the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 Jun 2025
Review article: Past and future climate–related hazards in Indonesia
Abstract. Indonesia, one of the most populous countries, ranked fifth globally for climate-related hazards and disasters in the past five years. This study aims to assess historical trends, future projections, and potential implications of climate-related hazards in Indonesia. We synthesize literature, analyze historical datasets, and examine CMIP projections to assess the trend of extreme climate events and their potential effects on climate hazards. Results show that current records and studies predominantly highlight floods as the most common hazard in the tropical–humid region, followed by landslides, droughts, extreme weather, and wildfires. Historical evidence indicates an increasing threat of flood by intensifying amplitude and frequency of rainfall extremes by around 25 %, especially in most northern parts of Indonesia (Kalimantan, northern Sumatera, parts of Sulawesi, and Papua), while the drought amplified by 60 % over the country across the dry regions from southern Sumatera, Java to Nusa Tenggara. These trends are projected to persist under future climate scenarios (SSPs or RCPs). We found that urban factors like land subsidence and landuse change, particularly in cities like Jakarta, may worsen flood impacts in the future. Future studies should also examine rainfall-induced landslides and flash floods in other vulnerable areas, such as steep areas. Additionally, drought, often overlooked in this country, requires comprehensive research given its unique slow onset and its agricultural and societal effects. Global teleconnection mechanisms (ENSO, IOD, MJO) have intensified wet and dry hazards in recent decades and must be considered. A more integrated approach, combining cascading process models, impact assessments, early warning systems, and adaptive land-use practices, is essential to enhance resilience against climate hazards in Indonesia.
- Preprint
(2275 KB) - Metadata XML
-
Supplement
(179 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
CC1: 'Comment on egusphere-2025-584', Ali S Akanda, 21 Dec 2025
- This is a large and meticulous review of various climate related natural hazards in Indonesia. Previous review papers on this topic has highlighted the vulnerability of Indonesia in general terms - but this manuscript does a good job of categorizing and mapping them with appropriate climate extremes and geographical significance. On that note, this manuscript attempts to provide a type-by-type reviews of the many climate hazards faced by Indonesia, and as reported in current literature and IPCC publications. Lastly, the geographical representation of the hazards and a list including less discussed hazards in the context of Indonesia (droughts, wildfires, and storm surges) is appreciated.
- The manuscript performed an important task by including the effect of the global and regional teleconnections, ENSO/IOD/MJO, on the climate extremes. Most other reviews do not combine the effect of climate change and teleconnections when discussing extremes. This is an unique contribution of this paper.
- In the introduction section, the Greater Jakarta example should be expanded to include the issue of excessive groundwater withdrawal and land subsidence. This issue, although anthropogenic, combines with climate extremes to exacerbate flooding during the rainy season and water scarcity during the dry season (lower groundwater table due to excessive withdrawal and at the same time, reduced rate of recharge due to land cover conversion).
- The selection methodology for literature review has been explained well in the methods section. It is very important to understand the criteria and methodology employed in a review work with such a vast scope. Although the final tally of selected literature (35) was surprisingly low, the careful selection method starting with a much larger selection affirms that important publications were included.
- In Figure 1, should there be a connection between Extreme Precipitation and Storm Surge – do these two types coincide to cause more aggravated coastal damage?
- Are the Yearly Count or Number of Events reported in Figures 2 and 3 related? Are they supposed to be the same? Figure 2 says “cumulative” while Figure 3 does not, but the shapes of the curves are very similar.
- The opening sentence in Discussion: should it be a thorough “review?”
- The summary figure in the discussion (Figure 8) depicting the clustering phenomena is very helpful to bring the vast review and its contents in a concluding discussion. Similarly, Figures 4 and 6 are appreciated that show the different types of hazards and their geographic distributions and future projections.
Citation: https://doi.org/10.5194/egusphere-2025-584-CC1 -
AC1: 'Reply on CC1', Vempi Satriya Adi Hendrawan, 20 May 2026
CC1: Ali S Akanda
We sincerely thank you for the careful and constructive review of our manuscript. Your comments and encouragement have helped us strengthen the framing of our paper, clarify the scope and limitations of the systematic selection, and tighten the discussion of teleconnections, figures, and cross-cutting synthesis. We have addressed each point in detail below, with corresponding revisions reflected in the manuscript.
