the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 31 May 2024
Global catalog of soil moisture droughts over the past four decades
Abstract. At the global scale, droughts can be described by many variables, expressing their extent, duration, dynamics and severity. To identify common features in global land drought events (GLDEs) based on soil moisture, we present a robust method for their identification and classification (cataloging). Gridded estimates of root-zone soil moisture from the SoilClim model and the mesoscale Hydrologic Model (mHM) were calculated over global land from 1980–2022. Using the 10th percentile thresholds of soil moisture anomalies and OPTICS clustering of the gridded data in a 10-day interval, a total of 775 GLDEs from SoilClim and 630 GLDEs from mHM were identified. By utilizing four spatiotemporal and three motion-related characteristics for each GLDE, we established threshold percentiles based on their distributions. This information enabled us to categorize droughts into seven severity categories (ranging from extremely weak to extremely severe) and seven dynamic categories (ranging from extremely static to extremely dynamic). Our global-scale synthesis revealed the highest relative proportions of extremely severe and extremely dynamic GLDEs in the South American region, followed by North America, while the single longest and most extensive GLDEs occurred in Eurasia. The severity and dynamic categories overlapped substantially for extremely severe and extremely dynamic droughts but very little for less severe/dynamic categories, despite some very small droughts that have occasionally been very dynamic. The frequency of GLDEs has generally increased in recent decades across different drought categories but is statistically significant only in some cases. Overall, the cataloging of GLDEs presents a unique opportunity to analyze the evolving features of spatiotemporally connected drought events in recent decades and provides a basis for future investigations of the drivers and impacts of dynamically evolving drought events.
- Preprint
(2767 KB) - Metadata XML
-
Supplement
(111 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2024-1434', Samuel Jonson Sutanto, 08 Jul 2024
Title: Global catalog of soil moisture droughts over the past four decades
Summary
This paper analyses global soil moisture drought using data derived from the SoilClim and mHM models from 1980 to 2022. A drought catalog was compiled based on severity and dynamic drought classifications utilizing a threshold approach and the OPTICS clustering technique. The authors identified hotspot regions for extremely severe and extremely dynamic drought events in the South American region and North America. The longest and most extensive droughts occurred in Eurasia. The authors highlighted an increase in global land drought events over the past decade. Furthermore, they also suggest that this study may serve as a basis for future investigations into drought drivers and impacts.
Assessment
This paper presents a new technique to analyze drought classification based on severity and dynamic classifications. The manuscript is interesting and well written. I have a few minor comments below and three general comments, but only for clarification and improvement. I believe this work is well suited for publication in HESS.
General Comments
I have two general comments regarding the manuscript but all of them are only for clarification, suggestion, and improvement of the manuscript.
- I am curious as to why did the authors combine Europe and Asia continents into a single region called Eurasia? This combination results in a significantly high number of Global Land Drought Events (GLDEs) in this region, as depicted in Figures 3, 6, and 10. Furthermore, as a reader, I would prefer to see separate results for Europe and Asia. I suggest splitting the Eurasia results into distinct sections for Europe and Asia. This approach would likely be more engaging for readers from these two continents.
- I am confused about the ongoing drought occurrences from November 2004 to December 2022 in Table 1, which spans over 18 years. How did you analyze this multi-year, prolonged drought duration in Eurasia? I believe this period likely consists of multiple drought events in Europe and Asia occurring from November 2004 to December 2022 in different regions. Please correct me if I am mistaken. Additionally, how does this analysis account for the European drought of 2003?
- In this paper, the authors present many figures but the explanation about the findings is limited. For example, Figures 1 and 2 are described in a single paragraph, whereas a more detailed discussion of the findings would be beneficial. In Figure 1, why is the scatter in (a)-(c) greater than in (b)-(c)? Similarly, in Figure 4, why is the relationship in (a)-(b) more linear than in the other panels? Additionally, I believe it would be more useful to provide the average findings from both the mHM and SoilClim models.
Line by line comments
L refers to line and P refers to page.
P2L41: Maybe write examples of traditional meteorological drought indices, e.g., SPI and SPEI?
P3L87: The evapotranspiration method should be mentioned.
P4L100-101: Please write what is D2 and what is S2 for readers who are not working with the US drought monitor and Czech drought monitor, respectively.
P4L115: What do the authors mean with the total sum of areal extent of 2nd percentile drought? Why not 10th percentile as well?
P5L150: Maybe replace the word “are variable” with “vary”?
P6: Figure 1. Maybe make the legend (colored circles) bigger? Also I prefer to label all figures with letter a, b, c, and so on.
P7: Figure 2. What are upper and lower quartile? Are they 75th and 25th percentiles?
P8: Please use comma to write number million in all tables. For example: 6,701,638.
P12L227-228: Here the authors mention about centroid movements. However, I could not see these values in the Table 2. These values should be written in the Table or somewhere.
P17L295: The authors may move the word “(Fig. 11)” to the end of sentence.
P24L380: Maybe provide references about the uncertainties of SoilClim model.
P24L397: The authors could explain the potential alternatives for drought clustering as proposed by the cited literature.
