Accelerated Hydrological Wet-to-Dry Transitions and Their Driving Mechanisms over Africa

Gurara, Megersa Adugna; Zhao, Gang; Liu, Xingcai; Xu, Ximeng; Gaffney, Paul P. J.; Tang, Qiuhong

doi:10.5194/egusphere-2026-1969

Preprints

https://doi.org/10.5194/egusphere-2026-1969

Preprints

27 Apr 2026

| 27 Apr 2026

Accelerated Hydrological Wet-to-Dry Transitions and Their Driving Mechanisms over Africa

Megersa Adugna Gurara, Gang Zhao, Xingcai Liu, Ximeng Xu, Paul P. J. Gaffney, and Qiuhong Tang

Abstract. Abrupt wet-to-dry (W2D) events represent a damaging natural disaster, exerting more far-reaching impacts on the environment and society than single extreme events. While sub-seasonal and interannual precipitation whiplash have recently been analyzed, hydrological W2D transitions, especially at smaller time scales such as sub-monthly, have yet to be examined. In this study, we quantify changes in the characteristics of hydrological W2D transitions based on soil moisture in Africa and identify the drivers behind these changes. The results show that the total W2D transition has accelerated markedly: transition speed increased by 19 % and duration shortened by 10 % from 1981 to 2024. The area averaged proportion of rapid W2D events to total W2D events has increased from 52 % during 1981–2000 to 58 % during 2001–2024. The spatial extent of rapid W2D transitions has increased significantly. On average, 13 % of the continent has experienced rapid W2D transition in the 1980s, increasing to 17 % after 2010. These findings suggest a general shift from slow to rapid hydrological W2D transitions on a sub-monthly timescale. We further find that the speeding up of W2D transition onset is driven by greater precipitation deficits, higher temperature, and higher evaporative demand during the transition onset period. Overall, the shift from slow to rapid W2D transitions reduces the predictability of hydrological volatility regimes, which has adverse impacts on agriculture, ecological stability, and water resources management.

Received: 08 Apr 2026 – Discussion started: 27 Apr 2026

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Megersa Adugna Gurara, Gang Zhao, Xingcai Liu, Ximeng Xu, Paul P. J. Gaffney, and Qiuhong Tang

Status: final response (author comments only)

RC1:
'Comment on egusphere-2026-1969', Shuo Wang, 24 Jun 2026
This manuscript investigates the acceleration of wet‑to‑dry (W2D) events across Africa and explores the potential mechanisms driving this trend. Utilizing ERA5‑Land soil moisture data and CHIRPS precipitation data, the authors systematically examine the spatiotemporal patterns and trends of W2D events across the continent. They further compare precipitation, evaporative demand, and temperature between rapid and slow W2D transitions. This topic is of particular relevance given Africa’s socioeconomic vulnerability to hydroclimatic extremes. Nevertheless, prior to publication, the manuscript would benefit from refinements in its conceptual framework, a more in‑depth mechanistic analysis, and clearer presentation of the key findings.
Major comments:
In the Abstract (Lines 15–17) and Introduction (Lines 56–57), the authors identify sub-monthly W2D events as an important research gap. However, the manuscript does not sufficiently justify why the sub-monthly timescale is scientifically or practically important, nor does it clarify how this timescale differs from monthly or longer ones. Before presenting this research gap, the authors should provide a stronger rationale for focusing on sub-monthly W2D transitions. Figure C2c is relevant to this point, but its current classification is too coarse. Rather than using broad categories such as "<30 days" and "30–60 days," the authors should consider subdividing transition durations into finer classes (e.g., 10, 15, or 20 days), as the present grouping does not directly address the sub-monthly issue.

The title emphasizes the driving mechanisms of W2D events, and in the Introduction (Lines 64–70), the authors summarize previous studies on the possible roles of precipitation, temperature, and evaporative demand. However, the results presented in Lines 25–26 and Section 2.3.5 do not extend beyond these three variables. The mechanistic analysis largely consists of comparing mean values of precipitation, evaporative demand, and temperature between rapid and slow W2D transitions. This approach appears too simplistic to support a strong claim about driving mechanisms and does not fully respond to the importance of "both precipitation and terrestrial hydrological indices" noted in Lines 73–74. If the authors wish to retain and emphasize the mechanistic contribution, Sections 2.3.5 and 3.3 should be strengthened. Beyond mean differences, the analysis should incorporate land–atmosphere interactions, variability, and extreme values.

The manuscript adopts the flash drought definition from Yuan et al. (2023) to define sub-monthly W2D events. According to Figure C3 and the sensitivity analysis in Section 2.3.4, the dry threshold is fixed at the 20th percentile, while wet thresholds of the 80th, 85th, and 90th percentiles are tested. In comparison, Yuan et al. (2023) identified sub-seasonal drought events as pentad-mean soil moisture declines from above the 40th percentile to below the 20th percentile, followed by recovery above the 20th percentile. Under the present method, events starting from above the 80th percentile represent a subset of those defined by Yuan et al.; thus, consistent conclusions are expected. Nevertheless, the Introduction and Discussion do not clearly articulate how W2D events differ from flash droughts or why this distinction is particularly relevant for Africa. The authors should clarify the added value of their W2D definition relative to existing flash drought frameworks.

The authors should add a sensitivity analysis concerning transition duration (Lines 137–140). Specifically, the manuscript should explain how many days between wet and dry conditions can still be considered part of the same W2D event. If the transition exceeds 60 days, it remains unclear whether the event can still be regarded as meaningful or impactful.

The authors should incorporate additional soil moisture datasets, such as the GLEAM root-zone soil moisture product. The results in Lines 177–185 could be revised to compare ERA5-Land with GLEAM. More importantly, the manuscript should demonstrate whether the key characteristics of rapid W2D events, including their acceleration and behavior across different timescales, remain consistent across multiple datasets.

Minor comments:
Figure 1 should be revised. The aridity index is not substantially discussed in the main text; the authors should either integrate it more fully into the manuscript or remove it from the figure. Additionally, Lines 155–156 refer to "agricultural and ecological implications," but Figure 1 does not directly support this claim. The figure would be more useful if it displayed the spatial distribution of soil moisture, precipitation, and heavy rainfall, and if it included a schematic definition of rapid and slow W2D events.

The statement in Lines 195–200 is not directly quantified in Figure 2c. These important results should be reorganized and presented more clearly in Figure 2.

For the regions discussed in Lines 221–228, the authors could add regional boundaries or color-coded boxes in Figure 3 to help readers connect the text with the spatial patterns shown. The argument also requires stronger evidence; for example, statements regarding "high climate variability" and "brief and intense rainfall, and rapid soil moisture depletion" should be supported by either additional data analysis or appropriate references.

The definition of rapid and slow W2D events in Section 2.3.3 should be moved before Table 1. Otherwise, readers encounter Table 1 without having been introduced to these definitions, which may cause confusion.

In Line 245, the manuscript introduces a new variable, the "rapid-event ratio." The authors should define this term clearly upon its first appearance and use it consistently thereafter. In addition, several terms and units are not uniformly written; for example, the units in Line 158 are inconsistent with those in Lines 150–151.

The paragraph in Lines 250–263 should be reorganized to improve its logical flow. I suggest presenting the overall pattern first, then discussing seasonal differences, and finally explaining the possible drivers.

A proper citation for the CHIRPS dataset should be added in Line 106.
Citation: https://doi.org/10.5194/egusphere-2026-1969-RC1
- AC1:
  'Reply on RC1', Qiuhong Tang, 01 Aug 2026
  Response to the Referee #1
  To the editor and reviewer,
  We sincerely thank the reviewer for their insightful evaluation of our manuscript and for providing constructive comments and valuable suggestions. We greatly appreciate the time and effort devoted to improving the quality of our work.
  A detailed, point-by-point response explaining is provided below. All reviewer comments are reproduced in black, followed by our responses in blue.
  This manuscript investigates the acceleration of wet‑to‑dry (W2D) events across Africa and explores the potential mechanisms driving this trend. Utilizing ERA5‑Land soil moisture data and CHIRPS precipitation data, the authors systematically examine the spatiotemporal patterns and trends of W2D events across the continent. They further compare precipitation, evaporative demand, and temperature between rapid and slow W2D transitions. This topic is of particular relevance given Africa’s socioeconomic vulnerability to hydroclimatic extremes. Nevertheless, prior to publication, the manuscript would benefit from refinements in its conceptual framework, a more in‑depth mechanistic analysis, and clearer presentation of the key findings.
  Major comments:
  In the Abstract (Lines 15–17) and Introduction (Lines 56–57), the authors identify sub-monthly W2D events as an important research gap. However, the manuscript does not sufficiently justify why the sub-monthly timescale is scientifically or practically important, nor does it clarify how this timescale differs from monthly or longer ones. Before presenting this research gap, the authors should provide a stronger rationale for focusing on sub-monthly W2D transitions. Figure C2c is relevant to this point, but its current classification is too coarse. Rather than using broad categories such as "<30 days" and "30–60 days," the authors should consider subdividing transition durations into finer classes (e.g., 10, 15, or 20 days), as the present grouping does not directly address the sub-monthly issue.
  