Comment 1: This is a large and meticulous review of various climate related natural hazards in Indonesia. Previous review papers on this topic has highlighted the vulnerability of Indonesia in general terms - but this manuscript does a good job of categorizing and mapping them with appropriate climate extremes and geographical significance. On that note, this manuscript attempts to provide a type-by-type reviews of the many climate hazards faced by Indonesia, and as reported in current literature and IPCC publications. Lastly, the geographical representation of the hazards and a list including less discussed hazards in the context of Indonesia (droughts, wildfires, and storm surges) is appreciated.
Response: We sincerely thank Dr. Akanda for this encouraging assessment. We believe that the inclusion of the discussed hazards: droughts, wildfires, and storm surges in this study was a deliberate choice to address the fragmented nature of existing Indonesia-focused hazard literature, which has tended to focus on individual hazard types or specific subregions.
Comment 2: The manuscript performed an important task by including the effect of the global and regional teleconnections, ENSO/IOD/MJO, on the climate extremes. Most other reviews do not combine the effect of climate change and teleconnections when discussing extremes. This is an unique contribution of this paper.
Response: Thank you. We have strengthened this aspect in the revised manuscript by explicitly citing the studies supporting each teleconnection–hazard linkage e.g., ENSO/IOD/MJO. By noting in our revised manuscript that no reviewed paper models the simultaneous multi-hazard ENSO response across all zones, we flagged it as a critical evidence gap.
Comment 3: In the introduction section, the Greater Jakarta example should be expanded to include the issue of excessive groundwater withdrawal and land subsidence. This issue, although anthropogenic, combines with climate extremes to exacerbate flooding during the rainy season and water scarcity during the dry season (lower groundwater table due to excessive withdrawal and at the same time, reduced rate of recharge due to land cover conversion).
Response: We agree. This point was underemphasized in the original manuscript. In the current revision, we have added a brief statement in the Introduction noting that in major coastal cities such as Jakarta, anthropogenic land subsidence driven by groundwater extraction and land cover change compounds climate-driven flood risk, with detailed treatment retained in the results section (Abidin et al., 2011; Aditiya and Ito, 2023; Takagi et al., 2016).
Comment 4: The selection methodology for literature review has been explained well in the methods section. It is very important to understand the criteria and methodology employed in a review work with such a vast scope. Although the final tally of selected literature (35) was surprisingly low, the careful selection method starting with a much larger selection affirms that important publications were included.
Response: We thank Dr. Akanda for this observation. The final selection of 35 papers from an initial 93 reflects the specific scope of our search query, which required studies to involve CMIP-based climate model projections applied to climate-related hazards in Indonesia. This criterion deliberately excluded observational-only studies, purely geophysical hazard studies, and studies not addressing the Indonesian climate context, which reduced the eligible pool substantially.
As noted in the limitations section, this also means that Indonesian-language publications and studies indexed outside Scopus are not represented. We have clarified in the Method Section that the 35 selected papers represent the projection literature specifically, and that additional related studies outside this systematic selection are incorporated in the synthesis where relevant, particularly for hazard types with limited reviewed paper coverage. The surprisingly low final count is itself an important finding, indicating that the evidence base for Indonesia-specific, CMIP-based climate hazard projections is still emerging, which reinforces the study's motivation and the research priorities identified in the Discussion.
Comment 5: In Figure 1, should there be a connection between Extreme Precipitation and Storm Surge – do these two types coincide to cause more aggravated coastal damage?
Response: Thank you for this insightful observation. Storm surges are primarily driven by extreme wind and low pressure from storm systems rather than by extreme precipitation directly, which is why the current Figure 1 does not show a direct connection between extreme precipitation and storm surge.
However, Dr. Akanda correctly identifies that in practice, compound coastal flooding events often involve simultaneous storm surge and extreme rainfall, particularly during tropical cyclone landfalls or monsoon-driven low-pressure systems, where both mechanisms co-occur and amplify total inundation. We have added a note in the paragraph prior to Figure 1 acknowledging this compound interaction: "These triggers do not always act in isolation: hazard types in this study can interact and compound, with storm surge amplified by concurrent extreme precipitation during cyclone landfalls and monsoon low-pressure events…" We have also added a brief reference to this compound mechanism in Section 3.4 in the revised manuscript, where coastal hazards are discussed, consistent with the broader compound hazard framing of the paper.
