the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 09 Sep 2022
Development of a novel daily-scale compound dry and hot index and its application across non-arid regions of China
Abstract. Drought and heat extremes often occur simultaneously or sequentially within a short period, named compound dry and hot events (CDHEs), enhancing damages caused by individual drought or heat extremes. Under global warming, occurrences of short-term CDHEs have increased, adversely impacting the ecosystem and society. However, current indicators generally monitor CDHEs at monthly scales, which cannot reflect short-term CDHEs. This study proposes a novel daily-scale compound dry and hot index (DCDHI) by jointing daily Standardized Moisture Anomaly Index (SZI) and Standardized Temperature Index (STI) using Copula. The applicability of daily SZI and DCDHI indices in monitoring droughts and CDHEs is verified across non-arid regions of China. The daily SZI agrees better with soil moisture variations than the Standardized Precipitation Index (SPI) and Standardized Precipitation and Temperature Index (SPEI) at multiple time scales, indicating it can be applied to construct daily DCDHI for detecting compound dry and hot events. The DCDHI can detect spatial evolutions of dry and hot conditions within a month and reflects vegetation losses, indicating the DCDHI is a good indicator for detecting compound CDHEs at different time scales (daily to monthly). The characteristics of CDHEs during growing seasons (April to October) are also investigated from 1961 to 2021. There is a significant increase in the area affected by CDHEs, which occur more frequently for the period of 1990–2021 than 1961–1989. The severity of compound dry and hot events decreases from the period 1961–1989 to 1990–2021 in northern regions but increases in southern especially southwestern regions. More extreme compound dry and hot events are more likely to occur under global warming. The new tool proposed in this study could detect evolutions and characteristics of short-term CDHEs and provide technical support for the risk management of extreme events.
- Preprint
(2748 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on hess-2022-217', Anonymous Referee #1, 05 Oct 2022
Compounding dry and hot events are challenging the world under a changing climate. This study attempted to expanded the time scale of the Standardized Moisture Anomaly Index (SZI) from months to days for evaluating short-term droughts. Then, it was combined with the hot index to construct a daily scale compound dry and hot (DCDHI) index using Copula theory for investigating important characteristics of compound dry and hot events. This topic is interesting and within the scope of HESS. However, the novelty is not well clarified. This paper is very similar with Li et al. (2021), including methods, structures and figures (e.g, Figure 2). Therefore, I would recommend a Rejection at the present stage. If the authors can well address the following concerns, I also welcome to review the resubmitted version.
Li, J., Wang, Z., Wu, X., Zscheischler, J., Guo, S., and Chen, X.: A standardized index for assessing sub-monthly compound dry and hot conditions with application in China, Hydrology and Earth System Sciences, 25(3), 1587–1601, DOI: 10.5194/hess-25-1587-2021, 2021.
Major concerns:
- There are so many indices for characterizing compound hot and dry events, such as Hao et al., 2019 and Wu et al. (2021). The authors claimed that these previous indices cannot monitor short-term events. Absolutely, monitoring the compounding events at a daily scale is important. Numerous studies have explored the extreme heat events by using daily indices, such as wet-bulb temperature and lethal heat stress index. The most challenge is to characterize the drought in a daily scale and then transform to the compounding events. This study used the existing drought indices by only changing the input from monthly to a daily scale. I am not persuasive that this is a great novelty in hydrological community. The authors should fully clarify this issue.
Hao, Z., Hao, F., Singh, V. P., and Zhang, X.: Changes in the severity of compound drought and hot extremes over global land areas, Environmental Research Letters, 13, 124022, DOI: 10.1088/1748-9326/aaee96, 2018.
Wu, X., Hao, Z., Zhang, X., Li, C., and Hao, F.: Evaluation of severity changes of compound dry and hot events in China based on a multivariate multi-index approach, Journal of Hydrology, 583: 124580, DOI: 10.1016/j.jhydrol.2020.124580, 2020.
- The authors claimed that they calculated the Standardized Temperature Index (STI) following the SPI estimation procedures. They fit the daily temperature data by using a normal distribution function; the normal assumption is OK. As the precipitation (temperature) in different months have large variations, the SPI (and STI) should be fitted in each month. Unfortunately, the authors fail to consider such seasonal variations in the present work.
- To validate that the developed compounding index is better than others, the authors used the NDVI data. There might raise two concerns here. First, the authors use the root-zone soil moisture data to represent drought, which is surely more straightforward to link with vegetation growth. So, it is not surprising that the soil-based index can better represent the drought condition than SPI (and SPEI), which is in the type of meteorological drought. Second, it is not persuasive that they use the monthly NDVI data to validate their daily compounding index.
- The main methods in this study (e.g., both using WSD, root-zone soil moisture, STI, copula, run theory, and the same study region) is very similar with Li et al. (2021). The main differences may be just the input data, i.e., from sub-monthly to daily data. These issue should be fully clarified.
Minor concerns:
- Abstract: Some quantitative results should be presented to represent the increasing severity of compound events.
- L50: “con not” to “can not”.
- L61: the “SDHEs” is not defined.
- L63: the “SCDHI” is not defined.
- The ongoing topic of flash droughts is relevant and should be discussed.
- L86: How do you define the “non-arid” region?
- L104: “The root zone soil moisture data were used for the appropriateness in characterizing drought because they have lower noise than land surface soil water.” The reason about why using root zone soil moisture is not correctly explained. I think you want to focus on vegetation, and the root-zone SM data is better.
- L121-125. The “es” and “ea” in Eq. (3) are not defined.
- Methods: So many parts are similar with Li et al. (2021).
- Section 2.2.2: The most commonly used copula in Archimedean family is Gumbel copula, rather than the Frank copula. Please strengthen your reason about only using Frank copula.
- L152-L153: “The compound dry and hot event was identified when the temperature was less than a threshold value and WSD was higher than a threshold”. I think this definition is wrong. For example, the temperature should be hot (rather than cold) when identifying a heat stress.
- Figure 2: The y-axis title “SCDHI” is the term in Li et al. (2021). Please revise it as your DCDHI.
- Section 3.2: Before using a copula, it is better that the dependence of variables can be explored. However, the authors wrongly present the correlation coefficients of DCDHI and SZI/STI. The two variables of copula function are SZI and STI. So, please present the correlation coefficients of SZI and STI. There are numerous copula-based literatures addressing this issue.
- Results: The main potential advances in this study is downscaling the sub-monthly drought indices to a daily scale. However, the authors mostly attempted to validate their compounding indices, failing to focus on the drought events.
- The authors use “significantly” in representing trends, but the significant test is missing.
- The main limitation and future works should be further discussed.
- The language should be polished.
Citation: https://doi.org/10.5194/hess-2022-217-RC1 -
AC1: 'Reply on RC1', Gengxi Zhang, 01 Nov 2022
We express our great appreciation for your constructive comments on improving the manuscript. We will revise our manuscript according to your comments. The answers to your concerns are as follows:
(1) There are so many indices for characterizing compound hot and dry events, such as Hao et al., 2019 and Wu et al. (2021). The authors claimed that these previous indices cannot monitor short-term events. Absolutely, monitoring the compounding events at a daily scale is important. Numerous studies have explored the extreme heat events by using daily indices, such as wet-bulb temperature and lethal heat stress index. The most challenge is to characterize the drought in a daily scale and then transform to the compounding events. This study used the existing drought indices by only changing the input from monthly to a daily scale. I am not persuasive that this is a great novelty in hydrological community. The authors should fully clarify this issue.
Response:
Thanks for your comments. Yes, there are some indices in monitoring daily heat events. Therefore, we changed the sentence to ‘these previous indices cannot monitor short-term compound dry and hot events. In addition, we will add some sentences about daily heat events indices (e.g., wet-bulb temperature and lethal heat stress index). We agree that the most challenge is to characterize the drought on a daily scale and then transform it into compound dry and hot events.
In the current study, we first extended the monthly SZI to a daily scale and then combined it with daily STI to monitor short-term compound dry and hot events. Although SZI is an existing drought index with good performance in drought monitoring across different climatic regions, we don’t know whether it can monitor short-term drought events, which occur more frequently in recent years due to climate change. Therefore, we should develop the daily SZI and verify its applicability.
Li et al. (2021) developed a short-term compound dry and hot events by combing SPEI and STI. We agree that SPEI is also a commonly used drought index, however, some studies (Ayantobo et al. 2020; Zhang et al. 2015, 2019, 2021) have reported that it always overestimates drought, especially across non-humid regions. The SPEI measures climatic water balance anomalies by incorporating the difference between precipitation (P; available water supply) and potential evapotranspiration (PET; atmospheric water demand). While PET is a good indicator for characterizing climate aridity, using it as a measure of atmospheric water demand for drought analysis leads to misrepresentation of droughts, especially over water-limited (non-humid) regions where the actual evapotranspiration is primarily dominated by water availability rather than energy (or PET) (Zhang et al., 2019). Compared to SPEI, SZI, using climatically appropriate precipitation for existing conditions as the atmospheric water demand metric, is physically more reasonable in reflecting surface water-energy balance over both humid and non-humid regions and can monitor different types of droughts in different climatic regions (Ayantobo et al. 2020; Zhang et al. 2015, 2019, 2021).
