basin in the northeastern Tibetan Plateau, China. Catena 2019, 176, 81-93.

[5] Sen, P.K.J.J.o.t.A.s.a. Estimates of the regression coefficient based on Kendall's tau. J. Am. Stat. Assoc. 1968, 63, 1379-1389.

[6] Li, L.-J.; Zhang, L.; Wang, H.; Wang, J.; Yang, J.-W.; Jiang, D.-J.; Li, J.-Y.; Qin, D.-Y. Assessing the impact of climate variability and human activities on streamflow from the Wuding River basin in China. Hydrol. Process. 2007, 21, 3485-3491.

[7] Tian, F.; Yang, Y.H.; Han, S.M. Using runoff slope-break to determine dominate factors of runoff decline in Hutuo River Basin, North China. Water Sci. Technol. 2009, 60, 2135-2144.

[8] Ye, X.; Zhang, Q.; Liu, J.; Li, X.; Xu, C.Y. Distinguishing the relative impacts of climate change and human activities on variation of streamflow in the Poyang Lake catchment, China. J. Hydrol. 2013, 494, 83-95.

[9] Zhang, X.; Li, P.; Li, D. Spatiotemporal Variations of Precipitation in the Southern Part of the Heihe River Basin (China), 1984–2014. Water 2018, 10, 410.

4. Line 279, What is "the compound dry and hot events in Inner Mongolia are mainly mild"? Does mild refer to the level of compound dry and hot events?

Thanks for your suggestions. Due to our inadvertent failure to align the description of SDHI levels in the manuscript with the level statements shown in Figure 5, we have changed the ambiguous sentence to "Overall, the results indicate that Inner Mongolia is dominated by the occurrence of abnormal compound dry and hot events…", see lines 295 for details.

5. Lines 392-393, the analysis of the relationship between STI and NDVI in meadow steppe seems to show a positive correlation?

Thanks for your suggestions. We rechecked the result in Figure 10 and redescribed this result in the manuscript as "Deserts and desert grasslands show a negative correlation, indicating that these two areas are more sensitive to temperature conditions compared to woodland", see lines 415-417 for details.

6. Line 477, what does "we studied the compound dry and hot events on a long-term scale, the temporal and spatial evolution characteristics of thermal events and NDVI" mean?

Thanks for your suggestions. We have changed the sentence that caused the ambiguity to "we study the spatial and temporal evolution characteristics of compound dry and hot events and NDVI on long-term scales from the climate-vegetation system with the help of multi-source data theory and methods, and explore the mechanisms of compound events affecting vegetation changes. "

7. Figure 11, what does R²=25.14 mean, please explain.

Thanks for your suggestions. A stepwise multiple regression model was used to assess the relative importance of STI, SPI and SDHI for the analysis of annual variation of NDVI in Inner Mongolia.

R² means the proportion of the total variation in the n observed values of the dependent variable that is explained by the overall regression model (Bowerman et al., 2005).

In this study R² = 25.14 which means in the multiple linear regression, the ensemble of climate events of SPI, STI and SDHI constituted the drivers of variability accounting for 25.14% of the NDVI variance in Inner Mongolia.

8. The conclusion should be largely shorted.

We have taken your advice and made appropriate deletions to the conclusion.

9. For all figures with latitudinal and longitudinal labels, the degree format would be enough, minute and second can be removed if they are both zeros.

Thanks for your suggestion, we have deleted the minutes and seconds when the latitude and longitude grid of each picture is zero, using a format that only retains degrees.

I am very appreciating all your suggestions, comments and favorable consideration again.

Sincerely yours,

Enliang Guo

Author Response File: Author Response.docx

Reviewer 3 Report

The comments are provided

Comments for author File: Comments.pdf

Author Response

Dear Reviewer:

Thank you very much for your valuable suggestions and comments on my manuscript. Your suggestions and comments are of great assistance to me for improving and revising our manuscript so as to be acceptable for publication in ''Remote Sensing'' .

I have tried to revise and improve the manuscript in line with the suggestions made by you, another reviewer and editor.