  Response
  We sincerely thank the reviewer for this insightful and constructive comment. We fully agree that the scientific and practical importance of focusing on sub-monthly wet-to-dry (W2D) transitions should be more clearly articulated and that the current presentation does not sufficiently distinguish the relevance of sub-monthly transitions from longer timescales.
  In the revised manuscript, we will strengthen both the abstract and the introduction by providing a more comprehensive justification for emphasizing sub-monthly W2D transitions. Specifically, we will explain that sub-monthly transitions represent rapid changes in land surface moisture conditions that can occur within a few weeks, leaving limited time for ecosystems, agriculture, and water resource management to adapt. We will also clarify how sub-monthly transitions differ from monthly or seasonal changes by highlighting their rapid onset and the associated implications for drought development, agricultural stress, and hydrological extremes.
  We also appreciate the reviewer's suggestion regarding Figure C2c and fully agree that the current duration categories are too broad to adequately illustrate the distribution of sub-monthly transition durations. If invited to revise the manuscript, we will update the figure by subdividing the transition durations into finer 15-day intervals, providing a more detailed characterization of the duration distribution and allowing readers to better distinguish transitions occurring within the sub-monthly timescale. We believe that this revised classification will more effectively support the motivation for focusing on sub-monthly W2D events and improve the overall clarity of the manuscript. We appreciate the reviewer's valuable recommendation and believe that these revisions will substantially strengthen the motivation for the study and improve the presentation of the transition duration characteristics.
  The title emphasizes the driving mechanisms of W2D events, and in the Introduction (Lines 64–70), the authors summarize previous studies on the possible roles of precipitation, temperature, and evaporative demand. However, the results presented in Lines 25–26 and Section 2.3.5 do not extend beyond these three variables. The mechanistic analysis largely consists of comparing mean values of precipitation, evaporative demand, and temperature between rapid and slow W2D transitions. This approach appears too simplistic to support a strong claim about driving mechanisms and does not fully respond to the importance of "both precipitation and terrestrial hydrological indices" noted in Lines 73–74. If the authors wish to retain and emphasize the mechanistic contribution, Sections 2.3.5 and 3.3 should be strengthened. Beyond mean differences, the analysis should incorporate land–atmosphere interactions, variability, and extreme values.
  
  Response
  We sincerely thank the reviewer for this thoughtful and constructive comment. We agree that the analyses presented in the current manuscript are primarily diagnostic and reveal statistical associations rather than definitive causal relationships. The intention of this analysis was to identify the potential hydroclimatic factors associated with the acceleration of rapid W2D transitions. Specifically, we aimed to demonstrate that, relative to slow W2D transitions, rapid W2D events are characterized by stronger negative precipitation anomalies and stronger positive anomalies in temperature and atmospheric evaporative demand during the transition period. However, we agree that these relationships should be interpreted as potential influencing factors associated with rapid W2D transitions, rather than definitive causal drivers.
  In the revised manuscript, we will revise the relevant statements throughout the manuscript to avoid implying direct causality based solely on composite anomaly differences. We will clarify that the observed climatic anomalies represent potential contributors that may favor rapid transitions by intensifying atmospheric water deficits, increasing evaporative demand, and accelerating terrestrial drying processes. Furthermore, we will strengthen the mechanistic interpretation by incorporating a more comprehensive assessment of the variability of the key hydroclimatic variables, including precipitation (P), PET, temperature, and climatic water balance, as well as precipitation extreme events. Specifically, we will analyze the temporal variability of these variables by removing the long-term seasonal climatological cycle from the raw monthly time series. The seasonal climatological cycle will be calculated as the long-term monthly mean for each month, and the corresponding monthly anomalies will be obtained by subtracting these climatological means from the original time series. The variability of each climatic factor will then be quantified using the standard deviation of the monthly anomaly series. We will apply the Spearman rank correlation test to examine the associations between the variability of climatic factors and key W2D metrics, including occurrence frequency and transition duration.
  In addition, we will add the discussion of land–atmosphere interactions and terrestrial hydrological processes. Therefore, in the revised manuscript, we will (1) clarify the diagnostic nature and limitations of the current attribution approach, (2) strengthen the analysis of hydroclimatic variability and its relationship with W2D characteristics, (3) incorporate discussion of land–atmosphere interactions and terrestrial hydrological processes, and (4) revise the wording regarding “drivers” to more accurately reflect potential contributing factors rather than the causal mechanisms, which were not addressed by the current analysis.
  The manuscript adopts the flash drought definition from Yuan et al. (2023) to define sub-monthly W2D events. According to Figure C3 and the sensitivity analysis in Section 2.3.4, the dry threshold is fixed at the 20th percentile, while wet thresholds of the 80th, 85th, and 90th percentiles are tested. In comparison, Yuan et al. (2023) identified sub-seasonal drought events as pentad-mean soil moisture declines from above the 40th percentile to below the 20th percentile, followed by recovery above the 20th percentile. Under the present method, events starting from above the 80th percentile represent a subset of those defined by Yuan et al.; thus, consistent conclusions are expected. Nevertheless, the Introduction and Discussion do not clearly articulate how W2D events differ from flash droughts or why this distinction is particularly relevant for Africa. The authors should clarify the added value of their W2D definition relative to existing flash drought frameworks.
  
  Response
  We thank the reviewer for the valuable comment and agree that the distinction between wet-to-dry (W2D) transitions and flash droughts, as well as the rationale for adopting a flash-drought-inspired framework, was not sufficiently articulated in the original manuscript. We will revise the Introduction and Discussion to make these ideas clear.
  Our goal was not to re-define flash drought, but to apply the event-based framework of flash drought to characterize rapid hydrological transitions from anomalously wet to anomalously dry conditions including in sub-monthly timescales. Flash drought definitions are designed by definition to identify rapid moisture loss and rapid hydroclimatic change and so should conceptually align well with W2D transitions. We follow the definition of events based on rapid declines of soil moisture percentile thresholds from Yuan et al. (2023), but we use a stricter wet-state threshold (≥75^th percentile) to specifically isolate transitions arising from distinctly wet antecedent conditions, rather than from near-normal moisture states.
  Many wet-to-dry shifts rely only on the sudden disappearance of rainfall without extreme heat; therefore, they are not flash droughts. Most flash droughts begin under average or mildly dry conditions, lacking the extreme wet precursor that is mandatory for wet-to-dry events, as defined by Yuan et al. (2023) among others. However, we would argue that coincident events involving a rapid wet-to-dry transition under intense heat and evapotranspiration would increase quickly in a changing climate. This would initially lead to waterlogging and crop root rot due to flooding, followed by compound heat-drought damage due to rapid moisture loss. This is a new type of compound hazard for agriculture. The added value of our W2D definition, compared to the existing definition of flash drought, is that we have identified a new type of emerging compound hazard for agriculture and have proposed a definition to quantify it. We will clarify the added value of our W2D definition in the revised paper.
  The authors should add a sensitivity analysis concerning transition duration (Lines 137–140). Specifically, the manuscript should explain how many days between wet and dry conditions can still be considered part of the same W2D event. If the transition exceeds 60 days, it remains unclear whether the event can still be regarded as meaningful or impactful.
  
  Response
  We appreciate the reviewer’s thoughtful comment on the definition and interpretation of W2D transition duration and the suggestion for a sensitivity analysis.
  We partially agree that it is important to clarify the role of transition duration (including those over 60 days) for interpretation. However, in our analysis framework, W2D events are not primarily defined or classified based on the total duration between the end of wet conditions and the onset of dry conditions. Instead, we differentiate events based on the rate of onset of moisture depletion in the W2D transition, the primary diagnostic metric used in this study.
  We specifically distinguish rapid transitions from slow transitions based on the onset speed of dry conditions. If the duration of time between the end of wet conditions and the beginning of dry conditions is longer than 60 days, the corresponding onset speed is usually below our pre-defined threshold. These cases are described as slow transitions, indicating a slow and long drying process, not an abrupt rapid shift transition.
  Importantly, we aim to identify transitions that based on the speed of rapid moisture loss, not the precise time interval between the wet and dry states. In the revised version, we will clarify the difference between transition duration and the classification based on the speed of transition onset. We will also explain how extended transition periods (including those longer than 60 days) are dealt with in the framework and why they are viewed as slow, gradual processes rather than rapid swift transitions.
  The authors should incorporate additional soil moisture datasets, such as the GLEAM root-zone soil moisture product. The results in Lines 177–185 could be revised to compare ERA5-Land with GLEAM. More importantly, the manuscript should demonstrate whether the key characteristics of rapid W2D events, including their acceleration and behavior across different timescales, remain consistent across multiple datasets.
  