In addition, we will extend our research regions from non-arid regions of China to the whole regions of China and verify the applicability of daily SZI in drought monitoring across different climatic regions of China by comparing correlation distributions of SPEI-SM and 1 (Figs. R1 and R2).
Fig. R1 Correlations for SPEI-SM and SZI-SM at 3- and 6-month scales across China (Supplement Figure R1)
Figure R2 Boxplots of correlations for SPEI-SM and SZI-SM at 3- and 6-month scales at seven climatic regions of China (Supplement Figure R2)
(2) The authors claimed that they calculated the Standardized Temperature Index (STI) following the SPI estimation procedures. They fit the daily temperature data by using a normal distribution function; the normal assumption is OK. As the precipitation (temperature) in different months have large variations, the SPI (and STI) should be fitted in each month. Unfortunately, the authors fail to consider such seasonal variations in the present work.
Response: Actually, we calculated daily SZI and STI considering seasonal variations by fitting the daily WSD and temperature using log-logistic and normal distribution functions. For SZI, the daily difference (WSD) between precipitation and climatically appropriate precipitation was calculated, and then the daily WSD was fitted to the log-logistic distribution function for each calendar day. For example, the WSD of January 1st for all years was extracted and fitted to the log-logistic distribution, and the cumulative probability (P) was obtained and transformed to SZI. And we obtained SZI values for all calendar days with the same procedure. Finally, we rearranged these SZI values according to time orders and obtained daily SZI time series.
(3) The main methods in this study (e.g., both using WSD, root-zone soil moisture, STI, copula, run theory, and the same study region) is very similar with Li et al. (2021). The main differences may be just the input data, i.e., from sub-monthly to daily data. These issue should be fully clarified.
Response: In the current study, we first extended the monthly SZI to a daily scale and then combined it with daily STI to monitor short-term compound dry and hot events. Although SZI is an existing drought index with good performance in drought monitoring across different climatic regions, we don’t know whether it can monitor short-term drought events, which occur more frequently in recent years due to climate change. Therefore, we should develop the daily SZI and verify its applicability.
Li et al. (2021) developed a short-term compound dry and hot events by combing SPEI and STI. We agree that SPEI is also a commonly used drought index, however, some studies (Ayantobo et al. 2020; Zhang et al. 2015, 2019, 2021) have reported that it always overestimates drought, especially across non-humid regions. The SPEI measures climatic water balance anomalies by incorporating the difference between precipitation (P; available water supply) and potential evapotranspiration (PET; atmospheric water demand). While PET is a good indicator for characterizing climate aridity, using it as a measure of atmospheric water demand for drought analysis leads to misrepresentation of droughts, especially over water-limited (non-humid) regions where the actual evapotranspiration is primarily dominated by water availability rather than energy (or PET) (Zhang et al., 2019). Compared to SPEI, SZI, using climatically appropriate precipitation for existing conditions as the atmospheric water demand metric, is physically more reasonable in reflecting surface water-energy balance over both humid and non-humid regions and can monitor different types of droughts in different climatic regions (Ayantobo et al. 2020; Zhang et al. 2015, 2019, 2021).
In addition, we will extend our research regions from non-arid regions of China to the whole regions of China and verify the applicability of daily SZI in drought monitoring across different climatic regions of China by comparing correlation distributions of SPEI-SM and 1 (Figs. R1 and R2).We will revise the manuscript according to your all comments (including minor comments), and ask native speakers to polish the language.
Ayantobo, O. O., and Wei, J.: Appraising regional multi-category and multi-scalar drought monitoring using standardized
moisture anomaly index (SZI): A water-energy balance approach, Journal of Hydrology, 579, 124–139, DOI:
10.1016/j.jhydrol.2019.124139, 2019.Zhang, B., Zhao, X., Jin, J., and Wu, P.: Development and evaluation of a physically based multiscalar drought index: The
Standardized Moisture Anomaly Index, Journal of Geophysical Research: Atmospheres, 120, 11575–11588, DOI:
10.1002/2015JD023772, 2015.Zhang, B., Xia, Y., Huning, L. S., Wei, J., Wang, G., and AghaKouchak, A.: A framework for global multicategory and
multiscalar drought characterization accounting for snow processes, Water Resources Research, 55, 9258–9278, DOI:
10.1029/2019WR025529, 2019.Zhang, G., Su, X., Singh, V.P., and Ayantobo, O.O.: Appraising standardized moisture anomaly index (SZI) in drought
projection across China under CMIP6 forcing scenarios, Journal of Hydrology: Regional Studies, 37, 100898, DOI:
10.1016/j.ejrh.2021.100898, 2021.
-
RC2: 'Comment on hess-2022-217', Anonymous Referee #2, 12 Oct 2022
General Comments:
Under climate change, the occurrence of compound dry and hot events (CDHEs) has increased significantly, adversely affecting the socio-society and ecosystem. This summer, the extreme heat and drought that has been roasting a vast swath of southern China for at least 70 straight days has no parallel in modern record-keeping in China, or elsewhere around the world (e.g., Europe) for that matter. Furthermore, occurrences of short-term CDHEs have increased with global warming. However, most current indicators generally monitor CDHEs at monthly scales, which cannot reflect short-term CDHEs. Thus, it is important to develop a daily-scale index for monitoring short-term CDHEs, which can provide useful insights for understanding short-term compound dry and hot events and valuable information timely for stakeholders. Additionally, in future research, maybe you can calculate the daily drought index and DCDHI across global lands, which is helpful to evaluate evolutions of flash droughts and CHDEs in other regions. In a word, it is a topic of interest to the researchers in the related research but it can need some improvement before acceptance for publication. The specific comments are as follows:
Major comments:
(1) Logic-wise, the abstract can be organized better. I suggest authors re-organize the abstract.
(2) In the Introduction section, the connection between some sentences is abrupt. For example, the content in line 34 suddenly begins to describe the influence factors that control the evolution of CDHEs, which is not mentioned in the previous and afterward sentences. In addition, in line 46, the authors mentioned that the threshold-based method did not consider other characteristics of the compound events, which was not accurate because Feng et al. (2020) calculated frequency, severity, and duration based on this method. I suggest you can rewrite this sentence. Please check the syntax and logic of this Introduction section carefully.
(3) “p” was used in the left side of both equation (1), equation (8), and equation (9), but they have different meanings or represent probability calculated by different distribution functions. Therefore, authors should distinguish them, maybe pt in equation (1), pz in equation (8), pj in equation (9). Please check it. Accordingly, L143 “Lastly, the daily SZI was obtained by standardizing p as Eq (2)” should be more clarified. You use log-logistic cumulative distribution to fit WSD when calculating SZI. Please clarify why you chose this distribution or add some references.
(4) The authors calculated daily SPI, SPEI, SZI, STI, and DCDHI, but are these datasets available for publication? If possible, others can use these datasets to assess short-term droughts and CDHEs, which is helpful for related research.
(5) There are a lot of long sentences in the paper (complete sentences followed by V-ing+sentence instead of sentences connected with conjunction words). In other words, too much information is conveyed in one sentence. This may make readers confused. Thus, I suggest authors to rewrite or break long sentence into a few shorter sentences and make it more readable without changing the meaning. For instance, the sentence “Other joint extremes indices combing drought and hot indices, such as the Standardized Compound Event Indicator and Blended Dry and Hot Events Index, are more flexible than the threshold-based indices when analyzing the spatial and temporal characteristics of compound extremes” in the second paragraph of the Introduction includes 40 words and introduced another common method and its advantage compared with threshold-based indices. It will be better to use a few shorter sentences to convey the same information.
(6) More/appropriate conjunction words can be used to make the writing more fluent.
Minor comments:
(1) As you have given the abbreviation of compound dry and hot events, you should change ‘compound dry and hot events’ to ‘CDHEs’ in lines 20, 25, and 26. In line 27, the ‘new tool’ should be changed to ‘new index’. In addition, the authors should unify the uses of ‘index’ and ‘indicator’.
(2) Line 50. The ‘can not’ should be changed to ‘cannot’.
(3) Line 57. Can you give some examples to indicate that short-term CDHEs have increased?
(4) Line 61. 2. Please also check other abbreviations to make sure their unity. For instance, there are “SDHEs” and “SCDHI” in L61 and L65 respectively. Do they have the same meaning?
