the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 24 Jun 2026
Tracking spatiotemporally contiguous heatwave hazards over mainland China: persistence, severity, spatial extent, and mobility
Abstract. Under continued warming, heatwaves increasingly evolve as regional processes that persist, expand, and migrate across space. Using ERA5-Land daily maximum temperature data, we identified spatiotemporally contiguous heatwave events over mainland China during May–September of 1986–2024 and quantified their persistence, severity, affected area, and mobility. The study period was divided into three phases according to changes in annual mean daily maximum temperature: 1986–1998, 1999–2011, and 2012–2024. A total of 609 events were identified, including 177, 219, and 213 events in the three phases, respectively. Although event number did not continue to increase after 2012, event characteristics intensified markedly. During 2012–2024, the fitted rates of change in duration, cumulative severity, affected area, and track length reached 0.264 days yr−1, 2.6 × 106 °C days km2 yr−1, 1.0 × 105 km2 yr−1, and 116.06 km yr−1, respectively. Track length was significantly correlated with duration, severity, and affected area, and these relationships were strongest after 2012. The results indicate that recent heatwave hazards in mainland China are characterized mainly by stronger contiguous event processes rather than by more frequent events.
- Preprint
(8359 KB) - Metadata XML
-
Supplement
(9836 KB) - BibTeX
- EndNote
Status: open (until 09 Aug 2026)
-
RC1: 'Comment on egusphere-2026-3287', Anonymous Referee #1, 03 Jul 2026
reply
-
AC1: 'Reply on RC1', Hao Guo, 12 Jul 2026
reply
Dear Reviewer,
We sincerely thank you for the positive assessment of our manuscript and for the constructive comments. We have carefully considered all suggestions and have prepared corresponding revisions, which will be incorporated into the revised manuscript after the discussion stage. In this public author comment, we provide point-by-point responses and indicate the main changes that will be made.
Reviewer’s overall assessment
This study identifies and tracks three-dimensional spatiotemporally continuous heatwave events across mainland China during 1986–2024 and investigates the evolution of heatwave characteristics across different warming phases. By treating heatwaves as continuously evolving regional processes rather than local extreme events defined at the conventional grid scale or through regional averaging, the study provides a novel perspective that contributes to understanding heatwave evolution in terms of persistence, severity, spatial extent, and migration characteristics. The manuscript is logically organized with a clear narrative structure. The study addresses a scientifically meaningful question and falls well within the scope of the journal and the interests of its readership. I recommend minor revisions prior to acceptance. The main issues relate to the transparency of methodological descriptions, the interpretability of metric definitions, the consistency of terminology, and aspects of figure and table presentation.
Response: We appreciate the reviewer’s encouraging evaluation. The comments are clear and constructive, and they mainly concern methodological transparency, metric definitions, terminology, and figure/table presentation. We agree with these suggestions and will revise the manuscript accordingly, as detailed below.
R1C1:
(1) Lines 120–124: The manuscript defines heatwaves as periods during which daily maximum temperature exceeds the daily 90th percentile threshold for at least three consecutive days. However, the description of the threshold calculation remains relatively brief. It would be helpful to clarify whether the threshold is calculated independently for each grid cell and to provide additional details on the implementation of the 15-day moving window.
Response: Thank you for pointing out this lack of clarity. We will clarify in Section 2.3.1 that the 90th percentile threshold was calculated independently for each grid cell. For each calendar day, the threshold was estimated from all daily maximum temperature values within a 15-day moving window, i.e., ±7 days around the target calendar day, over the full study period of 1986–2024. We will also explain that the use of a calendar-day-specific percentile threshold with a 15-day moving window follows common practice in heatwave identification and helps reduce discontinuities between adjacent calendar days. This clarification will improve the transparency and reproducibility of the heatwave identification procedure.
The relevant Methods text will be revised along the following lines: “At each grid cell, a heatwave event was defined as a period of at least three consecutive days during which the daily maximum temperature exceeded the calendar-day-specific 90th percentile threshold. The threshold was calculated independently for each grid cell. Specifically, for each calendar day, the threshold was estimated from all daily maximum temperature values within a 15-day moving window, i.e., ±7 days around the target calendar day, over the full study period of 1986–2024.”