  Response:
  We sincerely thank the reviewer for this valuable and constructive suggestion. We fully agree that incorporating an additional, independent soil moisture dataset would strengthen the robustness of the study and provide greater confidence in the identified characteristics of rapid wet-to-dry (W2D) transitions. In the revised manuscript, we will incorporate the GLEAM root-zone soil moisture product as an independent, observation-constrained dataset for comparison with ERA5-Land. We will revise the analysis presented in Lines 177–185 to include a comprehensive comparison of the spatial and temporal characteristics of root-zone soil moisture derived from the two datasets. This comparison will allow us to evaluate the consistency of the observed soil moisture variability across independently developed products.
  In addition, we agree that it is important to assess whether the spatiotemporal characteristics of W2D events are robust across multiple datasets. Accordingly, we will repeat the W2D detection framework using the GLEAM root-zone soil moisture product and compare the resulting transition characteristics with those derived from ERA5-Land. Specifically, we will evaluate the consistency of the spatial distribution, temporal evolution, transition frequency, duration, onset speed, and the observed acceleration of rapid W2D events. We will also discuss any similarities or discrepancies between the datasets and their implications for the interpretation and robustness of our findings. We appreciate the reviewer's insightful recommendation and believe that incorporating GLEAM and performing a multi-dataset comparison will substantially strengthen the reliability, robustness, and generality of the conclusions presented in this study.
  Minor comments:
  Figure 1 should be revised. The aridity index is not substantially discussed in the main text; the authors should either integrate it more fully into the manuscript or remove it from the figure. Additionally, Lines 155–156 refer to "agricultural and ecological implications," but Figure 1 does not directly support this claim. The figure would be more useful if it displayed the spatial distribution of soil moisture, precipitation, and heavy rainfall, and if it included a schematic definition of rapid and slow W2D events.
  
  Response
  We thank the reviewer for this constructive and insightful comment on Fig. 1. We agree that the role of the aridity index was underexplored in the original manuscript text and that its use in Figure 1 required a more explicit justification. Thus, we will update the description and interpretation of the aridity index within the main text, ensuring that its relevance to the study framework. We also acknowledge the suggestion regarding the interpretability and relevance of the spatial information in Figure 1. In the revised version, we will add spatial maps of soil moisture, precipitation and heavy rainfall to better support the discussion on agricultural and ecological implications mentioned earlier in Lines 155–156 as suggested. We will also include in the figure a schematic definition of rapid and slow W2D events in order to improve the conceptual clarity and to better illustrate the event classification adopted in this study. In summary, we believe these changes substantially improve the clarity, completeness, and interpretability of the figure, and better fit the narrative and objectives of the manuscript according to HESS standards.
  The statement in Lines 195–200 is not directly quantified in Figure 2c. These important results should be reorganized and presented more clearly in Figure 2.
  
  Response
  Thank you for this valuable comment. We agree that the percentage changes reported in Lines 195–200 were not explicitly quantified in Figure 2c, which may make it difficult for readers to understand how these values were obtained.
  The reported percentage changes were calculated from the linear trend of each time series over the study period (1981–2024). The trend magnitude was quantified using the non-parametric Sen's slope estimator, and the total change over the study period was calculated as the Sen's slope multiplied by the length of the analysis period. Percentage change was then expressed relative to the mean value over the study period.
  Percentage change (%) = ((Slope × Study period length)/ Mean value) × 100
  Where the slope was estimated from Sen's slope estimator, the study period length was 44 years (1981–2024), and the mean value was calculated from the entire time series. Thus, the reported values represent the relative magnitude of the long-term change over the study period with respect to the climatological mean.
  To improve clarity, we will revise Figure 2c by explicitly annotating the trend magnitude (percentage change and significance level) for each variable. In the revised version, we will also add a detailed description of the calculation method in the Methods section and clarified the corresponding text in the Results section. These revisions allow readers to directly associate the reported percentage changes with the trends shown in Figure 2c.
  For the regions discussed in Lines 221–228, the authors could add regional boundaries or color-coded boxes in Figure 3 to help readers connect the text with the spatial patterns shown. The argument also requires stronger evidence; for example, statements regarding "high climate variability" and "brief and intense rainfall, and rapid soil moisture depletion" should be supported by either additional data analysis or appropriate references.
  
  Response
  Thank you for this valuable suggestion. In the revised version, Figure 3 will be improved by adding clearly delineated regional boundaries (and/or color-coded boxes) corresponding to the regions discussed in the text (Lines 221–228). These additions enable readers to directly associate the spatial patterns shown in the figure with the regional descriptions provided in the manuscript.
  Second, we agree that the discussion required stronger supporting evidence. We will revise the corresponding text by providing additional justification for the statements regarding high climate variability, brief and intense rainfall events, and rapid soil moisture depletion. Where appropriate, we will incorporate the relevant literature to support these interpretations and clarify that these characteristics are consistent with previous studies conducted in the respective regions. In addition, we will revise the discussion to ensure that all regional interpretations are supported either by our analysis or by appropriate references.
  The definition of rapid and slow W2D events in Section 2.3.3 should be moved before Table 1. Otherwise, readers encounter Table 1 without having been introduced to these definitions, which may cause confusion.
  
  Response
  Thank you for this helpful comment. We agree that introducing the definitions of rapid and slow W2D events before Table 1 improves the logical flow and readability of the manuscript. Accordingly, we will move the definitions from Section 2.3.3 to precede Table 1 so that readers are familiar with these terms before encountering the table. This revision will be incorporated into the revised manuscript upon the invitation for revision.
  In Line 245, the manuscript introduces a new variable, the "rapid-event ratio." The authors should define this term clearly upon its first appearance and use it consistently thereafter. In addition, several terms and units are not uniformly written; for example, the units in Line 158 are inconsistent with those in Lines 150–151.
  
  Response
  Thank you for this helpful comment. We agree with the reviewer's suggestion and revise the manuscript accordingly. The term "rapid-event ratio" was not intended to represent a new variable; rather, it was used interchangeably with "rapid ratio." To avoid confusion, we will standardize the terminology throughout the manuscript and consistently use "rapid ratio" thereafter. In addition, we will careful review the manuscript and correct inconsistencies in the presentation of terms and units, including those noted in Lines 150–151 and Line 158, to ensure uniformity and clarity throughout the text.
  The paragraph in Lines 250–263 should be reorganized to improve its logical flow. I suggest presenting the overall pattern first, then discussing seasonal differences, and finally explaining the possible drivers.
  
  Response
  We thank the reviewer for this helpful suggestion. We agree that the logical flow of the paragraph can be improved by first presenting the overall pattern, followed by a discussion of seasonal differences, and then an explanation of the possible drivers. Accordingly, we will reorganize the paragraph in Lines 250–263 to follow this structure, which improves the clarity and coherence of the discussion.
  A proper citation for the CHIRPS dataset should be added in Line 106.
  
  Response
  We appreciate the reviewer’s valuable comment and suggestion. We will add the proper citation for the CHIRPS dataset in Line 106 in the revised manuscript. The corresponding reference will be included to ensure appropriate acknowledgment of the dataset source and its applicability and performance in Africa.
  
  Citation: https://doi.org/10.5194/egusphere-2026-1969-AC1
RC2:
'Comment on egusphere-2026-1969', Jiabo Yin, 17 Jul 2026

This study investigated the spatiotemporal characteristics and drivers of hydrological wet-to-dry (W2D) transitions across Africa using root-zone soil moisture from ERA5-Land. The topic is timely and relevant, given the increasing concerns about hydroclimatic volatility under climate change. The sensitivity tests on threshold choices are appreciated, and the topic is within the scope of HESS. Overall, this manuscript addressed an important and interesting issue int terms of W2D in a data-sparse continent. Before considering acceptance, the following concerns should be well addressed.