(5) Line 73. Maybe ‘flash drought’ is more suitable than ‘the short-term duration droughts’.
(6) Line 86. ‘north-western’ should be changed to ‘northwestern’, and ‘The’ should be changed to ‘the’.
(7) Line 135. ‘the jth month’ should be changed to ‘the jth month’.
(8) Line 152. The authors proposed DCDHI by linking SZI and STI. But the CDHE was still defined based on temperature and WSD (L152-L153 “The compound dry and hot event was identified when the temperature was less than a threshold value and WSD was higher than a threshold”) instead of SZI and STI, why is that? Moreover, the authors did not make the exact value of the threshold clear. What are the exact values of temperature and WSD to define the occurrence of dry and hot event in this study?
(9) Line 165. ‘compound characteristics’ should be changed to ‘characteristics of CDHEs’.
(10) Line 177. Before the evaluation of drought indices in drought monitoring, we don’t know which drought index is more suitable to construct DCDHI. Therefore, I suggest you change the sentence ‘The DCDHI was constructed based on daily SZI and STI’ to ‘The DCDHI was constructed based on daily drought and hot indices’.
(11) Line 178. Are ‘soil moisture or ‘normalized soil moisture used to evaluate drought indices? Please check the expression.
(12) Line 193. What is ‘SSMI’ mean? Please give the full name of the ‘SSMI’.
(13) Line 206. ‘north-eastern’ should be changed to ‘northeastern’. Change other similar expressions.
(14) Line 229. Remove the full stop before the word ‘accompanied’.
(15) Line 304, is it ‘between 1961-1989 and 1990-2021 periods’ or ‘between 1990-2021 and 1961-1989 periods’? please check the expression.
(16) You should add units for figures 10 and 11.
(17) Line 416, the journal name should be capitalized. Please check and unify reference formats carefully.
Reference:
Feng, S., Wu, X., Hao, Z., Hao, Y., Zhang, X., and Hao, F.: A database for characteristics and variations of global compound dry and hot events, Weather and Climate Extremes, 30, 100299, DOI: 10.1016/j.wace.2020.100299, 2020.
Citation: https://doi.org/10.5194/hess-2022-217-RC2 -
AC2: 'Reply on RC2', Gengxi Zhang, 01 Nov 2022
We express our great appreciation for your positive and constructive comments on improving the manuscript. We will revise our manuscript according to your comments. We will rewrite the abstract and revise the introduction when revising because we need to recalculate and reevaluate the index according to the comments of Reviewer#1 and Reviewer#3. We have compared different distributions for fitting SZI in Gengxi Zhang's thesis (Zhang, 2021), and we will give a simple description and refer to the thesis. We can provide our datasets as supplementary if we have a chance. And we will revise our manuscript according to your all comments.
Zhang, G. Effects of elevated CO2 concentration on future potential evapotranspiration and drought projection in China [D]. Yangling: Northwest A&F University, 2021.
Citation: https://doi.org/10.5194/hess-2022-217-AC2
-
AC2: 'Reply on RC2', Gengxi Zhang, 01 Nov 2022
-
RC3: 'Comment on hess-2022-217', Anonymous Referee #3, 19 Oct 2022
The study of Wang et al. proposes a novel daily-scale compound dry and hot index for assessing compound dry and hot events on a daily scale in non-arid regions in China. A standardized temperature based and a standardized soil moisture index are jointly modeled in a copula framework to create an index for short term drought and hot events. The index shows a general good agreement to soil moisture patterns in the study area and can reproduce compound dry and hot events for specific cases. However, in its current form I would recommend to reject the manuscript:
- First, the structure of the study is very similar to a study of Li et al. 2021. In my opinion, the main (technical) difference in the index derived by Li et al. and this study is the calculation of the evaporative demand. Moreover, the evaporative demand is only a constant for each month, divided by the number of days in the proposed index, where the potential evapotranspiration (Penman-Monteith) by Li et al. can be calculated for every day. Thereof, I am not sure if this study has the novelty to be published in HESS.
- Second, I am concerned about the application of copulas in the study. For the application of copulas the iid assumption has to be met. By applying a daily time series of soil moisture and temperature this is not true, as the time series are auto-correlated and the rising temperature over time violates the assumption of stationarity. Pitfalls and good practice for hydroclimatic applications of copulas can be found in e.g. Tootoonchi et al. 2022.
Additional general comments:
- Similar to the application of copulas the iid assumption also has to bet met for fitting a uniform distribution to soil moisture data and temperature.
- The evaluation is mainly performed against soil moisture data. Therefore, I suggest to use the term soil moisture drought explicitly throughout the manuscript, or to clearly define other drought types in dependency to soil moisture.
- The methods section is not very clear to me. I am missing the point which data is used for fitting the distribution for each of the indices (STI, SZI, DCDHI). Is each day fitted separately (as it is proposed by most standardized indices), or is one distribution fitted to the whole time series. Please check the formulas and restructure the methods section to make this more clear.
- One uncertainty for standardized indices as STI and SZI is to find a suitable distribution (Stagge et al. 2015). In case of the STI, I believe that the approximation by a normal distribution will be suitable for most data points. In the case of the SZI, I assume this may be dependent on the applied time-scale and on the geographic region. At least, I suggest to use a goodness of fit test for the log-logistic distribution and dismiss distributions that are not well-fitted to the data, or test other theoretical distributions.
Specific comments and technical corrections
Introduction
Line 36: I could not find Miralles et al. 2019 in the reference list.
Line 37: maybe delete “over global terrestrial regions”.
Line 38: Please check the use of the word “invaded” here.
Line 38 – Line 41: These results may be summarized into 1 sentence.
Line 45: I could not find Hao et al. 2013 in the reference list.
Line 51: “con not”
Line 56-59: Please add Li et al. 2021 as citation here.
Line 60: The index was called SAPEI in the study.
Line 63: SCDHI was not introduced.
Line 66: Please rephrase “over global maize areas”.
Line 73ff: The end of the introduction would may benefit if the research gap and the addressed research questions could be stated more explicitly. Additionally in the introduction there is no notice, that the indices are evaluated by SPI, SPEI, NDVI and soil moisture data.
Line 75: Please add what the hot index of your study is.
Data
Figure 1: Please check the legend of this figure. I am not sure if it is necessary to use a diverging color ramp, I suggest to at least report the middle threshold. Please check this throughout the whole manuscript.
Line 105 and Line 109: What is the purpose of transforming the soil moisture data and the NDVI?
Methods
Please check all formulas and their definitions.
Line 124ff: I could not find any reference where the data for the surface water balace is from? Please add this to the data section.
Line 149: which is a.
Line 153: Please check the sentence and the following formula.
Formula 10 and Table 1: In my opinion a cold and wet day would also yield a very low DCDHI, how was this approached in this study?
Line 165: What is run theory? Please define or at least add a citation.
Figure 2: The figure is very similar to Figure 3 of Li et al. 2021, either use this one and cite or create a new one. I believe on the y-axis there should be DCDHI and not the index of Li et al. - SCDHI.
Formula 11,12: I could not find any definition for t.
Results
Page 8: I am not quite sure, on which data the Pearson correlation coefficient (?) is calculated, but if it is the whole time series for each data point, then statements as Line185: “indicating SZI can better monitor drought across different climate regions”, or Line 189: “In a word, the daily SZI is a reliable indicator in monitoring drought at different time scales, …”, should be avoided. As the Pearson correlation coefficient only gives an impression of the overall correlation between the time series, which also includes medium and wet events.
Figure 3: I could not find a definition of SSMI in the caption. The resolution of the maps may be improved if areas where no results are shown are dismissed. Please check this also in all other figures. Additionally, the panel names ((a1),…) may be also discarded to improve the Figure.
Figure 4: A more informative plot may be the comparison between the different drought indices and dividing the panel by the different time scales.
Figure 5: I could not find a definition of the drought threshold, please clarify.
Line 210 (and 190): I could not find any definition of an agricultural drought. Which time-scales represent agricultural droughts for the soil moisture index? Please clarify.
Line 215: superior to what and how was this evaluated? Please clarify.
Line 218ff: I am not sure of the purpose of this paragraph, as the construction of a copula was to model the dependence between STI and SZI and now test the correlation between the input data (SZI, STI) and the modeled data (DCDHI). Please clarify, why this is necessary and not just report the parameters of the copula function?
Line 242: “with records in some published documents …” - please be specific, or at least cite the relevant documents.
Section 3.3. Please note my point on stationarity of the STI index. Further, why one period has 29 years and the other 32 years?
Figure 10: How are Duration, Severity, and Intensity defined for this plot – is it an average for each year as for the Frequency?
Literature
Stagge, J.H., Tallaksen, L.M., Gudmundsson, L., Van Loon, A.F. and Stahl, K. (2015), Candidate Distributions for Climatological Drought Indices (SPI and SPEI). Int. J. Climatol., 35: 4027-4040. https://doi.org/10.1002/joc.4267.