R1C2:
(2) Section 2.3.1: The three-dimensional continuous event identification method is a core component of this study; however, several key aspects of the methodology remain insufficiently described. For example, heatwave events may split into multiple sub-events during their temporal evolution, in which case event merging may be required to preserve continuity. It is currently unclear whether such situations were considered.
Response: We agree that this methodological detail needs to be stated more explicitly. In our identification procedure, heatwave events are treated as connected spatiotemporal components in a longitude-latitude-time framework. Spatial hot-temperature patches identified on adjacent days are linked when their overlap area exceeds the specified threshold. If one patch on day t is connected to multiple patches on day t+1, or if multiple patches on day t are connected to the same patch on day t+1, all connected patches are assigned to the same event object. Thus, splitting and merging cases are retained as part of the same evolving heatwave event, provided that they satisfy the temporal-adjacency and spatial-overlap criteria. We will add this explanation to Section 2.3.1 to clarify how event continuity is preserved during event splitting and merging.
R1C3:
(3) The definition of heatwave area (HWA) is currently unclear. Based on the current description, it is not entirely clear whether this metric represents the cumulative affected area over the event lifetime or the overall spatial footprint of the event. Further clarification of this definition would improve interpretability and help ensure consistency throughout the manuscript.
Response: This is an important clarification. In this study, HWA refers to the overall spatial footprint of a heatwave event, defined as the total area covered by all unique grid cells affected by the event at least once during its lifetime. It is therefore not the cumulative affected area summed over individual days. We will revise both the textual definition and the equation of HWA in Table 1 to avoid ambiguity. The revised equation will be expressed as HWA=Ai (i∈G_E), where G_E denotes the set of unique grid cells affected by event E during its lifetime and Ai is the area of grid cell i. We will also check all subsequent uses of HWA to ensure that the metric is consistently interpreted as an event footprint.
R1C4:
(4) Section 3.5 compares the events with the largest HWS in each warming phase, which provides useful illustrative examples for demonstrating differences in heatwave characteristics across periods. However, the term “representative events” may be misleading, as it could imply broader generalizability than intended. Consider using terminology that better reflects their illustrative nature, such as “typical events” or “case events”.
Response: We agree that “representative events” may overstate the generality of these examples. Because the three events were selected as the events with the largest HWS in each phase, we will replace “representative events” with “selected high-severity case events” in Section 3.5 and in the relevant figure captions. We will also adjust the wording in this section so that the case-event comparison is presented as illustrative evidence rather than as a basis for broad generalization.
R1C5:
Table 2 currently combines information on duration-class proportions and trend estimates for multiple indicators, resulting in a relatively dense presentation and making horizontal comparison difficult. Consider separating the information into two tables: one presenting the number of events and the proportions of different duration classes across the three phases, and another summarizing the trend estimates for HWF, HWD, HWS, HWA, HWL, and Speed.
Response: We agree that the original Table 2 was too dense and made horizontal comparison difficult. We will split it into two separate tables. The revised Table 2 will report both the number and proportion of events in each duration class across the three warming phases. The revised Table 3 will summarize the trend estimates of HWF, HWD, HWS, HWA, HWL, and Speed. This separation will make the duration composition and the trend estimates easier to compare. In particular, the revised duration-class table will include event counts together with percentages, rather than percentages alone.
R1C6:
(6) Consider reorganizing figure 7 into a 3 × 4 layout to improve clarity and facilitate comparison across panels. In addition, the meaning of the label “ALL”, which appears in several panels, is currently unclear and should be explicitly defined in either the figure caption or the main text.
Response: Thank you for this useful suggestion. We will reorganize Figure 7 into a 3 × 4 layout. In the revised layout, the columns will correspond to ALL, Phase I, Phase II, and Phase III, and the rows will show the relationships of HWL with HWD, HWS, and HWA, respectively. This arrangement will improve readability and facilitate comparison among phases and among relationships. We will also define “ALL” directly in the figure caption as all identified heatwave events during the full study period of 1986–2024. The caption will be revised to avoid ambiguity in the phrase “heatwave events track length”; for example, it will refer to “heatwave track length (HWL)” or “heatwave-event track length (HWL)”.