(1) The study relies almost exclusively on ERA5-Land (and ERA5 for comparison) soil moisture to identify W2D transitions. While the authors cite previous studies that found these reanalyses acceptable over Africa, this does not substitute for a direct evaluation in the context of this specific application. Reanalysis soil moisture in data-sparse regions is subject to considerable uncertainties from model physics and forcing. The comparison between ERA5-Land and ERA5 merely shows inter-product consistency, not accuracy against observations or independent satellite retrievals. A lack of independent validation undermines the confidence in the reported magnitudes and trends. The authors should either provide a direct validation against in situ observations or independent remote sensing products, or, at a minimum, explicitly and thoroughly discuss how biases and uncertainties in reanalysis soil moisture could affect the identification of wet/dry states and transition characteristics (e.g., frequency, onset speed).

(2) The framework directly adopts Yuan et al. (2023) flash drought identification criteria, altering only the starting condition from a “dry-to-drier” to a “wet-to-dry” transition (i.e., from >75th to <25th percentile). While this is a straightforward extension, the manuscript lacks a clear physical justification for why this particular set of thresholds appropriately captures hydrological wet-to-dry transitions that are distinct from simple drought onset. The current definition essentially identifies a drought event that happened to start from a wet state. The authors should elaborate on how a W2D transition, as defined here, differs from a flash drought or a general drought event, beyond the initial soil moisture condition. What specific hazards or impacts justify this classification?

(3) Although a sensitivity test is presented (Appendix C), the rationale for choosing the default 75th, 25th, and 5%/pentad thresholds is insufficiently motivated. In the main text, the authors should clarify why this configuration is considered the most appropriate for African ecosystems and agricultural contexts.

(4) The attribution of changes in W2D transitionsis limited to a comparison of composite anomalies of precipitation, PET, and temperature between rapid and slow onset periods. This analysis is purely diagnostic and largely correlational. The manuscript claims that “greater precipitation deficits, higher temperature, and higher evaporative demand” are the drivers, but fails to demonstrate a causal link, nor does it account for confounding factors or land-atmosphere feedbacks that are mentioned in the introduction. Statements such as “the main driver for the occurrence of rapid W2D transition events varies spatially” are not convincingly supported by simply mapping anomaly differences. More detailed discussions should be provided.

(5) No significance testing is shown for the spatial anomaly difference maps in Figure 5. At a minimum, stippling or hatching should indicate where the differences between rapid and slow composites are statistically significant.

(6) The time series in Figure 2c shows considerable interannual variability. The smoothed trends appear to be influenced heavily by the endpoints. The robustness of the trends should be tested, for example, by varying the start and end years, and the results should be discussed with more nuance.

(7) The “predictability” statement in the abstract and elsewhere (“shift from slow to rapid W2D transitions reduces the predictability of hydrological volatility regimes”) is an unsubstantiated leap. Reducing duration does not automatically imply reduced predictability; a formal predictability analysis is needed to make this claim.

(8) The Discussion section largely repeats the results and compares them with previous studies on precipitation whiplash. It lacks a critical self-assessment of the study’s limitations. There is no discussion of how the use of pentad data may smooth out higher-frequency variability relevant to “sub-monthly” transitions. The exclusion of hyper-arid regions is sensible, but the sensitivity of results around the margins of these masked areas is not discussed.
(9) The term “sub-monthly” is used to describe the transitions throughout the manuscript. However, the identified rapid W2D events have an area-averaged mean duration of 32.2 days, which is just over a month. The authors should reconcile this description with the reported duration, perhaps by clarifying that the onset phase is sub-monthly while the total event extends longer.
(10) Some sentences should be polished. For example “Anthropogenic activities have naturally complicated…” The word “naturally” is confusing here.

Citation: https://doi.org/10.5194/egusphere-2026-1969-RC2
- AC2:
  'Reply on RC2', Qiuhong Tang, 01 Aug 2026
  Response to the Referee #2
  To the editor and reviewer,
  We sincerely thank the reviewer for their insightful evaluation of our manuscript and for providing constructive comments and valuable suggestions. We greatly appreciate the time and effort devoted to improving the quality of our work.
  A detailed, point-by-point response is provided below. All reviewer comments are reproduced in black, followed by our responses in blue.
  This study investigated the spatiotemporal characteristics and drivers of hydrological wet-to-dry (W2D) transitions across Africa using root-zone soil moisture from ERA5-Land. The topic is timely and relevant, given the increasing concerns about hydroclimatic volatility under climate change. The sensitivity tests on threshold choices are appreciated, and the topic is within the scope of HESS. Overall, this manuscript addressed an important and interesting issue in terms of W2D in a data-sparse continent. Before considering acceptance, the following concerns should be well addressed.
  The study relies almost exclusively on ERA5-Land (and ERA5 for comparison) soil moisture to identify W2D transitions. While the authors cite previous studies that found these reanalyses acceptable over Africa, this does not substitute for a direct evaluation in the context of this specific application. Reanalysis soil moisture in data-sparse regions is subject to considerable uncertainties from model physics and forcing. The comparison between ERA5-Land and ERA5 merely shows inter-product consistency, not accuracy against observations or independent satellite retrievals. A lack of independent validation undermines the confidence in the reported magnitudes and trends. The authors should either provide a direct validation against in situ observations or independent remote sensing products, or, at a minimum, explicitly and thoroughly discuss how biases and uncertainties in reanalysis soil moisture could affect the identification of wet/dry states and transition characteristics (e.g., frequency, onset speed).
  
  Response:
  We sincerely thank the reviewer for this thoughtful and constructive comment. We fully agree that an independent evaluation of the ERA5-Land soil moisture dataset would strengthen the confidence in our analysis and improve the robustness of the study. We also appreciate the reviewer's distinction between inter-product consistency and independent validation.
  In the revised manuscript, we will incorporate two complementary validation analyses. First, we will compare ERA5-Land root-zone soil moisture with the Global Land Evaporation Amsterdam Model (GLEAM) root-zone soil moisture product, an independent, observation-constrained dataset that has been widely used for large-scale soil moisture evaluation. This comparison will enable us to assess the consistency of the spatial and temporal variability of soil moisture between two independently developed datasets across Africa and provide an additional evaluation of the suitability of ERA5-Land for detecting wet-to-dry transitions.
  Second, we will validate ERA5-Land root-zone soil moisture against available in situ observations from the International Soil Moisture Network (ISMN). Although the spatial coverage of ISMN stations in Africa is very limited, these ground-based observations provide an independent benchmark for assessing the temporal behavior of ERA5-Land soil moisture. We will quantify the agreement using appropriate statistical metrics, including the correlation coefficient, bias, and root mean square error (RMSE), where applicable, and discuss the representativeness and limitations of the available station network.
  In addition, we will substantially expand the discussion of uncertainties associated with reanalysis soil moisture. Specifically, we will discuss the potential influences of land surface model structure, meteorological forcing uncertainties, sparse observational constraints used in data assimilation, and the pronounced hydroclimatic heterogeneity across Africa. We will further explain how these sources of uncertainty may influence the identification of wet and dry states and the derived wet-to-dry transition characteristics, including transition frequency, duration, and onset speed. This expanded discussion will provide readers with a clearer understanding of the strengths and limitations of the dataset and the implications for interpreting the study results.
  We appreciate the reviewer's valuable recommendation and believe that incorporating these independent validation analyses and the expanded uncertainty assessment will substantially improve the robustness, transparency, and credibility of the study.
  The framework directly adopts Yuan et al. (2023) flash drought identification criteria, altering only the starting condition from a “dry-to-drier” to a “wet-to-dry” transition (i.e., from >75th to <25th percentile). While this is a straightforward extension, the manuscript lacks a clear physical justification for why this particular set of thresholds appropriately captures hydrological wet-to-dry transitions that are distinct from simple drought onset. The current definition essentially identifies a drought event that happened to start from a wet state. The authors should elaborate on how a W2D transition, as defined here, differs from a flash drought or a general drought event, beyond the initial soil moisture condition. What specific hazards or impacts justify this classification?
  