Tootoonchi, F., Sadegh, M., Haerter, J. O., Räty, O., Grabs, T., & Teutschbein, C. (2022). Copulas for hydroclimatic analysis: A practice-oriented overview. Wiley Interdisciplinary Reviews: Water, 9( 2), e1579. https://doi.org/10.1002/wat2.1579.
Li, J., Wang, Z., Wu, X., Zscheischler, J., Guo, S., and Chen, X.: A standardized index for assessing sub-monthly compound dry and hot conditions with application in China, Hydrol. Earth Syst. Sci., 25, 1587–1601, https://doi.org/10.5194/hess-25-1587-2021, 2021.
Citation: https://doi.org/10.5194/hess-2022-217-RC3 -
AC4: 'Reply on RC3', Gengxi Zhang, 09 Nov 2022
We express our great appreciation for your constructive comments on improving the manuscript. We will revise our manuscript according to your comments. The answers to your concerns are as follows:
First, the structure of the study is very similar to the study of Li et al. 2021. In my opinion, the main (technical) difference in the index derived by Li et al. and this study is the calculation of the evaporative demand. Moreover, the evaporative demand is only a constant for each month, divided by the number of days in the proposed index, where the potential evapotranspiration (Penman-Monteith) by Li et al. can be calculated for every day. Thereof, I am not sure if this study has the novelty to be published in HESS.
Second, I am concerned about the application of copulas in the study. For the application of Copulas the iid assumption has to be met. By applying a daily time series of soil moisture and temperature this is not true, as the time series are auto-correlated and the rising temperature over time violates the assumption of stationarity. Pitfalls and good practice for hydroclimatic applications of copulas can be found in e.g. Tootoonchi et al. 2022.
Response: In the current study, we first extended the monthly SZI to a daily scale and then combined it with daily STI to monitor short-term compound dry and hot events. Although SZI is an existing drought index with good performance in drought monitoring across different climatic regions, we don’t know whether it can monitor short-term drought events, which occur more frequently in recent years due to climate change. Therefore, we should develop the daily SZI and verify its applicability. Li et al. (2021) developed a short-term compound dry and hot events by combing SPEI and STI. We agree that SPEI is also a commonly used drought index, however, some studies (Ayantobo et al. 2020; Zhang et al. 2015, 2019, 2021) have reported that it always overestimates drought, especially across non-humid regions. The SPEI measures climatic water balance anomalies by incorporating the difference between precipitation (P; available water supply) and potential evapotranspiration (PET; atmospheric water demand). While PET is a good indicator for characterizing climate aridity, using it as a measure of atmospheric water demand for drought analysis leads to misrepresentation of droughts, especially over water-limited (non-humid) regions where the actual evapotranspiration is primarily dominated by water availability rather than energy (or PET) (Zhang et al., 2019). Compared to SPEI, SZI, using climatically appropriate precipitation for existing conditions (P ̂) as the atmospheric water demand metric, is physically more reasonable in reflecting surface water-energy balance over both humid and non-humid regions and can monitor different types of droughts in different climatic regions (Ayantobo et al. 2020; Zhang et al. 2015, 2019, 2021). In addition, we will extend our research regions from non-arid regions of China to the whole regions of China and verify the applicability of daily SZI in drought monitoring across different climatic regions of China.
Thanks for your suggestion about the application of copulas, we have read the reference carefully. Yes, we should test the iid assumption of variables before coupling them using copulas. We will test the iid assumption of the daily STI and SZI before using copulas when revising. If the variables are stationary and independent, we will join them using copulas, otherwise, we will first remove the autocorrelation using AR or ARMA models and then couple residuals using non-stationary (e.g., time-dependent parameter) copulas.
The evaluation is mainly performed against soil moisture data. Therefore, I suggest to use the term soil moisture drought explicitly throughout the manuscript, or to clearly define other drought types in dependency to soil moisture.
Response: We will use the term 'soil moisture drought' throughout the main text.
The methods section is not very clear to me. I am missing the point which data is used for fitting the distribution for each of the indices (STI, SZI, DCDHI). Is each day fitted separately (as it is proposed by most standardized indices), or is one distribution fitted to the whole time series. Please check the formulas and restructure the methods section to make this more clear.
Response: We fitted the daily temperature and SWD to normal and log-logistic distributions for each day. We will explain the procedure and restructure the methods when revising.
One uncertainty for standardized indices as STI and SZI is to find a suitable distribution (Stagge et al. 2015). In case of the STI, I believe that the approximation by a normal distribution will be suitable for most data points. In the case of the SZI, I assume this may be dependent on the applied time-scale and on the geographic region. At least, I suggest to use a goodness of fit test for the log-logistic distribution and dismiss distributions that are not well-fitted to the data, or test other theoretical distributions.
Response: Yes, we agree that geophysical locations or time scales may affect distribution types that WSD is subject to. We will compare and test different distributions for selecting the optimal distribution for SZI in different climatic regions and for various time scales.
We will revise the manuscript according to all your comments which are very helpful for improving our research. Thanks.
Citation: https://doi.org/10.5194/hess-2022-217-AC4
-
AC3: 'Comment on hess-2022-217', Gengxi Zhang, 02 Nov 2022
We express our great appreciation for your constructive comments on improving the manuscript. We will revise our manuscript according to your comments. The answers to your concerns are as follows:
First, the structure of the study is very similar to the study of Li et al. 2021. In my opinion, the main (technical) difference in the index derived by Li et al. and this study is the calculation of the evaporative demand. Moreover, the evaporative demand is only a constant for each month, divided by the number of days in the proposed index, where the potential evapotranspiration (Penman-Monteith) by Li et al. can be calculated for every day. Thereof, I am not sure if this study has the novelty to be published in HESS.
Second, I am concerned about the application of copulas in the study. For the application of Copulas the iid assumption has to be met. By applying a daily time series of soil moisture and temperature this is not true, as the time series are auto-correlated and the rising temperature over time violates the assumption of stationarity. Pitfalls and good practice for hydroclimatic applications of copulas can be found in e.g. Tootoonchi et al. 2022.Response: In the current study, we first extended the monthly SZI to a daily scale and then combined it with daily STI to monitor short-term compound dry and hot events. Although SZI is an existing drought index with good performance in drought monitoring across different climatic regions, we don’t know whether it can monitor short-term drought events, which occur more frequently in recent years due to climate change. Therefore, we should develop the daily SZI and verify its applicability. Li et al. (2021) developed a short-term compound dry and hot events by combing SPEI and STI. We agree that SPEI is also a commonly used drought index, however, some studies (Ayantobo et al. 2020; Zhang et al. 2015, 2019, 2021) have reported that it always overestimates drought, especially across non-humid regions. The SPEI measures climatic water balance anomalies by incorporating the difference between precipitation (P; available water supply) and potential evapotranspiration (PET; atmospheric water demand). While PET is a good indicator for characterizing climate aridity, using it as a measure of atmospheric water demand for drought analysis leads to misrepresentation of droughts, especially over water-limited (non-humid) regions where the actual evapotranspiration is primarily dominated by water availability rather than energy (or PET) (Zhang et al., 2019). Compared to SPEI, SZI, using climatically appropriate precipitation for existing conditions (P ̂) as the atmospheric water demand metric, is physically more reasonable in reflecting surface water-energy balance over both humid and non-humid regions and can monitor different types of droughts in different climatic regions (Ayantobo et al. 2020; Zhang et al. 2015, 2019, 2021). In addition, we will extend our research regions from non-arid regions of China to the whole regions of China and verify the applicability of daily SZI in drought monitoring across different climatic regions of China.
Thanks for your suggestion about the application of copulas, we have read the reference carefully. Yes, we should test the iid assumption of variables before coupling them using copulas. We will test the iid assumption of the daily STI and SZI before using copulas when revising. If the variables are stationary and independent, we will join them using copulas, otherwise, we will first remove the autocorrelation using AR or ARMA models and then couple residuals using non-stationary (e.g., time-dependent parameter) copulas.The evaluation is mainly performed against soil moisture data. Therefore, I suggest to use the term soil moisture drought explicitly throughout the manuscript, or to clearly define other drought types in dependency to soil moisture.
Response: We will use the term 'soil moisture drought' throughout the main text.
The methods section is not very clear to me. I am missing the point which data is used for fitting the distribution for each of the indices (STI, SZI, DCDHI). Is each day fitted separately (as it is proposed by most standardized indices), or is one distribution fitted to the whole time series. Please check the formulas and restructure the methods section to make this more clear.
Response: We fitted the daily temperature and SWD to normal and log-logistic distributions for each day. We will explain the procedure and restructure the methods when revising.