The followings are responses to your comments and remarks:

Comments:

1. Authors have not discussed about the selection of NDVI index over other indices such as SAVI, EVI etc. It would be better to provide the advantages and disadvantages of using this index. I have a big concern of using NDVI has several drawbacks such as topographic illumination, shading effect, solar angle issues which has not been discussed and elaborated. These aspects of NDVI has been mentioned and explained clearly in one of the recent studies by Kumari et al., 2021 and Shahzaman et al., 2021 from which authors can benefit.

Kumari, N., Srivastava, A., & Dumka, U. C. (2021). A Long-Term Spatiotemporal Analysis of Vegetation Greenness over the Himalayan Region Using Google Earth Engine. Climate, 9(7), 109. https://0-doi-org.brum.beds.ac.uk/10.3390/cli9070109

Shahzaman, M., Zhu, W., Bilal, M., Habtemicheal, B. A., Mustafa, F., Arshad, M., ... & Iqbal, R. (2021). Remote Sensing Indices for Spatial Monitoring of Agricultural Drought in South Asian Countries. Remote Sensing, 13(11), 2059. https://0-doi-org.brum.beds.ac.uk/10.3390/rs13112059

Thank you for your suggestions. We have benefited from your comments in your two papers on NDVI, and we point out the strengths and weaknesses of NDVI in the introduction and discussion sections, respectively, as follows:

Introduction:

Vegetation plays an important role in the energy exchange of different layers of the earth and is an important part of the ecological environment of the earth system and an important member of the biological carbon cycle [1-3]. Vegetation change is the result of the combined action of the interior and exterior of the Earth. The vegetation index covers a wealth of surface vegetation information, which is of great significance for the study of hydrology, ecology, and regional changes [4, 5]. NDVI is the most often used and is an operational, global-based vegetation index, partly due to its “ratio” properties, which enable the NDVI to cancel out a large proportion of the noise caused by changing sun angles, topography, clouds or shadow, and atmospheric conditions [6, 7]. The NDVI has been widely used to study vegetation growth in recent years [8-10]. The extraction of vegetation information using NDVI is commonly used to indicate the quantitative characteristics of vegetation and for monitoring seasonal changes in vegetation and land cover studies [11, 12].

Discussion:

“Among existing VIs, the Normalized Difference Vegetation Index (NDVI) is the most often used and is an operational, global-based vegetation index, NDVI is a good proxy for vegetation density parameters such as leaf area index (LAI), vegetation cover (FVC), and absorbed photosynthetically active radiation (fAPAR) [13]. However, NDVI has two major limitations in characterizing biomass and productivity. First, the relationship between NDVI and green biomass is non-linear and can be saturated in areas with high vegetation cover. The second limitation is that NDVI mainly reflects vegetation greenness rather than photosynthesis itself. However, total primary productivity (GPP) can decline without any reduction in LAI or chlorophyll. Combining the shortcomings of NDVI, a new vegetation index, kNDVI, was proposed by applying the theory of nuclear methods to NDVI, using machine learning. The kNDVI correlated with GPP similarly or better than other indices globally. kNDVI correlated with SIF better than other indices in general and in all biomes, especially in deciduous broadleaf forests and for herbaceous and cultivated crops. The correlations of kNDVI were higher in nearly all cases (e.g., Spearman correlation, distance correlation), thus confirming the advantage of kNDVI over other indices. Using the kNDVI index in future studies on vegetation will increase the significance of geo-monitoring and terrestrial biosphere studies [14].

Referrence:

[1] Parmesan, C.; Gary, Y.H. A globally coherent fingerprint of climate change impacts across natural systems. Nature. 2003, 421, 37-42.

[2] Forkel, M.; Carvalhais, N.; Verbesselt, J.; Mahecha, M.; Neigh, C.; Reichstein, M. Trend Change Detection in NDVI Time Series: Effects of Inter-Annual Variability and Methodology. Remote Sens. 2013, 5, 2113–2144.

[3] Pan, N.Q.; Feng, X.M.; Fu, B.J.; Wang, S.A.; Ji, F.; Pan, S.F. Increasing global vegetation browning hidden in over-all vegetation greening: Insights from time-varying trends. Remote Sens. Environ. 2018, 214, 59–72.