  Response:
  We sincerely thank the reviewer for this thoughtful and insightful comment. We agree that the physical basis of the proposed wet-to-dry (W2D) transition framework and its distinction from conventional flash drought and general drought events should be more clearly articulated.
  Our objective is not merely to redefine flash drought by modifying the initial soil moisture condition, but rather to characterize a distinct class of hydrological transition events that involve a rapid shift from anomalously wet to anomalously dry soil moisture conditions. Such transitions represent abrupt changes in terrestrial water availability and can occur following periods of excessive rainfall or unusually wet antecedent conditions. These rapid reversals may have important implications for agriculture, water resources, ecosystem functioning, and hazard management because systems that have adapted to wet conditions may subsequently experience rapid moisture depletion over relatively short timescales.
  Unlike conventional flash drought, which typically emphasizes rapid intensification of drought from near-normal or moderately dry antecedent conditions and focuses primarily on drought development, the proposed W2D framework explicitly requires an antecedent wet state before the transition. Consequently, the identified events represent rapid hydrological reversals between two contrasting moisture extremes rather than solely the onset of drought. This distinction allows the framework to quantify the frequency, onset speed, and duration of the transition from W2D hydrological extremes, thereby complementing existing flash drought metrics rather than replacing them.
  It should be noted that many wet-to-dry shifts rely only on the sudden disappearance of rainfall without extreme heat; therefore, they are not flash droughts. Most flash droughts begin under average or mildly dry conditions, lacking the extreme wet precursor that is mandatory for wet-to-dry events, as defined by Yuan et al. (2023) among others. However, we would argue that coincident events involving a rapid wet-to-dry transition under intense heat and evapotranspiration would increase quickly in a changing climate. This would initially lead to waterlogging and crop root rot due to flooding, followed by compound heat-drought damage due to rapid moisture loss. This is a new type of compound hazard for agriculture. The added value of our W2D definition, compared to the existing definition of flash drought, is that we have identified a new type of emerging compound hazard for agriculture and have proposed a definition to quantify it.
  In the revised manuscript, we clarify the differences between a rapid W2D transition, a flash drought, a general drought event, and a normal wet-to-dry transition. We will also include a more comprehensive discussion regarding the different impacts of rapid W2D transitions, flash drought and conventional drought events. We believe these revisions will clarify the novelty and physical relevance of the proposed framework and better demonstrate how rapid W2D transitions provide complementary information beyond existing flash drought and wet-to-dry transition.
  Although a sensitivity test is presented (Appendix C), the rationale for choosing the default 75th, 25th, and 5%/pentad thresholds is insufficiently motivated. In the main text, the authors should clarify why this configuration is considered the most appropriate for African ecosystems and agricultural contexts.
  
  Response:
  We sincerely thank the reviewer for this valuable comment. We agree that the rationale for selecting the default thresholds should be explained more clearly in the main text. Our intention was not to imply that the selected thresholds (75th percentile for wet conditions, 25th percentile for dry conditions, and an onset speed threshold of 5 percentile points per pentad) represent universally optimal values for African ecosystems or agricultural systems. Rather, these thresholds provide a practical and consistent framework for identifying W2D transitions while balancing the detection of sufficiently extreme wet and dry conditions against retaining an adequate sample size for robust statistical analysis across the diverse hydroclimatic regimes of Africa.
  To evaluate the robustness of our methodology and reduce dependence on a single threshold combination, we conducted a comprehensive sensitivity analysis using multiple threshold combinations, including wet thresholds of the 80th, 85th, and 90th percentiles, dry thresholds of the 20th and 25th percentiles, and onset speed thresholds of 5, 8, and 10 percentile points per pentad. The sensitivity analysis assesses how these alternative threshold combinations influence the detected W2D characteristics, particularly the rapid transition ratio and onset speed trends. In the revised manuscript, we will strengthen the main text by explicitly explaining the rationale for adopting the default thresholds and by emphasizing that they should be regarded as a reference configuration rather than uniquely optimal values for African ecosystems. We will also provide a clearer connection between the main methodology and the sensitivity analysis presented in Appendix C, highlighting that the principal conclusions remain robust across a range of reasonable threshold choices. We believe this clarification will improve the transparency of the methodological framework and better justify the selected threshold configuration.
  The attribution of changes in W2D transitions is limited to a comparison of composite anomalies of precipitation, PET, and temperature between rapid and slow onset periods. This analysis is purely diagnostic and largely correlational. The manuscript claims that “greater precipitation deficits, higher temperature, and higher evaporative demand” are the drivers, but fails to demonstrate a causal link, nor does it account for confounding factors or land-atmosphere feedbacks that are mentioned in the introduction. Statements such as “the main driver for the occurrence of rapid W2D transition events varies spatially” are not convincingly supported by simply mapping anomaly differences. More detailed discussions should be provided.
  
  Response:
  We sincerely thank the reviewer for this thoughtful and constructive comment. We agree that the analyses presented in the current manuscript are primarily diagnostic and reveal statistical associations rather than definitive causal relationships. The intention of this analysis was to identify the potential hydroclimatic factors associated with the acceleration of rapid W2D transitions. Specifically, we aimed to demonstrate that, relative to slow W2D transitions, rapid W2D events are characterized by stronger negative precipitation anomalies and stronger positive anomalies in temperature and atmospheric evaporative demand during the transition period. However, we agree that these relationships should be interpreted as potential influencing factors associated with rapid W2D transitions, rather than definitive causal drivers.
  In the revised manuscript, we will revise the relevant statements throughout the manuscript to avoid implying direct causality based solely on composite anomaly differences. We will clarify that the observed climatic anomalies represent potential contributors that may favor rapid transitions by intensifying atmospheric water deficits, increasing evaporative demand, and accelerating terrestrial drying processes. Furthermore, we will strengthen the mechanistic interpretation by incorporating a more comprehensive assessment of the variability of the key hydroclimatic variables, including precipitation, PET, temperature, and climatic water balance, as well as precipitation extreme events. Specifically, we will analyze the temporal variability of these variables by removing the long-term seasonal climatological cycle from the raw monthly time series. The seasonal climatological cycle will be calculated as the long-term monthly mean for each month, and the corresponding monthly anomalies will be obtained by subtracting these climatological means from the original time series. The variability of each climatic factor will then be quantified using the standard deviation of the monthly anomaly series. We will apply the Spearman rank correlation test to examine the associations between the variability of climatic factors and key W2D metrics, including occurrence frequency and transition duration.
  In addition, we will add the discussion of land–atmosphere interactions and terrestrial hydrological processes. Therefore, in the revised manuscript, we will (1) clarify the diagnostic nature and limitations of the current attribution approach, (2) strengthen the analysis of hydroclimatic variability and its relationship with W2D characteristics, (3) incorporate discussion of land–atmosphere interactions and terrestrial hydrological processes, and (4) revise the wording regarding “drivers” to more accurately reflect potential contributing factors rather than the causal mechanisms, which were not addressed by the current analysis.
  No significance testing is shown for the spatial anomaly difference maps in Figure 5. At a minimum, stippling or hatching should indicate where the differences between rapid and slow composites are statistically significant.
  
  Author Response:
  We sincerely thank the reviewer for this valuable suggestion. We agree that statistical significance testing is important for evaluating whether the spatial differences between rapid and slow W2D transition composites represent robust differences rather than random variability.
  In response to this comment, we will perform statistical significance testing for the spatial anomaly difference maps presented in Figure 5. Specifically, we will evaluate the significance of the differences between the composite anomalies of precipitation (P), potential evapotranspiration (PET), and temperature during rapid and slow W2D transition onset stages using an appropriate statistical test. The regions where the differences are statistically significant will be indicated using stippling/hatching in the updated figure.
  The revised Figure 5 will therefore present the spatial distributions of the differences in composite anomalies of (a) precipitation, (b) potential evapotranspiration, and (c) temperature between the onset stages of rapid and slow W2D transitions, with statistical significance information included. We believe that adding significance testing will improve the interpretation of the spatial patterns and provide stronger evidence for the robustness of the hydroclimatic differences associated with rapid versus slow W2D transitions.
  The time series in Figure 2c shows considerable interannual variability. The smoothed trends appear to be influenced heavily by the endpoints. The robustness of the trends should be tested, for example, by varying the start and end years, and the results should be discussed with more nuance.
  
  Response:
  We sincerely thank the reviewer for this insightful comment. We agree that the time series exhibits substantial interannual variability, which demands careful interpretation of the long-term changes.
  However, we would like to clarify that the reported long-term trends in this study were not estimated from the smoothed curves shown in Figure 2c, nor were they derived from the endpoint values of the time series. Instead, the trend magnitude was quantified using the non-parametric Sen's slope estimator, and the total change over the study period was calculated as the Sen's slope multiplied by the length of the analysis period. Percentage change was then expressed relative to the mean value over the study period. Because Sen's slope is estimated from the median of slopes computed from all pairwise observations, it is considerably less sensitive to individual endpoint values and short-term fluctuations than estimates based solely on the beginning and ending years.
  Percentage change was then expressed relative to the mean value over the study period.
  Percentage change (%) = ((Slope × Study period length)/ Mean value) × 100
  Where the slope was estimated from Sen's slope estimator, the study period length was 44 years (1981–2024), and the mean value was calculated from the entire time series.
  To avoid potential misunderstanding, we will revise the manuscript to more clearly describe the trend estimation methodology and explicitly state that the smoothed curves in Figure 2c are intended only to aid visualization of low-frequency variability and are not used for trend estimation or statistical inference.
  In addition, we will expand the discussion to acknowledge the pronounced interannual variability observed in the time series and emphasize that the reported long-term changes represent robust monotonic trends estimated using a non-parametric approach that is less influenced by outliers and endpoint effects. This clarification will provide a more balanced interpretation of the temporal evolution of W2D transition characteristics. We appreciate the reviewer's suggestion, as it will help us better communicate the distinction between visual smoothing and the statistical trend analysis adopted in this study.
  The “predictability” statement in the abstract and elsewhere (“shift from slow to rapid W2D transitions reduces the predictability of hydrological volatility regimes”) is an unsubstantiated leap. Reducing duration does not automatically imply reduced predictability; a formal predictability analysis is needed to make this claim.
  