One uncertainty for standardized indices as STI and SZI is to find a suitable distribution (Stagge et al. 2015). In case of the STI, I believe that the approximation by a normal distribution will be suitable for most data points. In the case of the SZI, I assume this may be dependent on the applied time-scale and on the geographic region. At least, I suggest to use a goodness of fit test for the log-logistic distribution and dismiss distributions that are not well-fitted to the data, or test other theoretical distributions.
Response: Yes, we agree that geophysical locations or time scales may affect distribution types that WSD is subject to. We will compare and test different distributions for selecting the optimal distribution for SZI in different climatic regions and for various time scales.
We will revise the manuscript according to all your comments which are very helpful for improving our research. Thanks.
Citation: https://doi.org/10.5194/hess-2022-217-AC3
Status: closed
-
RC1: 'Comment on hess-2022-217', Anonymous Referee #1, 05 Oct 2022
Compounding dry and hot events are challenging the world under a changing climate. This study attempted to expanded the time scale of the Standardized Moisture Anomaly Index (SZI) from months to days for evaluating short-term droughts. Then, it was combined with the hot index to construct a daily scale compound dry and hot (DCDHI) index using Copula theory for investigating important characteristics of compound dry and hot events. This topic is interesting and within the scope of HESS. However, the novelty is not well clarified. This paper is very similar with Li et al. (2021), including methods, structures and figures (e.g, Figure 2). Therefore, I would recommend a Rejection at the present stage. If the authors can well address the following concerns, I also welcome to review the resubmitted version.
Li, J., Wang, Z., Wu, X., Zscheischler, J., Guo, S., and Chen, X.: A standardized index for assessing sub-monthly compound dry and hot conditions with application in China, Hydrology and Earth System Sciences, 25(3), 1587–1601, DOI: 10.5194/hess-25-1587-2021, 2021.
Major concerns:
- There are so many indices for characterizing compound hot and dry events, such as Hao et al., 2019 and Wu et al. (2021). The authors claimed that these previous indices cannot monitor short-term events. Absolutely, monitoring the compounding events at a daily scale is important. Numerous studies have explored the extreme heat events by using daily indices, such as wet-bulb temperature and lethal heat stress index. The most challenge is to characterize the drought in a daily scale and then transform to the compounding events. This study used the existing drought indices by only changing the input from monthly to a daily scale. I am not persuasive that this is a great novelty in hydrological community. The authors should fully clarify this issue.
Hao, Z., Hao, F., Singh, V. P., and Zhang, X.: Changes in the severity of compound drought and hot extremes over global land areas, Environmental Research Letters, 13, 124022, DOI: 10.1088/1748-9326/aaee96, 2018.
Wu, X., Hao, Z., Zhang, X., Li, C., and Hao, F.: Evaluation of severity changes of compound dry and hot events in China based on a multivariate multi-index approach, Journal of Hydrology, 583: 124580, DOI: 10.1016/j.jhydrol.2020.124580, 2020.
- The authors claimed that they calculated the Standardized Temperature Index (STI) following the SPI estimation procedures. They fit the daily temperature data by using a normal distribution function; the normal assumption is OK. As the precipitation (temperature) in different months have large variations, the SPI (and STI) should be fitted in each month. Unfortunately, the authors fail to consider such seasonal variations in the present work.
- To validate that the developed compounding index is better than others, the authors used the NDVI data. There might raise two concerns here. First, the authors use the root-zone soil moisture data to represent drought, which is surely more straightforward to link with vegetation growth. So, it is not surprising that the soil-based index can better represent the drought condition than SPI (and SPEI), which is in the type of meteorological drought. Second, it is not persuasive that they use the monthly NDVI data to validate their daily compounding index.
- The main methods in this study (e.g., both using WSD, root-zone soil moisture, STI, copula, run theory, and the same study region) is very similar with Li et al. (2021). The main differences may be just the input data, i.e., from sub-monthly to daily data. These issue should be fully clarified.
Minor concerns:
- Abstract: Some quantitative results should be presented to represent the increasing severity of compound events.
- L50: “con not” to “can not”.
- L61: the “SDHEs” is not defined.
- L63: the “SCDHI” is not defined.
- The ongoing topic of flash droughts is relevant and should be discussed.
- L86: How do you define the “non-arid” region?
- L104: “The root zone soil moisture data were used for the appropriateness in characterizing drought because they have lower noise than land surface soil water.” The reason about why using root zone soil moisture is not correctly explained. I think you want to focus on vegetation, and the root-zone SM data is better.
- L121-125. The “es” and “ea” in Eq. (3) are not defined.
- Methods: So many parts are similar with Li et al. (2021).
- Section 2.2.2: The most commonly used copula in Archimedean family is Gumbel copula, rather than the Frank copula. Please strengthen your reason about only using Frank copula.
- L152-L153: “The compound dry and hot event was identified when the temperature was less than a threshold value and WSD was higher than a threshold”. I think this definition is wrong. For example, the temperature should be hot (rather than cold) when identifying a heat stress.
- Figure 2: The y-axis title “SCDHI” is the term in Li et al. (2021). Please revise it as your DCDHI.
- Section 3.2: Before using a copula, it is better that the dependence of variables can be explored. However, the authors wrongly present the correlation coefficients of DCDHI and SZI/STI. The two variables of copula function are SZI and STI. So, please present the correlation coefficients of SZI and STI. There are numerous copula-based literatures addressing this issue.
- Results: The main potential advances in this study is downscaling the sub-monthly drought indices to a daily scale. However, the authors mostly attempted to validate their compounding indices, failing to focus on the drought events.
- The authors use “significantly” in representing trends, but the significant test is missing.
- The main limitation and future works should be further discussed.
- The language should be polished.
Citation: https://doi.org/10.5194/hess-2022-217-RC1 -
AC1: 'Reply on RC1', Gengxi Zhang, 01 Nov 2022
We express our great appreciation for your constructive comments on improving the manuscript. We will revise our manuscript according to your comments. The answers to your concerns are as follows:
(1) There are so many indices for characterizing compound hot and dry events, such as Hao et al., 2019 and Wu et al. (2021). The authors claimed that these previous indices cannot monitor short-term events. Absolutely, monitoring the compounding events at a daily scale is important. Numerous studies have explored the extreme heat events by using daily indices, such as wet-bulb temperature and lethal heat stress index. The most challenge is to characterize the drought in a daily scale and then transform to the compounding events. This study used the existing drought indices by only changing the input from monthly to a daily scale. I am not persuasive that this is a great novelty in hydrological community. The authors should fully clarify this issue.
Response:
Thanks for your comments. Yes, there are some indices in monitoring daily heat events. Therefore, we changed the sentence to ‘these previous indices cannot monitor short-term compound dry and hot events. In addition, we will add some sentences about daily heat events indices (e.g., wet-bulb temperature and lethal heat stress index). We agree that the most challenge is to characterize the drought on a daily scale and then transform it into compound dry and hot events.
In the current study, we first extended the monthly SZI to a daily scale and then combined it with daily STI to monitor short-term compound dry and hot events. Although SZI is an existing drought index with good performance in drought monitoring across different climatic regions, we don’t know whether it can monitor short-term drought events, which occur more frequently in recent years due to climate change. Therefore, we should develop the daily SZI and verify its applicability.
Li et al. (2021) developed a short-term compound dry and hot events by combing SPEI and STI. We agree that SPEI is also a commonly used drought index, however, some studies (Ayantobo et al. 2020; Zhang et al. 2015, 2019, 2021) have reported that it always overestimates drought, especially across non-humid regions. The SPEI measures climatic water balance anomalies by incorporating the difference between precipitation (P; available water supply) and potential evapotranspiration (PET; atmospheric water demand). While PET is a good indicator for characterizing climate aridity, using it as a measure of atmospheric water demand for drought analysis leads to misrepresentation of droughts, especially over water-limited (non-humid) regions where the actual evapotranspiration is primarily dominated by water availability rather than energy (or PET) (Zhang et al., 2019). Compared to SPEI, SZI, using climatically appropriate precipitation for existing conditions as the atmospheric water demand metric, is physically more reasonable in reflecting surface water-energy balance over both humid and non-humid regions and can monitor different types of droughts in different climatic regions (Ayantobo et al. 2020; Zhang et al. 2015, 2019, 2021).
In addition, we will extend our research regions from non-arid regions of China to the whole regions of China and verify the applicability of daily SZI in drought monitoring across different climatic regions of China by comparing correlation distributions of SPEI-SM and 1 (Figs. R1 and R2).