R1C7:
(7) The manuscript uses multiple related terms interchangeably, including heatwaves, heatwave event, and heat event. This may introduce unnecessary ambiguity regarding the specific unit of analysis being discussed. Consider adopting a more consistent terminology throughout the manuscript (e.g., using “heatwave event” or its abbreviation where appropriate) to improve clarity and maintain consistency.
Response: We agree that the terminology should be made more consistent. We will use “heatwave event(s)” when referring to the identified three-dimensional event objects and event-level metrics, such as HWD, HWS, HWA, HWL, and Speed. The broader term “heatwaves” will be retained only when referring to the general phenomenon or when summarizing previous studies. The term “heat event” will be removed to avoid ambiguity. This revision will make the unit of analysis clearer throughout the manuscript.
R1C8:
(8) Line 156: The meaning of “DD” is unclear. Table 1 defines duration using the abbreviation HWD and already provides an explanation of this metric, making the introduction of an additional notation potentially unnecessary. Consider using consistent notation throughout the manuscript and replacing “DD” with “HWD” in the HWS equation if they refer to the same quantity.
Response: We agree with this comment. In the revised manuscript, we will replace “DD” with “HWD” in the HWS equation and in the explanatory text below Table 1. HWD will be used consistently to denote the duration of a single heatwave event. This change will avoid introducing an unnecessary additional notation for the same quantity.
R1C9:
(9) The term “HW Days” used in Figure 8 and Figure S4 appears to correspond to HWD. To improve consistency and avoid potential confusion, consider using the abbreviation HWD uniformly throughout the manuscript. In addition, the scientific notation labels in the upper-left corners of panels d, f, i, j, and l in Figure S6 appear to overlap with other figure elements, which affects readability. Figure presentation could be further adjusted to improve visual clarity.
Response: Thank you for pointing this out. We agree that the label “HW Days” may cause confusion with the event-level metric HWD. However, the maps in Figure 8 and Figure S4 describe annual heatwave days at the grid-cell scale, whereas HWD denotes the duration of a single identified heatwave event. To avoid conflating these two related but different quantities, we will revise the label “HW Days” to “annual heatwave days” in Figure 8 and Figure S4 and define it explicitly in the figure captions. The abbreviation HWD will be retained only for event-level duration. In addition, we will adjust the layout of Figure S6 to remove the overlap of the scientific-notation labels in panels d, f, i, j, and l, thereby improving readability and visual clarity.
Closing note: We again thank the reviewer for the constructive suggestions. The corresponding revisions will be incorporated into the revised manuscript and response file after the discussion stage.
-
AC1: 'Reply on RC1', Hao Guo, 12 Jul 2026
reply
-
RC2: 'Comment on egusphere-2026-3287', Anonymous Referee #2, 06 Jul 2026
reply
This manuscript investigates heatwaves over mainland China during May–September of 1986–2024 from a three-dimensional, spatiotemporally contiguous event perspective. Using ERA5-Land daily maximum temperature data, the authors identify heatwave events and quantify their persistence, severity, affected area, track length, and moving speed. The study shows that recent heatwave changes, especially after 2012, were characterized less by a continued increase in event numbers than by the strengthening of event-level characteristics. This event-based perspective is valuable because it helps describe heatwaves as evolving regional processes rather than as isolated local temperature anomalies.
Overall, I find the manuscript scientifically meaningful, generally well structured, and relevant to the scope of Natural Hazards and Earth System Sciences. The methodology is appropriate for the research question, and the main conclusions are potentially useful for understanding changes in regional heatwave hazards under warming. I also appreciate the authors’ provision of supporting data and code, which improves the reproducibility and transparency of the work. However, several methodological details, figure/table descriptions, and formatting issues still require clarification or improvement. These issues are mostly minor and should be straightforward to address. I therefore recommend minor revision.
General comments
- Some methodological details should be clarified to improve reproducibility and interpretation. For example, the manuscript should state the reference period used to calculate the calendar-day-specific 90th percentile threshold, explain the rationale for the 1.6% area threshold and the 6400 km² overlap threshold, clarify the exact definition of heatwave area, and define the weighting factor used in centroid calculation. These points are further detailed in Specific Comments 6–8 and 12–13.