P25L402-403: Here the authors mention drought event in North America, which appeared around the center of the continent. However, I could not see any Figure showing the centroid of the drought events. Why don’t the authors provide this figure in the appendix or supplementary material?
P26: Table 5. Maybe in this table the authors can provide the average results of SoilClim and mHM.
Citation: https://doi.org/10.5194/egusphere-2024-1434-RC1 - AC1: 'Reply on RC1', Jan Řehoř, 23 Sep 2024
-
RC2: 'Comment on egusphere-2024-1434', Louise Mimeau, 17 Jul 2024
The identification and classification of droughts events and the consideration of the spatial dynamics of events is both a novelty and an interesting contribution to the study of droughts at the global scale. I find it interesting that the authors have taken into account two different hydrological models to analyze the model uncertainty in their method. The paper is well written and can be suited for publication in HESS after a few major remarks are taken into account, mainly concerning a more in-depth discussion of the results.
General Comments
I agree with the general comments of RC1 and I would like to add the following points in addition to his:
1. The paper would gain in clarity by regrouping parts 4.1 and 4.2 (the results obtained with SoilClim and mHM could be grouped together in the same figures and analyzed at the same time), and by discussing in greater depth the identifiaction of drought events with the OPTICS method (cf general comment n°2 from RC1). I also noticed in the supplementary materials that for each continent there are often distinct events with the same start and end dates (e.g. for the SoilClim model the events ranked 193 and 373 in S. America start at the end of April 1983 and end at the beginning of January 1984, or the events ranked 463, 532, 735 and 736 in N. America which all begin in August 2017 and end in November 2017). Can these events really be considered climatically distinct? What are the reasons for these distinctions between events? I think these questions should be addressed in the discussion section.
2. I have a few reservations about the severity and dynamics indicators, which in my opinion are too interdependent and may have an impact on the classification. I suggest to replace some of these indicators with new ones (see line by line comments below) or at least to discuss the implications of selecting these indicators further in the discussion.
3. Presenting some of the results in map form would help showing the severity and dynamics of the drought events. In particular, it would be interesting to show maps with the maximum spatial extent and the trajectory of the centroids for some of the most extreme events or for the events mentionned in the discussion section (e.g. L395, L402, L463).
Line by line comments
L refers to line and P refers to page.
P3L74 : Please provide references for LAI, landuse and terrain inputs used for the modelling with SoilClim model.
P3L86 : Why are the SoilClim and mHM models forced with two different datasets for the daily meteorological inputs (ERA5-Land and ERA5) ? Using the same inputs would make it easier to compare results obtained with SoilClim and mHM.
P4L105 : Please provide the parameter values for the OPTICS clustering.
P4L106 : How many clusters were removed from this filtering ? This information would be useful for analyzing the fraction of identified droughts that are local events versus the fraction of drought events on a broader continental scale.
P5L148 : The relationships between the severity characteristics seem to be due to the fact that some the characteristics are inter-dependant. Characteristics b and d are directly related to c : the longer the event, the higher is the total sum of areal extents. b should perhaps be replaced with the averaged areal extent of a drought event during its duration and d with the average fraction of the drought events area with a AWR/SM value under the 2nd-percentile threshold.
Figure 4, 5, 6, 7, 11, 12, 13, 14, 15 : Please use different color scales for dynamic classification to avoid confusion with the severity classification.
P13L244 : Can the authors explain this relationship between categories S and D: why events with a wide spatial and temporal coverage are also the most dynamic? I believe this might due to the method used to calculate the dynamics characteristics, which are dependent on the duration of the event (especially for the indicator a: the longer the event, the greater the sum of the distances) and the spatial extent of the event (an event can be very extensive spatially and relatively static, but because of its spatial extent, a small shift in the centroid can give a large absolute distance). The dynamics characteristics should rather be computed as ratios between distances between the centroids and the width of the spatial exent (or number of grid cells) of the events and averaged per time interval.
Figure 7 and 14 : Please clarify axis labels (e.g. Severity category or Dynamic category)
Figure 15 : A clear separation between the 5 continents (a and b) and categories (c) would make the figure easier to read.
P24L385 : Could changes in land cover or irrigation, which are not taken into account in the modelling, also be sources of uncertainty and have an impact on the classification of drought events?
P24L390 : The authors should discuss in more detail the sensitivity of drought event identification to OPTICS clustering parameters.
P26 Table 5 : To make the table easier to read, please replace A and B in the table respectively with 1980-200 and 2000-2020, and perhaps just show the relative frequencies instead of showing both absolute and relative frequencies.
P26L455 : It should be pointed out in the discussion that this statistical analysis over two 20-year periods is a little short-sighted for identifying trends (especially when some drought events can last several years).
L553, L590, L651, L663, L699 : Some doi or url are missing in the references.
L964 : A line break is missing before Vincente-Serrano et al, 2022
Citation: https://doi.org/10.5194/egusphere-2024-1434-RC2 - AC2: 'Reply on RC2', Jan Řehoř, 23 Sep 2024
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 411 | 94 | 29 | 534 | 18 | 14 | 18 |
- HTML: 411
- PDF: 94
- XML: 29
- Total: 534
- Supplement: 18
- BibTeX: 14
- EndNote: 18
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1