Fig. R1 Correlations for SPEI-SM and SZI-SM at 3- and 6-month scales across China (Supplement Figure R1)
Figure R2 Boxplots of correlations for SPEI-SM and SZI-SM at 3- and 6-month scales at seven climatic regions of China (Supplement Figure R2)
(2) The authors claimed that they calculated the Standardized Temperature Index (STI) following the SPI estimation procedures. They fit the daily temperature data by using a normal distribution function; the normal assumption is OK. As the precipitation (temperature) in different months have large variations, the SPI (and STI) should be fitted in each month. Unfortunately, the authors fail to consider such seasonal variations in the present work.
Response: Actually, we calculated daily SZI and STI considering seasonal variations by fitting the daily WSD and temperature using log-logistic and normal distribution functions. For SZI, the daily difference (WSD) between precipitation and climatically appropriate precipitation was calculated, and then the daily WSD was fitted to the log-logistic distribution function for each calendar day. For example, the WSD of January 1st for all years was extracted and fitted to the log-logistic distribution, and the cumulative probability (P) was obtained and transformed to SZI. And we obtained SZI values for all calendar days with the same procedure. Finally, we rearranged these SZI values according to time orders and obtained daily SZI time series.
(3) The main methods in this study (e.g., both using WSD, root-zone soil moisture, STI, copula, run theory, and the same study region) is very similar with Li et al. (2021). The main differences may be just the input data, i.e., from sub-monthly to daily data. These issue should be fully clarified.
Response: In the current study, we first extended the monthly SZI to a daily scale and then combined it with daily STI to monitor short-term compound dry and hot events. Although SZI is an existing drought index with good performance in drought monitoring across different climatic regions, we don’t know whether it can monitor short-term drought events, which occur more frequently in recent years due to climate change. Therefore, we should develop the daily SZI and verify its applicability.
Li et al. (2021) developed a short-term compound dry and hot events by combing SPEI and STI. We agree that SPEI is also a commonly used drought index, however, some studies (Ayantobo et al. 2020; Zhang et al. 2015, 2019, 2021) have reported that it always overestimates drought, especially across non-humid regions. The SPEI measures climatic water balance anomalies by incorporating the difference between precipitation (P; available water supply) and potential evapotranspiration (PET; atmospheric water demand). While PET is a good indicator for characterizing climate aridity, using it as a measure of atmospheric water demand for drought analysis leads to misrepresentation of droughts, especially over water-limited (non-humid) regions where the actual evapotranspiration is primarily dominated by water availability rather than energy (or PET) (Zhang et al., 2019). Compared to SPEI, SZI, using climatically appropriate precipitation for existing conditions as the atmospheric water demand metric, is physically more reasonable in reflecting surface water-energy balance over both humid and non-humid regions and can monitor different types of droughts in different climatic regions (Ayantobo et al. 2020; Zhang et al. 2015, 2019, 2021).
In addition, we will extend our research regions from non-arid regions of China to the whole regions of China and verify the applicability of daily SZI in drought monitoring across different climatic regions of China by comparing correlation distributions of SPEI-SM and 1 (Figs. R1 and R2).We will revise the manuscript according to your all comments (including minor comments), and ask native speakers to polish the language.
Ayantobo, O. O., and Wei, J.: Appraising regional multi-category and multi-scalar drought monitoring using standardized
moisture anomaly index (SZI): A water-energy balance approach, Journal of Hydrology, 579, 124–139, DOI:
10.1016/j.jhydrol.2019.124139, 2019.Zhang, B., Zhao, X., Jin, J., and Wu, P.: Development and evaluation of a physically based multiscalar drought index: The
Standardized Moisture Anomaly Index, Journal of Geophysical Research: Atmospheres, 120, 11575–11588, DOI:
10.1002/2015JD023772, 2015.Zhang, B., Xia, Y., Huning, L. S., Wei, J., Wang, G., and AghaKouchak, A.: A framework for global multicategory and
multiscalar drought characterization accounting for snow processes, Water Resources Research, 55, 9258–9278, DOI:
10.1029/2019WR025529, 2019.Zhang, G., Su, X., Singh, V.P., and Ayantobo, O.O.: Appraising standardized moisture anomaly index (SZI) in drought
projection across China under CMIP6 forcing scenarios, Journal of Hydrology: Regional Studies, 37, 100898, DOI:
10.1016/j.ejrh.2021.100898, 2021.
-
RC2: 'Comment on hess-2022-217', Anonymous Referee #2, 12 Oct 2022
General Comments:
Under climate change, the occurrence of compound dry and hot events (CDHEs) has increased significantly, adversely affecting the socio-society and ecosystem. This summer, the extreme heat and drought that has been roasting a vast swath of southern China for at least 70 straight days has no parallel in modern record-keeping in China, or elsewhere around the world (e.g., Europe) for that matter. Furthermore, occurrences of short-term CDHEs have increased with global warming. However, most current indicators generally monitor CDHEs at monthly scales, which cannot reflect short-term CDHEs. Thus, it is important to develop a daily-scale index for monitoring short-term CDHEs, which can provide useful insights for understanding short-term compound dry and hot events and valuable information timely for stakeholders. Additionally, in future research, maybe you can calculate the daily drought index and DCDHI across global lands, which is helpful to evaluate evolutions of flash droughts and CHDEs in other regions. In a word, it is a topic of interest to the researchers in the related research but it can need some improvement before acceptance for publication. The specific comments are as follows:
Major comments:
(1) Logic-wise, the abstract can be organized better. I suggest authors re-organize the abstract.
(2) In the Introduction section, the connection between some sentences is abrupt. For example, the content in line 34 suddenly begins to describe the influence factors that control the evolution of CDHEs, which is not mentioned in the previous and afterward sentences. In addition, in line 46, the authors mentioned that the threshold-based method did not consider other characteristics of the compound events, which was not accurate because Feng et al. (2020) calculated frequency, severity, and duration based on this method. I suggest you can rewrite this sentence. Please check the syntax and logic of this Introduction section carefully.
(3) “p” was used in the left side of both equation (1), equation (8), and equation (9), but they have different meanings or represent probability calculated by different distribution functions. Therefore, authors should distinguish them, maybe pt in equation (1), pz in equation (8), pj in equation (9). Please check it. Accordingly, L143 “Lastly, the daily SZI was obtained by standardizing p as Eq (2)” should be more clarified. You use log-logistic cumulative distribution to fit WSD when calculating SZI. Please clarify why you chose this distribution or add some references.
(4) The authors calculated daily SPI, SPEI, SZI, STI, and DCDHI, but are these datasets available for publication? If possible, others can use these datasets to assess short-term droughts and CDHEs, which is helpful for related research.
(5) There are a lot of long sentences in the paper (complete sentences followed by V-ing+sentence instead of sentences connected with conjunction words). In other words, too much information is conveyed in one sentence. This may make readers confused. Thus, I suggest authors to rewrite or break long sentence into a few shorter sentences and make it more readable without changing the meaning. For instance, the sentence “Other joint extremes indices combing drought and hot indices, such as the Standardized Compound Event Indicator and Blended Dry and Hot Events Index, are more flexible than the threshold-based indices when analyzing the spatial and temporal characteristics of compound extremes” in the second paragraph of the Introduction includes 40 words and introduced another common method and its advantage compared with threshold-based indices. It will be better to use a few shorter sentences to convey the same information.
(6) More/appropriate conjunction words can be used to make the writing more fluent.
Minor comments:
(1) As you have given the abbreviation of compound dry and hot events, you should change ‘compound dry and hot events’ to ‘CDHEs’ in lines 20, 25, and 26. In line 27, the ‘new tool’ should be changed to ‘new index’. In addition, the authors should unify the uses of ‘index’ and ‘indicator’.
(2) Line 50. The ‘can not’ should be changed to ‘cannot’.
(3) Line 57. Can you give some examples to indicate that short-term CDHEs have increased?
(4) Line 61. 2. Please also check other abbreviations to make sure their unity. For instance, there are “SDHEs” and “SCDHI” in L61 and L65 respectively. Do they have the same meaning?
(5) Line 73. Maybe ‘flash drought’ is more suitable than ‘the short-term duration droughts’.
(6) Line 86. ‘north-western’ should be changed to ‘northwestern’, and ‘The’ should be changed to ‘the’.
(7) Line 135. ‘the jth month’ should be changed to ‘the jth month’.
(8) Line 152. The authors proposed DCDHI by linking SZI and STI. But the CDHE was still defined based on temperature and WSD (L152-L153 “The compound dry and hot event was identified when the temperature was less than a threshold value and WSD was higher than a threshold”) instead of SZI and STI, why is that? Moreover, the authors did not make the exact value of the threshold clear. What are the exact values of temperature and WSD to define the occurrence of dry and hot event in this study?
(9) Line 165. ‘compound characteristics’ should be changed to ‘characteristics of CDHEs’.
(10) Line 177. Before the evaluation of drought indices in drought monitoring, we don’t know which drought index is more suitable to construct DCDHI. Therefore, I suggest you change the sentence ‘The DCDHI was constructed based on daily SZI and STI’ to ‘The DCDHI was constructed based on daily drought and hot indices’.