- The presentation of the manuscript should be improved before publication. For example, Table 2 is difficult to read in its current layout, some figure captions are too brief for multi-panel figures, and unit formatting is not fully consistent, such as the mixed use of “℃ days km2,” “°C days km²,” “km day−1,” and “km yr−” In addition, there are minor formatting issues such as missing spaces before citations, for example “different events(Ren et al., 2025; ...).” These issues are further noted in Specific Comments 9–10, 14, 17, 19, and 20.
Specific comments
- Lines 11–20. The abstract is clear, but it may be useful to briefly mention that the heatwave definition is based on the calendar-day-specific 90th percentile threshold and a 3-day duration criterion. This would make the abstract more self-contained. For example, the authors may add the following sentence after introducing the ERA5-Land daily maximum temperature data: “Heatwave events were defined using a calendar-day-specific 90th percentile threshold with a minimum duration of three consecutive days and were then tracked as spatiotemporally contiguous three-dimensional events.”
- Lines 23–32. The opening paragraph effectively introduces the broad impacts and increasing risks of heatwaves, including their effects on human health, ecosystems, agriculture, and energy systems. However, these background statements are relatively general. The authors may consider slightly condensing this part and moving more quickly to the key motivation of the study, namely the need to understand heatwaves as spatiotemporally contiguous regional processes. For example, the sentence beginning with “Over recent decades...” could be shortened or linked more directly to the following discussion of heatwave persistence, expansion, and migration.
- Lines 40–46. The limitation of lower-dimensional heatwave analyses is well stated. However, the phrase “such dimensional reduction” could be made more specific. For example, specify whether it refers to grid-cell, station-based, or regional-average analyses.
- Lines 87–90. The study objectives are clearly stated, but the sentence is rather long. Please consider splitting it into two shorter sentences to improve readability.
- Figure 1(a) should include a scale bar where appropriate.
- Lines 106–108. The caption of Figure 1 could be more informative. Please indicate the data period used for the mean daily maximum temperature in panel (c), and clarify whether the inset bar shows monthly means from May to September.
- Lines 110–117. The authors should add a brief note on potential uncertainties or biases in reanalysis temperature from ERA5-Land, especially in complex terrain such as the Qinghai–Tibet Plateau and mountainous regions. A short limitation statement would be sufficient.
- Lines 120–124. Please clarify the reference period used to calculate the calendar-day-specific 90th percentile threshold. Was the threshold calculated using the full 1986–2024 period? This is important for interpreting long-term changes in heatwave characteristics.
- Line 138. There is a formatting issue in “different events(Ren et al., 2025; ...).” Please insert a space before the parenthesis: “different events (Ren et al., 2025; ...).”
- Lines 144–146. The caption of Figure 2 should be revised grammatically. “The 3D structure of a typical heatwave event occurred in June 2024” should be changed to “The 3D structure of a typical heatwave event that occurred in June 2024” or “A typical 3D heatwave event in June 2024.”
- Lines 150–152. The term “Speed” is capitalized differently from the other abbreviations. Please use a consistent style, for example “heatwave moving speed (HWSpeed)” or “moving speed (Speed),” and keep the same form throughout the manuscript and figures.
- Table 1. The definition of HWA needs clarification. It is not fully clear whether HWA represents the maximum daily affected area, the mean daily affected area, the union area over the event lifetime, or a cumulative space–time area. Since HWA is central to the analysis, the definition should be precise.
- Table 1 and lines 153–157. The weighting factor used in the centroid calculation should be defined. Please specify whether the centroid is weighted by temperature exceedance, grid-cell area, or another quantity. This is important because centroid positions are used to calculate track length and moving speed.
- Lines 215–225. Lines 215–225 and Table 2. The manuscript reports positive fitted slopes for several heatwave characteristics across the three phases, while Table 2 marks statistical significance only for selected indicators. To avoid potential overinterpretation, please make the wording slightly more cautious and ensure consistency between the text and Table 2. For example, the sentence may be revised as: “The fitted slopes of HWF, HWD, HWS, HWA, HWL, and Speed were generally positive across the three phases, indicating overall increasing tendencies, while the statistical significance of these fitted trends is reported separately in Table 2.”
- Table 2. The table is currently difficult to read because of its dense layout and rotated/fragmented text. Please improve table formatting or try to split it into two parts.