[4] Kennedy, R.E.; Yang, Z.; Cohen, W.B.; Detecting trends in forest disturbance and recovery using yearly Landsat time series: 1. LandTrendr—Temporal segmentation algorithms. Remote Sensing of Environment. 2010, 114, 2897-2910.

[5] Baret, F.; Guyot, G. Potentials and limits of vegetation indices for LAI and APAR assessment. Remote Sens. Environ. 1991, 35, 161–173.

[6] Kumari, N.; Srivastava, A.; Dumka, U.C. A Long-Term Spatiotemporal Analysis of Vegetation Greenness over the Himalayan Region Using Google Earth Engine. Climate. 2021, 9, 109.

[7] Shahzaman, M.; Zhu, W.; Bilal, M.; Habtemicheal, B.A.; Mustafa, F.; Arshad, M.; Ullah, I.; Ishfaq, S.; Iqbal, R. Remote Sensing Indices for Spatial Monitoring of Agricultural Drought in South Asian Countries. Remote Sens. 2021, 13, 2059.

[8] Zhou, X.; Yamaguchi, Y.; Arjasakusuma, S. Distinguishing the vegetation dynamics induced by anthro-pogenic factors using vegetation optical depth and AVHRR NDVI: A cross-border study on the Mongolian Plateau. Sci. Total Environ. 2018, 616,730–743.

[9] Rivas-Tabares, D.A.; Saa-Requejo, A.; Martín-Sotoca, J.J.; Tarquis, A.M. Multiscaling NDVI Series Analysis of Rainfed Cereal in Central Spain. Remote. Sens. 2021, 13, 568.

[10] Fensholt, R.; Proud, S.R. Evaluation of Earth Observation based global long term vegetation trends — Comparing GIMMS and MODIS global NDVI time series. Remote Sens. Environ. 2012, 119, 131-47.

[11] Gottfried, M.; Pauli, H.; Futschik, A.; Akhalkatsi, M.; Baran, P.; Alonso, J.L.B.; Coldea, G.; Dick, J.; Erschbamer, B.; Calzado, M.R.F.; George, K.; Larsson, P.; Mallaun, M.; Michelsen, O.; Moiseev, D.; Moiseev, P.; Molau, U.; Merzouki, A.; Nagy, L.; Nakhutsrishvili, G.; Pedersen, B.; Pelino, G.; Puscas, M.; Rossi, G.; Stanisci, A.; Theurillat, J.P.; Tomaselli, M.; Villar, L.; Vittoz, P.; Vogiatzakis, I.; Grabherr, G. Continent-wide response of mountain vegetation to climate change. Nature Climate Change. 2012, 2, 111-115.

[12] He, D.; Huang, X.L.; Tian, Q.J.; Zhang, Z.C. Changes in Vegetation Growth Dynamics and Relations with Climate in Inner Mongolia under More Strict Multiple Pre-Processing (2000–2018). Sustainability. 2020, 12, 2534.

[13] Matsushita, B., Yang, W., Chen, J., Onda, Y., Qiu, G. Sensitivity of the enhanced vegetation index (EVI) and normalized difference vegetation index (NDVI) to topographic effects: A case study in high-density cypress forest. Sensors 2007, 7, 2636–2651.

[14] Camps-Valls, G.; Campos-Taberner, M.; Moreno-Martínez, L.; Walther, S.; Duveiller, G.; Cescatti, A.; Mahecha, M.D.; Muoz-Marí, J.; García-Haro, F.J.; Guanter, L. A unified vegetation index for quantifying the terrestrial biosphere. Sci. Adv. 2021, 7, eabc7447.

2. Please provide the NDVI vs elevation, authors can utilise the Figure 1.

Thanks for your suggestion. We have included the comparison of elevation and NDVI in Inner Mongolia in Figure 1.

Figure 1. The distribution of elevation, NDVI, vegetation types, and meteorological stations in Inner Mongolia.