(11) Line 178. Are ‘soil moisture or ‘normalized soil moisture used to evaluate drought indices? Please check the expression.
(12) Line 193. What is ‘SSMI’ mean? Please give the full name of the ‘SSMI’.
(13) Line 206. ‘north-eastern’ should be changed to ‘northeastern’. Change other similar expressions.
(14) Line 229. Remove the full stop before the word ‘accompanied’.
(15) Line 304, is it ‘between 1961-1989 and 1990-2021 periods’ or ‘between 1990-2021 and 1961-1989 periods’? please check the expression.
(16) You should add units for figures 10 and 11.
(17) Line 416, the journal name should be capitalized. Please check and unify reference formats carefully.
Reference:
Feng, S., Wu, X., Hao, Z., Hao, Y., Zhang, X., and Hao, F.: A database for characteristics and variations of global compound dry and hot events, Weather and Climate Extremes, 30, 100299, DOI: 10.1016/j.wace.2020.100299, 2020.
Citation: https://doi.org/10.5194/hess-2022-217-RC2 -
AC2: 'Reply on RC2', Gengxi Zhang, 01 Nov 2022
We express our great appreciation for your positive and constructive comments on improving the manuscript. We will revise our manuscript according to your comments. We will rewrite the abstract and revise the introduction when revising because we need to recalculate and reevaluate the index according to the comments of Reviewer#1 and Reviewer#3. We have compared different distributions for fitting SZI in Gengxi Zhang's thesis (Zhang, 2021), and we will give a simple description and refer to the thesis. We can provide our datasets as supplementary if we have a chance. And we will revise our manuscript according to your all comments.
Zhang, G. Effects of elevated CO2 concentration on future potential evapotranspiration and drought projection in China [D]. Yangling: Northwest A&F University, 2021.
Citation: https://doi.org/10.5194/hess-2022-217-AC2
-
AC2: 'Reply on RC2', Gengxi Zhang, 01 Nov 2022
-
RC3: 'Comment on hess-2022-217', Anonymous Referee #3, 19 Oct 2022
The study of Wang et al. proposes a novel daily-scale compound dry and hot index for assessing compound dry and hot events on a daily scale in non-arid regions in China. A standardized temperature based and a standardized soil moisture index are jointly modeled in a copula framework to create an index for short term drought and hot events. The index shows a general good agreement to soil moisture patterns in the study area and can reproduce compound dry and hot events for specific cases. However, in its current form I would recommend to reject the manuscript:
- First, the structure of the study is very similar to a study of Li et al. 2021. In my opinion, the main (technical) difference in the index derived by Li et al. and this study is the calculation of the evaporative demand. Moreover, the evaporative demand is only a constant for each month, divided by the number of days in the proposed index, where the potential evapotranspiration (Penman-Monteith) by Li et al. can be calculated for every day. Thereof, I am not sure if this study has the novelty to be published in HESS.
- Second, I am concerned about the application of copulas in the study. For the application of copulas the iid assumption has to be met. By applying a daily time series of soil moisture and temperature this is not true, as the time series are auto-correlated and the rising temperature over time violates the assumption of stationarity. Pitfalls and good practice for hydroclimatic applications of copulas can be found in e.g. Tootoonchi et al. 2022.
Additional general comments:
- Similar to the application of copulas the iid assumption also has to bet met for fitting a uniform distribution to soil moisture data and temperature.
- The evaluation is mainly performed against soil moisture data. Therefore, I suggest to use the term soil moisture drought explicitly throughout the manuscript, or to clearly define other drought types in dependency to soil moisture.
- The methods section is not very clear to me. I am missing the point which data is used for fitting the distribution for each of the indices (STI, SZI, DCDHI). Is each day fitted separately (as it is proposed by most standardized indices), or is one distribution fitted to the whole time series. Please check the formulas and restructure the methods section to make this more clear.
- One uncertainty for standardized indices as STI and SZI is to find a suitable distribution (Stagge et al. 2015). In case of the STI, I believe that the approximation by a normal distribution will be suitable for most data points. In the case of the SZI, I assume this may be dependent on the applied time-scale and on the geographic region. At least, I suggest to use a goodness of fit test for the log-logistic distribution and dismiss distributions that are not well-fitted to the data, or test other theoretical distributions.
Specific comments and technical corrections
Introduction
Line 36: I could not find Miralles et al. 2019 in the reference list.
Line 37: maybe delete “over global terrestrial regions”.
Line 38: Please check the use of the word “invaded” here.
Line 38 – Line 41: These results may be summarized into 1 sentence.
Line 45: I could not find Hao et al. 2013 in the reference list.
Line 51: “con not”
Line 56-59: Please add Li et al. 2021 as citation here.
Line 60: The index was called SAPEI in the study.
Line 63: SCDHI was not introduced.
Line 66: Please rephrase “over global maize areas”.
Line 73ff: The end of the introduction would may benefit if the research gap and the addressed research questions could be stated more explicitly. Additionally in the introduction there is no notice, that the indices are evaluated by SPI, SPEI, NDVI and soil moisture data.
Line 75: Please add what the hot index of your study is.
Data
Figure 1: Please check the legend of this figure. I am not sure if it is necessary to use a diverging color ramp, I suggest to at least report the middle threshold. Please check this throughout the whole manuscript.
Line 105 and Line 109: What is the purpose of transforming the soil moisture data and the NDVI?
Methods
Please check all formulas and their definitions.
Line 124ff: I could not find any reference where the data for the surface water balace is from? Please add this to the data section.
Line 149: which is a.
Line 153: Please check the sentence and the following formula.
Formula 10 and Table 1: In my opinion a cold and wet day would also yield a very low DCDHI, how was this approached in this study?
Line 165: What is run theory? Please define or at least add a citation.
Figure 2: The figure is very similar to Figure 3 of Li et al. 2021, either use this one and cite or create a new one. I believe on the y-axis there should be DCDHI and not the index of Li et al. - SCDHI.
Formula 11,12: I could not find any definition for t.
Results
Page 8: I am not quite sure, on which data the Pearson correlation coefficient (?) is calculated, but if it is the whole time series for each data point, then statements as Line185: “indicating SZI can better monitor drought across different climate regions”, or Line 189: “In a word, the daily SZI is a reliable indicator in monitoring drought at different time scales, …”, should be avoided. As the Pearson correlation coefficient only gives an impression of the overall correlation between the time series, which also includes medium and wet events.
Figure 3: I could not find a definition of SSMI in the caption. The resolution of the maps may be improved if areas where no results are shown are dismissed. Please check this also in all other figures. Additionally, the panel names ((a1),…) may be also discarded to improve the Figure.
Figure 4: A more informative plot may be the comparison between the different drought indices and dividing the panel by the different time scales.
Figure 5: I could not find a definition of the drought threshold, please clarify.
Line 210 (and 190): I could not find any definition of an agricultural drought. Which time-scales represent agricultural droughts for the soil moisture index? Please clarify.
Line 215: superior to what and how was this evaluated? Please clarify.
Line 218ff: I am not sure of the purpose of this paragraph, as the construction of a copula was to model the dependence between STI and SZI and now test the correlation between the input data (SZI, STI) and the modeled data (DCDHI). Please clarify, why this is necessary and not just report the parameters of the copula function?
Line 242: “with records in some published documents …” - please be specific, or at least cite the relevant documents.
Section 3.3. Please note my point on stationarity of the STI index. Further, why one period has 29 years and the other 32 years?
Figure 10: How are Duration, Severity, and Intensity defined for this plot – is it an average for each year as for the Frequency?
Literature
Stagge, J.H., Tallaksen, L.M., Gudmundsson, L., Van Loon, A.F. and Stahl, K. (2015), Candidate Distributions for Climatological Drought Indices (SPI and SPEI). Int. J. Climatol., 35: 4027-4040. https://doi.org/10.1002/joc.4267.
Tootoonchi, F., Sadegh, M., Haerter, J. O., Räty, O., Grabs, T., & Teutschbein, C. (2022). Copulas for hydroclimatic analysis: A practice-oriented overview. Wiley Interdisciplinary Reviews: Water, 9( 2), e1579. https://doi.org/10.1002/wat2.1579.
Li, J., Wang, Z., Wu, X., Zscheischler, J., Guo, S., and Chen, X.: A standardized index for assessing sub-monthly compound dry and hot conditions with application in China, Hydrol. Earth Syst. Sci., 25, 1587–1601, https://doi.org/10.5194/hess-25-1587-2021, 2021.