- Lines 250–251. The caption of Figure 6 is too general. Please specify what the colors, directions, panels, and duration classes represent. Since Figure 6 contains multiple directional and phase-based elements, a more detailed caption would improve readability.
- Lines 279–292. Please standardize the terminology used for the spatial heatwave metric in this section. I suggest reserving “heatwave severity (HWS)” for the event-level indicator defined in Table 1, and using “annual cumulative heatwave temperature” consistently for the gridded spatial metric shown in Figure S5. To avoid ambiguity, please add a short clarification when this term first appears, for example: “Here, annual cumulative heatwave temperature refers to the grid-cell-level annual sum of temperature exceedance during heatwave days, whereas HWS denotes the cumulative severity of an individual three-dimensional heatwave event.”
- Lines 303–316. The description of monthly differences would be clearer if the authors added a brief interpretive sentence after summarizing the May–September patterns. This does not require additional mechanism analysis, but a cautious explanation would help readers understand why the dominant heatwave characteristics differ among months. For example, the authors may consider: “These monthly contrasts may be also related to seasonal changes in the background thermal state and regional circulation conditions during the warm season, which can influence the spatial extent, persistence, and frequency of contiguous heatwave events.”.
- Figure 9. Please ensure that all six heatwave indicators have clearly labeled units.
- Lines 335–341. Please clarify the terminology used for the selected case events. Since these events were selected based on the maximum HWS in each phase, they should not be described in a way that implies they are typical of all events in that phase. I suggest using “the most severe events” or “illustrative high-severity cases” instead of “representative events.”.
- Lines 364–374. The word “stalling” may imply a quantitatively defined stagnation process, but the manuscript appears to use it as a qualitative description of the event trajectory. To avoid overstatement, I suggest replacing “stalling over Qinghai, Sichuan, Shanxi, Hunan, and Ningxia” with a more neutral expression, such as “passing through Qinghai, Sichuan, Shanxi, Hunan, and Ningxia” or “showing prolonged influence over Qinghai, Sichuan, Shanxi, Hunan, and Ningxia.”
- Lines 385–409. Lines 385–409. This paragraph partly repeats the results already presented in Section 3.2, especially the statements that Phase III was characterized by concurrent increases in duration, cumulative severity, affected extent, and track length. I suggest condensing this paragraph by deleting or merging some repetitive sentences. For example, the sentence beginning with “In other words, changes in heatwaves during the rapid warming phase...” can be merged with the preceding sentence. The sentence beginning with “Therefore, the most noteworthy feature of Phase III...” also repeats the same point and could be shortened. In addition, the final sentence beginning with “The results of this study further indicate...” overlaps with the preceding discussion and could be removed or rewritten to emphasize interpretation rather than restating the result.
- Lines 435–439. The limitation paragraph is appropriate. Please consider moving part of this limitation discussion to a broader “limitations and future work” paragraph, including ERA5-Land uncertainty and the lack of physical mechanism diagnosis.
- Lines 445–470. The Conclusions are clear, but they read partly like a shortened repetition of the Results section. I do not suggest removing most numerical information, as some key values are useful for supporting the main take-home messages. However, the authors may consider retaining only the most important numbers that directly support the central conclusions, while reducing detailed numerical repetition. For example, in conclusion (1), it would be sufficient to retain the contrast that Phase III had slightly fewer events than Phase II but a larger contribution from long-duration events, without listing all duration-class proportions. In conclusion (3), the authors may keep one concise statement showing that HWL was more strongly related to HWD, HWS, and HWA than Speed, but the full set of correlation coefficients need not be repeated. In conclusion (4), the month-by-month description could be condensed into a broader take-home message that heatwave intensification showed clear intra-seasonal differences. Overall, I suggest revising the Conclusions to emphasize the conceptual contribution of the contiguous-event perspective while retaining a small number of key quantitative results for support.