3. Please cite the original citation of NDVI

Thank you for your suggestion. We have changed the data references for NDVI to the original citations and acknowledgements recommended in the official website. (Pedelty, J., Devadiga, S., Masuoka, E., Brown, M., Pinzon, J., Tucker, C., et al. (2007). Generating a long-term land data record from the AVHRR and MODIS instruments. Geoscience and Remote Sensing Symposium, 2007. IGARSS 2007 (pp. 1021–1025). :IEEE International.)

“These data products are processed and distributed by the MEaSUREs Long-Term Data Record project by the Level-1 and Atmosphere Archive & Distribution System (LAADS) Distributed Active Archive Center (DAAC) at the Goddard Space Flight Center.”

4. Provide a tabular format of the hydrometeorogical and other datasets (e.g., satellite) with their source and time period.

Thanks for your suggestion. We have included in the text a table containing specific details of the climate data and satellite data, in new lines 172-173.

5. Although AVHRR can meet the demand of this study, but if we want to obtain more accurate results, the omission error will become one of the fatal defects. The ideal way to solve this problem is to combined use of remote sensing images with different time and spatial frequencies, such as Sentinel images with high spatial-temporal resolution launched in recent years with high time resolution. The combine use of MODIS, Sentinel, Landsat and SPOT sensors can provide remote sensing database with high temporal and spatial resolution.

Thanks for your suggestion. Previous studies on NDVI changes in Inner Mongolia using Landsat MODIS and Sentinel data have confirmed the applicability of the above datasets, but the applicability of the daily AVHRRLTDR NDVI data in Inner Mongolia is not yet clear. The goal of NASA’s Land Long Term Data Record (LTDR) project is to produce a consistent long term data set from the AVHRR and MODIS instruments for land climate studies. The project will create daily surface reflectance and normalized difference vegetation index (NDVI) products at a resolution of 0.05°, which is identical to the Climate Modeling Grid (CMG) used for MODIS products from EOS Terra and Aqua. They more easily capture the sensitivity of the algorithm to surface (e.g., vegetation phenology), atmospheric (e.g., aerosol loading), and remote sensing (e.g., solar-surface-sensor geometry) conditions. As a typical arid and semi-arid region, Inner Mongolia is also one of the most sensitive regions for climate change response. We consider that the AVHRR LTDR dataset, which is sensitive to capture climate and vegetation changes, can well meet the needs of this special climate condition in Inner Mongolia, so we conducted its applicability analysis in this manuscript in Inner Mongolia. The results show that the AVHRR data have the ability to solve the research questions raised in our introduction and satisfactory results were obtained. In the next step, we will combine multi-source remote sensing data, as you mentioned, and integrate remote sensing images with different temporal and spatial frequencies to further conduct a comprehensive analysis of the vegetation in Inner Mongolia.

6. Please add a summary of the results, comparing them with the initial motivation of the study. Have you answered all the research questions described in the Introduction? If yes, how? If not, what are the next steps needed for answering them?

Thanks for your suggestion. Combining your comments and those of other reviewers, we have further sorted out and revised the research questions raised in the introduction, and have re-summarized them as follows：(1) Constructing compound dry and hot index in Inner Mongolia from meteorological station data. (2) Based on the NDVI and the compound dry and hot index constructed in Objective 1, the spatial and temporal variation characteristics of compound dry and hot events and vegetation dynamics in Inner Mongolia are analyzed by utilizing the Theil-Sen slope and Mann-Kendall test. (3) The driving effects of three climatic conditions (drought, high temperature and compound dry and hot condition) on the overall and different grassland types of vegetation in Inner Mongolia, and the dominant climatic condition factors affecting NDVI changes in different grassland types were quantitatively revealed with the help of partial correlation analysis and in a multiple stepwise regression framework. Through our efforts, we have solved the above three research problems. For research objective 1, we have constructed a compound dry and hot index for Inner Mongolia region using meteorological station data of precipitation and temperature. For research objective 2, the period 1982-2020 was dominated by the occurrence of abnormal compound dry and hot events in Inner Mongolia, and the SDHI in general and in each grassland type showed a decreasing trend in time and space without abrupt changes; in the temporal analysis, the NDVI in Inner Mongolia as a whole changed abruptly in 2001, and in each grassland type, except for the desert steppe and desert areas, showed an increasing trend, and more than 70% of the areas in space showed an increasing trend. For research objective 3, temperature conditions had a greater influence on the overall NDVI changes in Inner Mongolia, and SDHI and NDVI were positively correlated in more than 80% of Inner Mongolia; among the vegetations types, forests were most significantly affected positively by STI, and desert areas were more sensitive to temperature conditions than woodlands; five vegetation types were positively correlated with SDHI, and the area most affected by compound dry and hot conditions was desert steppe; the response of different vegetation types to SPI varied widely.