Citation: https://doi.org/10.5194/hess-2022-217-RC3 -
AC4: 'Reply on RC3', Gengxi Zhang, 09 Nov 2022
We express our great appreciation for your constructive comments on improving the manuscript. We will revise our manuscript according to your comments. The answers to your concerns are as follows:
First, the structure of the study is very similar to the study of Li et al. 2021. In my opinion, the main (technical) difference in the index derived by Li et al. and this study is the calculation of the evaporative demand. Moreover, the evaporative demand is only a constant for each month, divided by the number of days in the proposed index, where the potential evapotranspiration (Penman-Monteith) by Li et al. can be calculated for every day. Thereof, I am not sure if this study has the novelty to be published in HESS.
Second, I am concerned about the application of copulas in the study. For the application of Copulas the iid assumption has to be met. By applying a daily time series of soil moisture and temperature this is not true, as the time series are auto-correlated and the rising temperature over time violates the assumption of stationarity. Pitfalls and good practice for hydroclimatic applications of copulas can be found in e.g. Tootoonchi et al. 2022.
Response: In the current study, we first extended the monthly SZI to a daily scale and then combined it with daily STI to monitor short-term compound dry and hot events. Although SZI is an existing drought index with good performance in drought monitoring across different climatic regions, we don’t know whether it can monitor short-term drought events, which occur more frequently in recent years due to climate change. Therefore, we should develop the daily SZI and verify its applicability. Li et al. (2021) developed a short-term compound dry and hot events by combing SPEI and STI. We agree that SPEI is also a commonly used drought index, however, some studies (Ayantobo et al. 2020; Zhang et al. 2015, 2019, 2021) have reported that it always overestimates drought, especially across non-humid regions. The SPEI measures climatic water balance anomalies by incorporating the difference between precipitation (P; available water supply) and potential evapotranspiration (PET; atmospheric water demand). While PET is a good indicator for characterizing climate aridity, using it as a measure of atmospheric water demand for drought analysis leads to misrepresentation of droughts, especially over water-limited (non-humid) regions where the actual evapotranspiration is primarily dominated by water availability rather than energy (or PET) (Zhang et al., 2019). Compared to SPEI, SZI, using climatically appropriate precipitation for existing conditions (P ̂) as the atmospheric water demand metric, is physically more reasonable in reflecting surface water-energy balance over both humid and non-humid regions and can monitor different types of droughts in different climatic regions (Ayantobo et al. 2020; Zhang et al. 2015, 2019, 2021). In addition, we will extend our research regions from non-arid regions of China to the whole regions of China and verify the applicability of daily SZI in drought monitoring across different climatic regions of China.
Thanks for your suggestion about the application of copulas, we have read the reference carefully. Yes, we should test the iid assumption of variables before coupling them using copulas. We will test the iid assumption of the daily STI and SZI before using copulas when revising. If the variables are stationary and independent, we will join them using copulas, otherwise, we will first remove the autocorrelation using AR or ARMA models and then couple residuals using non-stationary (e.g., time-dependent parameter) copulas.
The evaluation is mainly performed against soil moisture data. Therefore, I suggest to use the term soil moisture drought explicitly throughout the manuscript, or to clearly define other drought types in dependency to soil moisture.
Response: We will use the term 'soil moisture drought' throughout the main text.
The methods section is not very clear to me. I am missing the point which data is used for fitting the distribution for each of the indices (STI, SZI, DCDHI). Is each day fitted separately (as it is proposed by most standardized indices), or is one distribution fitted to the whole time series. Please check the formulas and restructure the methods section to make this more clear.
Response: We fitted the daily temperature and SWD to normal and log-logistic distributions for each day. We will explain the procedure and restructure the methods when revising.
One uncertainty for standardized indices as STI and SZI is to find a suitable distribution (Stagge et al. 2015). In case of the STI, I believe that the approximation by a normal distribution will be suitable for most data points. In the case of the SZI, I assume this may be dependent on the applied time-scale and on the geographic region. At least, I suggest to use a goodness of fit test for the log-logistic distribution and dismiss distributions that are not well-fitted to the data, or test other theoretical distributions.
Response: Yes, we agree that geophysical locations or time scales may affect distribution types that WSD is subject to. We will compare and test different distributions for selecting the optimal distribution for SZI in different climatic regions and for various time scales.
We will revise the manuscript according to all your comments which are very helpful for improving our research. Thanks.
Citation: https://doi.org/10.5194/hess-2022-217-AC4
-
AC3: 'Comment on hess-2022-217', Gengxi Zhang, 02 Nov 2022
We express our great appreciation for your constructive comments on improving the manuscript. We will revise our manuscript according to your comments. The answers to your concerns are as follows:
First, the structure of the study is very similar to the study of Li et al. 2021. In my opinion, the main (technical) difference in the index derived by Li et al. and this study is the calculation of the evaporative demand. Moreover, the evaporative demand is only a constant for each month, divided by the number of days in the proposed index, where the potential evapotranspiration (Penman-Monteith) by Li et al. can be calculated for every day. Thereof, I am not sure if this study has the novelty to be published in HESS.
Second, I am concerned about the application of copulas in the study. For the application of Copulas the iid assumption has to be met. By applying a daily time series of soil moisture and temperature this is not true, as the time series are auto-correlated and the rising temperature over time violates the assumption of stationarity. Pitfalls and good practice for hydroclimatic applications of copulas can be found in e.g. Tootoonchi et al. 2022.Response: In the current study, we first extended the monthly SZI to a daily scale and then combined it with daily STI to monitor short-term compound dry and hot events. Although SZI is an existing drought index with good performance in drought monitoring across different climatic regions, we don’t know whether it can monitor short-term drought events, which occur more frequently in recent years due to climate change. Therefore, we should develop the daily SZI and verify its applicability. Li et al. (2021) developed a short-term compound dry and hot events by combing SPEI and STI. We agree that SPEI is also a commonly used drought index, however, some studies (Ayantobo et al. 2020; Zhang et al. 2015, 2019, 2021) have reported that it always overestimates drought, especially across non-humid regions. The SPEI measures climatic water balance anomalies by incorporating the difference between precipitation (P; available water supply) and potential evapotranspiration (PET; atmospheric water demand). While PET is a good indicator for characterizing climate aridity, using it as a measure of atmospheric water demand for drought analysis leads to misrepresentation of droughts, especially over water-limited (non-humid) regions where the actual evapotranspiration is primarily dominated by water availability rather than energy (or PET) (Zhang et al., 2019). Compared to SPEI, SZI, using climatically appropriate precipitation for existing conditions (P ̂) as the atmospheric water demand metric, is physically more reasonable in reflecting surface water-energy balance over both humid and non-humid regions and can monitor different types of droughts in different climatic regions (Ayantobo et al. 2020; Zhang et al. 2015, 2019, 2021). In addition, we will extend our research regions from non-arid regions of China to the whole regions of China and verify the applicability of daily SZI in drought monitoring across different climatic regions of China.
Thanks for your suggestion about the application of copulas, we have read the reference carefully. Yes, we should test the iid assumption of variables before coupling them using copulas. We will test the iid assumption of the daily STI and SZI before using copulas when revising. If the variables are stationary and independent, we will join them using copulas, otherwise, we will first remove the autocorrelation using AR or ARMA models and then couple residuals using non-stationary (e.g., time-dependent parameter) copulas.The evaluation is mainly performed against soil moisture data. Therefore, I suggest to use the term soil moisture drought explicitly throughout the manuscript, or to clearly define other drought types in dependency to soil moisture.
Response: We will use the term 'soil moisture drought' throughout the main text.
The methods section is not very clear to me. I am missing the point which data is used for fitting the distribution for each of the indices (STI, SZI, DCDHI). Is each day fitted separately (as it is proposed by most standardized indices), or is one distribution fitted to the whole time series. Please check the formulas and restructure the methods section to make this more clear.
Response: We fitted the daily temperature and SWD to normal and log-logistic distributions for each day. We will explain the procedure and restructure the methods when revising.
One uncertainty for standardized indices as STI and SZI is to find a suitable distribution (Stagge et al. 2015). In case of the STI, I believe that the approximation by a normal distribution will be suitable for most data points. In the case of the SZI, I assume this may be dependent on the applied time-scale and on the geographic region. At least, I suggest to use a goodness of fit test for the log-logistic distribution and dismiss distributions that are not well-fitted to the data, or test other theoretical distributions.
Response: Yes, we agree that geophysical locations or time scales may affect distribution types that WSD is subject to. We will compare and test different distributions for selecting the optimal distribution for SZI in different climatic regions and for various time scales.
We will revise the manuscript according to all your comments which are very helpful for improving our research. Thanks.
Citation: https://doi.org/10.5194/hess-2022-217-AC3
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 780 | 320 | 53 | 1,153 | 41 | 35 |
- HTML: 780
- PDF: 320
- XML: 53
- Total: 1,153
- BibTeX: 41
- EndNote: 35
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1