Citation: https://doi.org/10.5194/egusphere-2026-3287-RC2 -
RC3: 'RC3', Anonymous Referee #3, 12 Jul 2026
reply
This study identifies heatwaves over mainland China during May–September of 1986–2024 as spatiotemporally contiguous three-dimensional events using ERA5-Land daily maximum temperature, and compares event persistence, cumulative severity, affected area, track length, and moving speed across three warming phases (1986–1998, 1999–2011, 2012–2024). The central finding, that recent heatwave change after 2012 is expressed less through a rise in event counts and more through the joint strengthening of event-level characteristics, is well supported by the analyses and is a useful contribution. The event-based framing is timely, the manuscript is clearly organized, and I appreciate that the event catalogue, gridded products, and analysis scripts are archived, which supports reproducibility.
Two earlier referees have already commented in detail on the transparency of the identification procedure, the definition of HWA, terminology, and table and figure presentation. I agree with those comments and will not repeat them. My review focuses on complementary points concerning the phase division, the interpretation of the correlations among event metrics, the trend estimation within phases, and event assignment near the seasonal and phase boundaries. None of these requires new data, and most can be addressed with clarifications, modest additional analysis, or more careful wording. I therefore recommend minor revisions, with General Comments 1 and 2 deserving the most attention.
S.no.
General Comments
Specific Comments
1
The Pettitt test identifies a single change point, and the manuscript reports 2012 as that turning point (lines 170–172). The basis for the second boundary at 1998/1999 is less clear. The 3 phases are each exactly 13 years long, which gives the impression that the earlier boundary was chosen at least partly for symmetry, and the supporting argument currently rests on the global warming hiatus literature (Li et al., 2019) rather than on a test applied to this dataset. Please state explicitly how the 1998/1999 boundary was determined, for example through a sequential or multiple change-point test and briefly demonstrate that the main conclusions are robust to reasonable shifts of this boundary (for instance, moving it by one to three years). Related to this, the fitted warming trends in Phases I and II are not statistically significant (p > 0.05 in Figure 3), so describing all three periods as distinct warming phases should be worded with care; the essential and well-supported contrast is between the pre-2012 and post-2012 periods.
(Lines 135–136). At 0.1° resolution the grid-cell area varies substantially with latitude across the study domain, from roughly 120 km² in the far south to below 80 km² in the north. Please confirm that Ai in Table 1 and the 1.6% area criterion are computed from true (latitude-dependent) cell areas rather than from cell counts, and state this in the Methods.
2
HWL is defined as the sum of DD − 1 daily centroid displacements, so a longer-lasting event accumulates a longer track by construction, even if its per-day movement is unchanged. Similarly, HWS integrates temperature exceedance over both days and area, so it embeds HWD and HWA in its definition. The strong positive correlations in Figure 7, and their interpretation in Section 4.2 as evidence that track length carries particular process significance, are therefore partly built into the metric definitions rather than purely empirical findings. I do not think this invalidates the analysis, but the manuscript should acknowledge the construction-based component of these relationships explicitly. To show that mobility carries information beyond duration, I suggest a short supplementary analysis, for example the correlation of per-day displacement (HWL/HWD, which is Speed) or of duration-detrended residuals with the other characteristics, or partial correlations between HWL and HWS or HWA controlling for HWD. Interestingly, the weak correlations of Speed with HWS and HWD already reported in Section 3.3 are relevant here: they suggest that much of the HWL signal is indeed carried by duration. A franker treatment of this point would make the argument of Section 4.2 more convincing, and the conclusion (3) statement that track length better captures the integrated evolution of events should be qualified accordingly.
Lines 160–167). Please report the p values of the Phase I and Phase II temperature trends in the text, not only in Figure 3, and temper the wording accordingly (see GC1).
3
Each phase contains only 13 annual values, so fitted slopes are sensitive to single years and endpoints (for example, the very active years 2022 and 2024 fall in Phase III). Please state in the Methods which trend estimator and significance test were used (ordinary least squares, Sen's slope with Mann-Kendall, or other) and justify the choice of p < 0.1 as the significance level in Table 2; marking p < 0.05 as well would help readers judge the strength of the evidence. Please also clarify how the annual values of the event metrics were constructed, that is, whether each annual value is the mean over all events starting in that year, and how many events enter each annual mean, since annual means over roughly 15 events can be noisy. A brief caution about the short window length would be appropriate in the limitations paragraph.