  Response:
  We sincerely thank the reviewer for this insightful comment. We agree that the original wording could be interpreted as claiming that our study directly quantified changes in the predictability of hydrological volatility, which is beyond the scope of the analyses presented.
  Our intention was not to claim that we performed a formal predictability assessment. Rather, we intended to convey that a shorter transition duration between wet and dry states reduces the available lead time for anticipating and responding to hydrological extremes. In this context, more rapid W2D transitions may provide less opportunity for preparedness and adaptation, thereby posing greater challenges for operational water resources management, agriculture, and ecosystem management.
  To avoid overstating our findings, we will revise the wording throughout the manuscript, including the abstract, to clearly distinguish between reduced lead time for anticipation and predictability in the strict forecasting sense. Specifically, we will replace statements implying reduced predictability with language indicating that the observed shortening of transition duration reduces the time available for preparedness and adaptation, rather than claiming that the intrinsic predictability of hydrological extremes has been quantified or reduced. We appreciate the reviewer's comment, which will help us present our conclusions in a more precise and scientifically rigorous manner while preserving the intended implication of the results.
  The discussion section largely repeats the results and compares them with previous studies on precipitation whiplash. It lacks a critical self-assessment of the study’s limitations. There is no discussion of how the use of pentad data may smooth out higher-frequency variability relevant to “sub-monthly” transitions. The exclusion of hyper-arid regions is sensible, but the sensitivity of results around the margins of these masked areas is not discussed.
  
  Response:
  We sincerely thank the reviewer for this valuable suggestion. We agree that the discussion should provide a more balanced assessment of the study's limitations. In the revised manuscript, we will include a discussion of the methodological limitations. Specifically, we will clarify that pentad soil moisture data were intentionally used to reduce high-frequency daily noise and better capture persistent wet-to-dry (W2D) transitions. However, we will acknowledge that this temporal aggregation may smooth very short-lived variability and influence the detection of the most rapid transitions.
  We will also expand the discussion regarding the exclusion of hyper-arid regions (aridity index < 0.05), explaining that these areas were masked because persistent moisture limitation makes meaningful W2D identification unreliable. In addition, we will acknowledge that greater uncertainty may exist near the boundaries of the masked regions and discuss its potential influence on the results. We appreciate the reviewer's comment, which will improve the quality of the manuscript accordingly.
  The term “sub-monthly” is used to describe the transitions throughout the manuscript. However, the identified rapid W2D events have an area-averaged mean duration of 32.2 days, which is just over a month. The authors should reconcile this description with the reported duration, perhaps by clarifying that the onset phase is sub-monthly while the total event extends longer.
  
  Response:
  We sincerely thank the reviewer for this helpful comment. We agree that the terminology requires clarification to avoid potential confusion. Our intention in using the term "sub-monthly" is to describe the transition phase from the wet state to the dry state, which can occur over a wide range of durations, beginning with very rapid transitions of one pentad (5 days) and extending up to less than 30 days. Thus, the term refers to the duration of the wet-to-dry transition itself rather than to the area-averaged duration of all detected rapid W2D events.
  The reported area-averaged mean duration of 32.2 days represents the spatial average across all rapid W2D events over Africa and reflects the combined influence of substantial regional variability. While some regions exhibit transitions longer than one month, many individual rapid transitions occur within the sub-monthly timescale. Therefore, the term "sub-monthly" is intended to characterize the occurrence of rapid transitions that take place within less than 30 days, rather than to imply that the continental mean duration must also be less than one month.
  To improve clarity, we will revise the manuscript to explicitly state that "sub-monthly" refers to the duration of individual wet-to-dry transition phases (i.e., transitions occurring within less than 30 days), whereas the reported mean duration represents a spatially averaged characteristic across all detected rapid W2D events. This revision will ensure that the terminology is used consistently and is fully aligned with the reported results.
  Some sentences should be polished. For example “Anthropogenic activities have naturally complicated…” The word “naturally” is confusing here.
  
  Response:
  We sincerely thank the reviewer for pointing out this wording issue. We agree that the use of the word "naturally" in this context is inappropriate and may cause confusion. In the revised manuscript, we will revise the sentence by removing the word "naturally" to improve clarity and scientific accuracy. The revised sentence will read:
  "Anthropogenic activities have complicated the hydrological cycle and consequent precipitation variability (Afuecheta and Omar, 2021; Marra et al., 2025)."
  We have also carefully reviewed the manuscript to improve similar wording and enhance the overall readability and clarity.
  
  Citation: https://doi.org/10.5194/egusphere-2026-1969-AC2
- AC3:
  'Reply on RC2', Qiuhong Tang, 01 Aug 2026
  Response to the Referee #2
  To the editor and reviewer,
  We sincerely thank the reviewer for their insightful evaluation of our manuscript and for providing constructive comments and valuable suggestions. We greatly appreciate the time and effort devoted to improving the quality of our work.
  A detailed, point-by-point response is provided below. All reviewer comments are reproduced in black, followed by our responses in blue.
  This study investigated the spatiotemporal characteristics and drivers of hydrological wet-to-dry (W2D) transitions across Africa using root-zone soil moisture from ERA5-Land. The topic is timely and relevant, given the increasing concerns about hydroclimatic volatility under climate change. The sensitivity tests on threshold choices are appreciated, and the topic is within the scope of HESS. Overall, this manuscript addressed an important and interesting issue in terms of W2D in a data-sparse continent. Before considering acceptance, the following concerns should be well addressed.
  The study relies almost exclusively on ERA5-Land (and ERA5 for comparison) soil moisture to identify W2D transitions. While the authors cite previous studies that found these reanalyses acceptable over Africa, this does not substitute for a direct evaluation in the context of this specific application. Reanalysis soil moisture in data-sparse regions is subject to considerable uncertainties from model physics and forcing. The comparison between ERA5-Land and ERA5 merely shows inter-product consistency, not accuracy against observations or independent satellite retrievals. A lack of independent validation undermines the confidence in the reported magnitudes and trends. The authors should either provide a direct validation against in situ observations or independent remote sensing products, or, at a minimum, explicitly and thoroughly discuss how biases and uncertainties in reanalysis soil moisture could affect the identification of wet/dry states and transition characteristics (e.g., frequency, onset speed).
  
  Response:
  We sincerely thank the reviewer for this thoughtful and constructive comment. We fully agree that an independent evaluation of the ERA5-Land soil moisture dataset would strengthen the confidence in our analysis and improve the robustness of the study. We also appreciate the reviewer's distinction between inter-product consistency and independent validation.
  In the revised manuscript, we will incorporate two complementary validation analyses. First, we will compare ERA5-Land root-zone soil moisture with the Global Land Evaporation Amsterdam Model (GLEAM) root-zone soil moisture product, an independent, observation-constrained dataset that has been widely used for large-scale soil moisture evaluation. This comparison will enable us to assess the consistency of the spatial and temporal variability of soil moisture between two independently developed datasets across Africa and provide an additional evaluation of the suitability of ERA5-Land for detecting wet-to-dry transitions.
  Second, we will validate ERA5-Land root-zone soil moisture against available in situ observations from the International Soil Moisture Network (ISMN). Although the spatial coverage of ISMN stations in Africa is very limited, these ground-based observations provide an independent benchmark for assessing the temporal behavior of ERA5-Land soil moisture. We will quantify the agreement using appropriate statistical metrics, including the correlation coefficient, bias, and root mean square error (RMSE), where applicable, and discuss the representativeness and limitations of the available station network.
  In addition, we will substantially expand the discussion of uncertainties associated with reanalysis soil moisture. Specifically, we will discuss the potential influences of land surface model structure, meteorological forcing uncertainties, sparse observational constraints used in data assimilation, and the pronounced hydroclimatic heterogeneity across Africa. We will further explain how these sources of uncertainty may influence the identification of wet and dry states and the derived wet-to-dry transition characteristics, including transition frequency, duration, and onset speed. This expanded discussion will provide readers with a clearer understanding of the strengths and limitations of the dataset and the implications for interpreting the study results.
  We appreciate the reviewer's valuable recommendation and believe that incorporating these independent validation analyses and the expanded uncertainty assessment will substantially improve the robustness, transparency, and credibility of the study.
  The framework directly adopts Yuan et al. (2023) flash drought identification criteria, altering only the starting condition from a “dry-to-drier” to a “wet-to-dry” transition (i.e., from >75th to <25th percentile). While this is a straightforward extension, the manuscript lacks a clear physical justification for why this particular set of thresholds appropriately captures hydrological wet-to-dry transitions that are distinct from simple drought onset. The current definition essentially identifies a drought event that happened to start from a wet state. The authors should elaborate on how a W2D transition, as defined here, differs from a flash drought or a general drought event, beyond the initial soil moisture condition. What specific hazards or impacts justify this classification?
  