7. Also, it would be great if authors can comment that how the data was processed as in the methodology section it is not clear. For instance, the current satellite product has many issues related to BRDF corrections so did authors perform any kind of such corrections. If not then please argue that.

Thanks for your suggestion. The goal of NASA’s Land Long Term Data Record (LTDR) project is to produce a consistent long term data set from the AVHRR and MODIS instruments for land climate studies. The project will create daily surface reflectance and normalized difference vegetation index (NDVI) products at a resolution of 0.05°, which is identical to the Climate Modeling Grid (CMG) used for MODIS products from EOS Terra and Aqua. The LTDR project will reprocess Global Area Coverage (GAC) data from AVHRR sensors onboard NOAA satellites by applying the preprocessing improvements identified in the AVHRR Pathfinder II project and atmospheric and BRDF corrections used in MODIS processing. The site BRDF parameterization is used to predict the surface reflectance for any set of solar illumination and viewing geometries. The spectral translation equations are used to adjust the predicted reflectance to AVHRR spectral bands. The ratio of the observed surface reflectance to the modeled reflectance is then used to predict the sensor degradation and offer a calibration adjustment coefficient. After the above series of processing, the surface reflectance and normalized vegetation index (NDVI) products from NASA's Land Long Term Data Record (LTDR) project were generated to create. After we downloaded the daily NDVI products, we obtained the monthly and annual NDVI data by the maximum synthesis method in order to solve the problems such as omission errors in AVHRR data, erasing the details of errors on the time scale.

8. Some of the statements are not supported by published works. Authors may like to find studies in line with their statements to add scientific weight to their observations. I believe that after duly addressing the comments authors can improve the quality of the manuscript substantially to make it more insightful. For instance, discussion is very weakly supported by the previous studies, at line 432-435 authors just cite one study already cited in the previous line. I highly recommend the study conducted by Kumari et al., 2021 in the Himalayas to support these lines.

Thanks for your suggestion. Based on your suggestion we have added literature in the Discussion and Analysis section of the manuscript that is consistent with our conclusions. The study you recommended by Kumari et al., 2021 in the Himalayas, effectively supports our results and we have benefited a lot from it, thank you again for your valuable suggestions.

We have added a consistent conclusion on the response of different vegetation types to climate in Inner Mongolia in page 14, line 431

In page 15, line 453, we have added Kumari et al.'s study, which you recommended and which has the same conclusion as our study

We have added consistent findings on the climatic response of vegetation in central Inner Mongolia in page 15, line 457

Added useful conclusions on the driving effects of high temperature and drought conditions on different vegetation types in lines 466 and 469

Lines 472-488 add solid literature on the shortcomings of NDVI and what can be done to address this issue in the future.

I will express my deep gratitude and respect to you for your time and effort in reviewing the manuscript. From the review process can be seen that you are a very rigorous scientific researcher. You have given me detailed corrections and suggestions for every point in my paper, which has benefited me a lot.

I hope that these revisions and the improved text will be satisfactory and make the paper be acceptable for publication in ''Remote Sensing''.

I would like to express my heartfelt thanks and gratitude to you once again.

Sincerely yours,

Enliang Guo

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

The authors have put diligent efforts in responding to reviewers' comments and they are revised the paper accordingly. The paper is ready. I am happy to recommend acceptance of the paper.
Best regards.

Reviewer 3 Report

Overall I recommend for accepting the manuscript as authors have clearly incorporated all the suggestions adequately and in a well structured manner.