(Lines 253–259 and Figure 7). The text states that the relationships are nonlinear and that the increases in HWD and HWS steepen when HWL exceeds approximately 1×10⁴ km. Please state what functional form was fitted to produce the curves in Figure 7, how the confidence bands were obtained, and how the 1×10⁴ km value was identified (visually or through a formal breakpoint fit). Only a small number of events appear to exceed this value, so the steepening should be described with appropriate caution.
(Lines 269–275). The inverted U-shaped relationship between Speed and the other characteristics is one of the more interesting results and deserves one or two sentences of physical interpretation. A plausible reading is that very fast-moving events pass through a region too quickly to accumulate severity, while near-stationary events accumulate heat locally but produce short tracks. Even a cautious remark along these lines would help readers, and it also connects naturally to Luo et al. (2024) on slower-moving large heatwaves.
(Lines 303–316 and Figure 9). The finding that May shows the highest mean HWS, HWA, HWL, and Speed is somewhat counterintuitive, since July is climatologically the hottest month. Two clarifications would help. First, because a calendar-day-relative threshold is used, May events represent large anomalies relative to a cooler climatology, which should be said explicitly so that readers do not interpret May events as the hottest in an absolute sense. Second, please check whether the May maxima are driven by a small number of very large events (the Phase I case event of May 1995 is itself a 30-day event starting on 5 May); showing medians alongside means in Figure 9, or reporting the event counts per month, would indicate how robust the monthly contrasts are.
Recommendation
This is good research just needs some minor revisions. The dataset, method, and central message are sound, and the requested changes concern justification, interpretation, and presentation rather than the core analysis.
Citation: https://doi.org/10.5194/egusphere-2026-3287-RC3 -
RC4: 'Comment on egusphere-2026-3287', Anonymous Referee #4, 13 Jul 2026
reply
This manuscript examines heatwave events over mainland China during May–September of 1986–2024 using a three-dimensional, spatiotemporally contiguous event framework. The topic is timely and relevant to Natural Hazards and Earth System Sciences. The event-based perspective is useful because it allows heatwaves to be interpreted as evolving regional processes rather than as isolated grid-cell anomalies. The manuscript is clearly organized, the methodology is generally appropriate, and the results provide useful information on changes in heatwave persistence, severity, affected area, and mobility. I recommend minor revision. The following comments may help improve the clarity and readability of the manuscript.
- The introduction explains the value of a three-dimensional event perspective, but the specific novelty of this study relative to previous studies on contiguous heatwaves in China could be stated more directly. Several recent studies have already examined heatwave migration, propagation, and three-dimensional heatwave characteristics. The authors may wish to clarify more explicitly whether the main contribution of this study lies in the phase-based comparison, the joint analysis of persistence, severity, affected area, and track length, or the interpretation of track length as a process-level indicator.
- Some abbreviations and technical terms could be introduced more carefully when they first appear. For example, event-level indicators such as HWD, HWS, HWA, HWL, and Speed are central to the analysis, and their meanings should be easy for readers to follow throughout the manuscript. The authors may consider briefly reminding readers of the meaning of these indicators in key result sections, especially when several metrics are discussed together.
- The discussion of track length is one of the most interesting parts of the manuscript. However, the current interpretation could be made slightly more cautious. A strong correlation between HWL and HWD, HWS, or HWA does not necessarily imply that migration itself causes longer duration or greater severity. The authors may revise the wording to emphasize that track length is statistically associated with these characteristics and can serve as an integrated descriptor of event evolution, rather than implying a direct causal mechanism.
- Some figure captions could be made more self-contained. For multi-panel figures, especially Figures 6, 7, and 10, the captions should briefly indicate what each group of panels represents, how phases are organized, and what the main visual elements mean. This would improve readability, particularly for readers who first inspect the figures before reading the detailed text.
- The conclusions are generally clear, but they could further emphasize the broader implication of the contiguous-event framework. The authors may consider adding one concise sentence explaining how this framework can complement conventional grid-cell or regional-average heatwave analyses, for example by better capturing event persistence, spatial expansion, and migration as connected components of regional heatwave hazards.
- In Figure 6, please clarify what the length of the arrows represents.
- In Figure 10, the color scheme could be made more intuitive. Although the current color assignment is not incorrect, it would be clearer if the colors of the axes, labels, and corresponding legend elements were matched consistently.