  Response:
  We sincerely thank the reviewer for this thoughtful and insightful comment. We agree that the physical basis of the proposed wet-to-dry (W2D) transition framework and its distinction from conventional flash drought and general drought events should be more clearly articulated.
  Our objective is not merely to redefine flash drought by modifying the initial soil moisture condition, but rather to characterize a distinct class of hydrological transition events that involve a rapid shift from anomalously wet to anomalously dry soil moisture conditions. Such transitions represent abrupt changes in terrestrial water availability and can occur following periods of excessive rainfall or unusually wet antecedent conditions. These rapid reversals may have important implications for agriculture, water resources, ecosystem functioning, and hazard management because systems that have adapted to wet conditions may subsequently experience rapid moisture depletion over relatively short timescales.
  Unlike conventional flash drought, which typically emphasizes rapid intensification of drought from near-normal or moderately dry antecedent conditions and focuses primarily on drought development, the proposed W2D framework explicitly requires an antecedent wet state before the transition. Consequently, the identified events represent rapid hydrological reversals between two contrasting moisture extremes rather than solely the onset of drought. This distinction allows the framework to quantify the frequency, onset speed, and duration of the transition from W2D hydrological extremes, thereby complementing existing flash drought metrics rather than replacing them.
  It should be noted that many wet-to-dry shifts rely only on the sudden disappearance of rainfall without extreme heat; therefore, they are not flash droughts. Most flash droughts begin under average or mildly dry conditions, lacking the extreme wet precursor that is mandatory for wet-to-dry events, as defined by Yuan et al. (2023) among others. However, we would argue that coincident events involving a rapid wet-to-dry transition under intense heat and evapotranspiration would increase quickly in a changing climate. This would initially lead to waterlogging and crop root rot due to flooding, followed by compound heat-drought damage due to rapid moisture loss. This is a new type of compound hazard for agriculture. The added value of our W2D definition, compared to the existing definition of flash drought, is that we have identified a new type of emerging compound hazard for agriculture and have proposed a definition to quantify it.
  In the revised manuscript, we clarify the differences between a rapid W2D transition, a flash drought, a general drought event, and a normal wet-to-dry transition. We will also include a more comprehensive discussion regarding the different impacts of rapid W2D transitions, flash drought and conventional drought events. We believe these revisions will clarify the novelty and physical relevance of the proposed framework and better demonstrate how rapid W2D transitions provide complementary information beyond existing flash drought and wet-to-dry transition.
  Although a sensitivity test is presented (Appendix C), the rationale for choosing the default 75th, 25th, and 5%/pentad thresholds is insufficiently motivated. In the main text, the authors should clarify why this configuration is considered the most appropriate for African ecosystems and agricultural contexts.
  
  Response:
  We sincerely thank the reviewer for this valuable comment. We agree that the rationale for selecting the default thresholds should be explained more clearly in the main text. Our intention was not to imply that the selected thresholds (75th percentile for wet conditions, 25th percentile for dry conditions, and an onset speed threshold of 5 percentile points per pentad) represent universally optimal values for African ecosystems or agricultural systems. Rather, these thresholds provide a practical and consistent framework for identifying W2D transitions while balancing the detection of sufficiently extreme wet and dry conditions against retaining an adequate sample size for robust statistical analysis across the diverse hydroclimatic regimes of Africa.
  To evaluate the robustness of our methodology and reduce dependence on a single threshold combination, we conducted a comprehensive sensitivity analysis using multiple threshold combinations, including wet thresholds of the 80th, 85th, and 90th percentiles, dry thresholds of the 20th and 25th percentiles, and onset speed thresholds of 5, 8, and 10 percentile points per pentad. The sensitivity analysis assesses how these alternative threshold combinations influence the detected W2D characteristics, particularly the rapid transition ratio and onset speed trends. In the revised manuscript, we will strengthen the main text by explicitly explaining the rationale for adopting the default thresholds and by emphasizing that they should be regarded as a reference configuration rather than uniquely optimal values for African ecosystems. We will also provide a clearer connection between the main methodology and the sensitivity analysis presented in Appendix C, highlighting that the principal conclusions remain robust across a range of reasonable threshold choices. We believe this clarification will improve the transparency of the methodological framework and better justify the selected threshold configuration.
  The attribution of changes in W2D transitions is limited to a comparison of composite anomalies of precipitation, PET, and temperature between rapid and slow onset periods. This analysis is purely diagnostic and largely correlational. The manuscript claims that “greater precipitation deficits, higher temperature, and higher evaporative demand” are the drivers, but fails to demonstrate a causal link, nor does it account for confounding factors or land-atmosphere feedbacks that are mentioned in the introduction. Statements such as “the main driver for the occurrence of rapid W2D transition events varies spatially” are not convincingly supported by simply mapping anomaly differences. More detailed discussions should be provided.
  
  Response:
  We sincerely thank the reviewer for this thoughtful and constructive comment. We agree that the analyses presented in the current manuscript are primarily diagnostic and reveal statistical associations rather than definitive causal relationships. The intention of this analysis was to identify the potential hydroclimatic factors associated with the acceleration of rapid W2D transitions. Specifically, we aimed to demonstrate that, relative to slow W2D transitions, rapid W2D events are characterized by stronger negative precipitation anomalies and stronger positive anomalies in temperature and atmospheric evaporative demand during the transition period. However, we agree that these relationships should be interpreted as potential influencing factors associated with rapid W2D transitions, rather than definitive causal drivers.
  In the revised manuscript, we will revise the relevant statements throughout the manuscript to avoid implying direct causality based solely on composite anomaly differences. We will clarify that the observed climatic anomalies represent potential contributors that may favor rapid transitions by intensifying atmospheric water deficits, increasing evaporative demand, and accelerating terrestrial drying processes. Furthermore, we will strengthen the mechanistic interpretation by incorporating a more comprehensive assessment of the variability of the key hydroclimatic variables, including precipitation, PET, temperature, and climatic water balance, as well as precipitation extreme events. Specifically, we will analyze the temporal variability of these variables by removing the long-term seasonal climatological cycle from the raw monthly time series. The seasonal climatological cycle will be calculated as the long-term monthly mean for each month, and the corresponding monthly anomalies will be obtained by subtracting these climatological means from the original time series. The variability of each climatic factor will then be quantified using the standard deviation of the monthly anomaly series. We will apply the Spearman rank correlation test to examine the associations between the variability of climatic factors and key W2D metrics, including occurrence frequency and transition duration.
  In addition, we will add the discussion of land–atmosphere interactions and terrestrial hydrological processes. Therefore, in the revised manuscript, we will (1) clarify the diagnostic nature and limitations of the current attribution approach, (2) strengthen the analysis of hydroclimatic variability and its relationship with W2D characteristics, (3) incorporate discussion of land–atmosphere interactions and terrestrial hydrological processes, and (4) revise the wording regarding “drivers” to more accurately reflect potential contributing factors rather than the causal mechanisms, which were not addressed by the current analysis.
  No significance testing is shown for the spatial anomaly difference maps in Figure 5. At a minimum, stippling or hatching should indicate where the differences between rapid and slow composites are statistically significant.
  
  Author Response:
  We sincerely thank the reviewer for this valuable suggestion. We agree that statistical significance testing is important for evaluating whether the spatial differences between rapid and slow W2D transition composites represent robust differences rather than random variability.
  In response to this comment, we will perform statistical significance testing for the spatial anomaly difference maps presented in Figure 5. Specifically, we will evaluate the significance of the differences between the composite anomalies of precipitation (P), potential evapotranspiration (PET), and temperature during rapid and slow W2D transition onset stages using an appropriate statistical test. The regions where the differences are statistically significant will be indicated using stippling/hatching in the updated figure.
  The revised Figure 5 will therefore present the spatial distributions of the differences in composite anomalies of (a) precipitation, (b) potential evapotranspiration, and (c) temperature between the onset stages of rapid and slow W2D transitions, with statistical significance information included. We believe that adding significance testing will improve the interpretation of the spatial patterns and provide stronger evidence for the robustness of the hydroclimatic differences associated with rapid versus slow W2D transitions.
  The time series in Figure 2c shows considerable interannual variability. The smoothed trends appear to be influenced heavily by the endpoints. The robustness of the trends should be tested, for example, by varying the start and end years, and the results should be discussed with more nuance.
  
  Response:
  We sincerely thank the reviewer for this insightful comment. We agree that the time series exhibits substantial interannual variability, which demands careful interpretation of the long-term changes.
  However, we would like to clarify that the reported long-term trends in this study were not estimated from the smoothed curves shown in Figure 2c, nor were they derived from the endpoint values of the time series. Instead, the trend magnitude was quantified using the non-parametric Sen's slope estimator, and the total change over the study period was calculated as the Sen's slope multiplied by the length of the analysis period. Percentage change was then expressed relative to the mean value over the study period. Because Sen's slope is estimated from the median of slopes computed from all pairwise observations, it is considerably less sensitive to individual endpoint values and short-term fluctuations than estimates based solely on the beginning and ending years.
  Percentage change was then expressed relative to the mean value over the study period.
  Percentage change (%) = ((Slope × Study period length)/ Mean value) × 100
  Where the slope was estimated from Sen's slope estimator, the study period length was 44 years (1981–2024), and the mean value was calculated from the entire time series.
  To avoid potential misunderstanding, we will revise the manuscript to more clearly describe the trend estimation methodology and explicitly state that the smoothed curves in Figure 2c are intended only to aid visualization of low-frequency variability and are not used for trend estimation or statistical inference.
  In addition, we will expand the discussion to acknowledge the pronounced interannual variability observed in the time series and emphasize that the reported long-term changes represent robust monotonic trends estimated using a non-parametric approach that is less influenced by outliers and endpoint effects. This clarification will provide a more balanced interpretation of the temporal evolution of W2D transition characteristics. We appreciate the reviewer's suggestion, as it will help us better communicate the distinction between visual smoothing and the statistical trend analysis adopted in this study.
  The “predictability” statement in the abstract and elsewhere (“shift from slow to rapid W2D transitions reduces the predictability of hydrological volatility regimes”) is an unsubstantiated leap. Reducing duration does not automatically imply reduced predictability; a formal predictability analysis is needed to make this claim.
  
  Response:
  We sincerely thank the reviewer for this insightful comment. We agree that the original wording could be interpreted as claiming that our study directly quantified changes in the predictability of hydrological volatility, which is beyond the scope of the analyses presented.
  Our intention was not to claim that we performed a formal predictability assessment. Rather, we intended to convey that a shorter transition duration between wet and dry states reduces the available lead time for anticipating and responding to hydrological extremes. In this context, more rapid W2D transitions may provide less opportunity for preparedness and adaptation, thereby posing greater challenges for operational water resources management, agriculture, and ecosystem management.
  To avoid overstating our findings, we will revise the wording throughout the manuscript, including the abstract, to clearly distinguish between reduced lead time for anticipation and predictability in the strict forecasting sense. Specifically, we will replace statements implying reduced predictability with language indicating that the observed shortening of transition duration reduces the time available for preparedness and adaptation, rather than claiming that the intrinsic predictability of hydrological extremes has been quantified or reduced. We appreciate the reviewer's comment, which will help us present our conclusions in a more precise and scientifically rigorous manner while preserving the intended implication of the results.
  The discussion section largely repeats the results and compares them with previous studies on precipitation whiplash. It lacks a critical self-assessment of the study’s limitations. There is no discussion of how the use of pentad data may smooth out higher-frequency variability relevant to “sub-monthly” transitions. The exclusion of hyper-arid regions is sensible, but the sensitivity of results around the margins of these masked areas is not discussed.
  
  Response:
  We sincerely thank the reviewer for this valuable suggestion. We agree that the discussion should provide a more balanced assessment of the study's limitations. In the revised manuscript, we will include a discussion of the methodological limitations. Specifically, we will clarify that pentad soil moisture data were intentionally used to reduce high-frequency daily noise and better capture persistent wet-to-dry (W2D) transitions. However, we will acknowledge that this temporal aggregation may smooth very short-lived variability and influence the detection of the most rapid transitions.
  We will also expand the discussion regarding the exclusion of hyper-arid regions (aridity index < 0.05), explaining that these areas were masked because persistent moisture limitation makes meaningful W2D identification unreliable. In addition, we will acknowledge that greater uncertainty may exist near the boundaries of the masked regions and discuss its potential influence on the results. We appreciate the reviewer's comment, which will improve the quality of the manuscript accordingly.
  The term “sub-monthly” is used to describe the transitions throughout the manuscript. However, the identified rapid W2D events have an area-averaged mean duration of 32.2 days, which is just over a month. The authors should reconcile this description with the reported duration, perhaps by clarifying that the onset phase is sub-monthly while the total event extends longer.
  
  Response:
  We sincerely thank the reviewer for this helpful comment. We agree that the terminology requires clarification to avoid potential confusion. Our intention in using the term "sub-monthly" is to describe the transition phase from the wet state to the dry state, which can occur over a wide range of durations, beginning with very rapid transitions of one pentad (5 days) and extending up to less than 30 days. Thus, the term refers to the duration of the wet-to-dry transition itself rather than to the area-averaged duration of all detected rapid W2D events.
  The reported area-averaged mean duration of 32.2 days represents the spatial average across all rapid W2D events over Africa and reflects the combined influence of substantial regional variability. While some regions exhibit transitions longer than one month, many individual rapid transitions occur within the sub-monthly timescale. Therefore, the term "sub-monthly" is intended to characterize the occurrence of rapid transitions that take place within less than 30 days, rather than to imply that the continental mean duration must also be less than one month.
  To improve clarity, we will revise the manuscript to explicitly state that "sub-monthly" refers to the duration of individual wet-to-dry transition phases (i.e., transitions occurring within less than 30 days), whereas the reported mean duration represents a spatially averaged characteristic across all detected rapid W2D events. This revision will ensure that the terminology is used consistently and is fully aligned with the reported results.
  Some sentences should be polished. For example “Anthropogenic activities have naturally complicated…” The word “naturally” is confusing here.
  
  Response:
  We sincerely thank the reviewer for pointing out this wording issue. We agree that the use of the word "naturally" in this context is inappropriate and may cause confusion. In the revised manuscript, we will revise the sentence by removing the word "naturally" to improve clarity and scientific accuracy. The revised sentence will read:
  "Anthropogenic activities have complicated the hydrological cycle and consequent precipitation variability (Afuecheta and Omar, 2021; Marra et al., 2025)."
  We have also carefully reviewed the manuscript to improve similar wording and enhance the overall readability and clarity.
  
  Citation: https://doi.org/10.5194/egusphere-2026-1969-AC3

Megersa Adugna Gurara, Gang Zhao, Xingcai Liu, Ximeng Xu, Paul P. J. Gaffney, and Qiuhong Tang

Viewed

Total article views: 374 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
261	86	27	374	19	16

HTML: 261
PDF: 86
XML: 27
Total: 374
BibTeX: 19
EndNote: 16

Views and downloads (calculated since 27 Apr 2026)

Month	HTML	PDF	XML	Total
Apr 2026	78	26	3	107
May 2026	120	30	14	164
Jun 2026	8	7	4	19
Jul 2026	39	18	4	61
Aug 2026	16	5	2	23

Cumulative views and downloads (calculated since 27 Apr 2026)

Month	HTML	PDF	XML	Total
Apr 2026	78	26	3	107
May 2026	120	30	14	164
Jun 2026	8	7	4	19
Jul 2026	39	18	4	61
Aug 2026	16	5	2	23

Viewed (geographical distribution)

Total article views: 364 (including HTML, PDF, and XML) Thereof 364 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 11 Aug 2026

Short summary

Climate change accelerates hydroclimate swings, which significantly impact water supply, agriculture, and ecosystems. We evaluate the spatiotemporal characteristics of hydrological wet-to-dry transitions based on root-zone soil moisture from 1981 to 2024. We then classify these transitions into rapid and slow events based on the onset speed. Wet-to-dry transitions become increasingly rapid and frequent as their duration decreases to a sub-monthly timescale, demanding better planning.


Total:	0
HTML:	0
PDF:	0
XML:	0