- Several important findings should also be highlighted in the Abstract. For example, although the number of contiguous heatwave events does not differ substantially among the three phases, the events in Phase III are characterized by much greater duration, severity and affected area. This is one of the key findings revealed by the three-dimensional event framework and deserves to be summarized in the Abstract.
- Table 1: Please explicitly state that track length is calculated from the movement of the event centroid.
Citation: https://doi.org/10.5194/egusphere-2026-3287-RC4
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 18 | 2 | 0 | 20 | 1 | 0 | 0 |
- HTML: 18
- PDF: 2
- XML: 0
- Total: 20
- Supplement: 1
- BibTeX: 0
- EndNote: 0
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This study identifies and tracks three-dimensional spatiotemporally continuous heatwave events across mainland China during 1986–2024 and investigates the evolution of heatwave characteristics across different warming phases. By treating heatwaves as continuously evolving regional processes rather than local extreme events defined at the conventional grid scale or through regional averaging, the study provides a novel perspective that contributes to understanding heatwave evolution in terms of persistence, severity, spatial extent, and migration characteristics.
The manuscript is logically organized with a clear narrative structure. The study addresses a scientifically meaningful question and falls well within the scope of the journal and the interests of its readership. I recommend minor revisions prior to acceptance. The main issues relate to the transparency of methodological descriptions, the interpretability of metric definitions, the consistency of terminology, and aspects of figure and table presentation.
(1) Lines 120–124: The manuscript defines heatwaves as periods during which daily maximum temperature exceeds the daily 90th percentile threshold for at least three consecutive days. However, the description of the threshold calculation remains relatively brief. It would be helpful to clarify whether the threshold is calculated independently for each grid cell and to provide additional details on the implementation of the 15-day moving window.
(2) Section 2.3.1: The three-dimensional continuous event identification method is a core component of this study; however, several key aspects of the methodology remain insufficiently described. For example, heatwave events may split into multiple sub-events during their temporal evolution, in which case event merging may be required to preserve continuity. It is currently unclear whether such situations were considered.
(3) The definition of heatwave area (HWA) is currently unclear. Based on the current description, it is not entirely clear whether this metric represents the cumulative affected area over the event lifetime or the overall spatial footprint of the event. Further clarification of this definition would improve interpretability and help ensure consistency throughout the manuscript.
(4) Section 3.5 compares the events with the largest HWS in each warming phase, which provides useful illustrative examples for demonstrating differences in heatwave characteristics across periods. However, the term “representative events” may be misleading, as it could imply broader generalizability than intended. Consider using terminology that better reflects their illustrative nature, such as “typical events” or “case events”.
(5) Table 2 currently combines information on duration-class proportions and trend estimates for multiple indicators, resulting in a relatively dense presentation and making horizontal comparison difficult. Consider separating the information into two tables: one presenting the number of events and the proportions of different duration classes across the three phases, and another summarizing the trend estimates for HWF, HWD, HWS, HWA, HWL, and Speed.
(6) Consider reorganizing figure 7 into a 3 × 4 layout to improve clarity and facilitate comparison across panels. In addition, the meaning of the label “ALL”, which appears in several panels, is currently unclear and should be explicitly defined in either the figure caption or the main text.
(7) The manuscript uses multiple related terms interchangeably, including heatwaves, heatwave event, and heat event. This may introduce unnecessary ambiguity regarding the specific unit of analysis being discussed. Consider adopting a more consistent terminology throughout the manuscript (e.g., using “heatwave event” or its abbreviation where appropriate) to improve clarity and maintain consistency.
(8) Line 156: The meaning of “DD” is unclear. Table 1 defines duration using the abbreviation HWD and already provides an explanation of this metric, making the introduction of an additional notation potentially unnecessary. Consider using consistent notation throughout the manuscript and replacing “DD” with “HWD” in the HWS equation if they refer to the same quantity.
(9) The term “HW Days” used in Figure 8 and Figure S4 appears to correspond to HWD. To improve consistency and avoid potential confusion, consider using the abbreviation HWD uniformly throughout the manuscript. In addition, the scientific notation labels in the upper-left corners of panels d, f, i, j, and l in Figure S6 appear to overlap with other figure elements, which affects readability. Figure presentation could be further adjusted to improve visual clarity.