Comment 6: Are the Yearly Count or Number of Events reported in Figures 2 and 3 related? Are they supposed to be the same? Figure 2 says “cumulative” while Figure 3 does not, but the shapes of the curves are very similar.
Response: We thank Dr. Akanda for identifying this potential source of confusion. In the current revised manuscript, Figure 3 (annual publication count and cumulative trend of the 35 reviewed papers) and Figure 4 (trends in annual number of recorded DIBI disaster events) are entirely different datasets and should not be compared directly. Figure 3 shows the annual count of peer-reviewed papers identified through the systematic literature review, with a cumulative curve showing the total number of papers over time. Figure 4 shows the annual number of climate-related disaster events recorded in the DIBI national database from 2000 to 2023, which involves thousands of events per year rather than the single-digit annual paper counts in Figure 3. The visual similarity of their trend shapes (both showing increasing trends over time) is coincidental and reflects two independent phenomena: the growing body of climate projection literature on Indonesia, and the increasing frequency of recorded disaster events. We have added a clarifying statement in the text when Figure 3 is introduced to explicitly distinguish it from the DIBI disaster trend analysis presented later in Figure 4, preventing any confusion between these two different datasets.
Comment 7: The opening sentence in Discussion: should it be a thorough “review?”
Response: Agreed. We have changed most parts of the opening paragraph of the discussion. The Discussion opening has now been revised from "comprehensive assessment" to "an integrated multi-evidence assessment," which more accurately reflects the integration of four complementary evidence streams and aligns with the stated objectives and acknowledged limitations.
Comment 8: The summary figure in the discussion (Figure 8) depicting the clustering phenomena is very helpful to bring the vast review and its contents in a concluding discussion. Similarly, Figures 4 and 6 are appreciated that show the different types of hazards and their geographic distributions and future projections.
Response: Thank you very much for your appreciation. In addition, in the revised Section 3.5, we have made the connection between the current version of Figure 10 and Figures 5 and 8 explicit, stating directly that the three delineated hazard zones spatially align with the disaster patterns in these figures, to reinforce the north–south climate hazard divide as the dominant spatial pattern of disaster risk across the archipelago. We have additionally added a Supplementary Figure to clearly show the pattern of the major island average distribution of each hazard.
Citation: https://doi.org/10.5194/egusphere-2025-584-AC1
-
RC1: 'Comment on egusphere-2025-584', Anonymous Referee #1, 27 Jan 2026
The abstract is unclear. It does not specify what the authors actually did: it mentions a search of bibliographic sources and models but does not indicate a precise methodology. Furthermore, a series of well-known issues are listed, but it is not explained how they were identified in this study.
The first part of the introduction discusses the effects of natural disasters, mixing case studies and literature definitions, but it does not clearly connect to the work carried out by the authors. It gives the impression of presenting their general perspective on disasters rather than introducing the study itself.
From lines 110 to 120, there is confusion: the definitions of the phenomena are approximate and inconsistent. Overall, the work is not well organized: the phenomena are listed in succession without providing unambiguous definitions and seem to change from one paragraph to another. It would be better to create a single, clear, and complete list of the studied phenomena, avoiding adding new items in different sections of the text.
At line 136, a temporal trend analysis based on a certain number of events is mentioned, but the number of events is not specified.
In the third paragraph, while discussing the literature review, climatic extremes of the study area are introduced. It is unclear whether these data come from the authors’ own analysis or from literature sources.
Figure 4 is rather picturesque, but it does not seem useful for understanding what the authors actually did: it merely lists disasters that affected their country, without providing new insights.
There is likely a lot of work behind this article, but the authors fail to clearly communicate what they actually did. It is not clear which statements are mere opinions, which results are drawn from the consulted literature, and which are the original findings derived from analyzing the disaster data of their country. In my opinion, the article needs to be completely rewritten to clarify its objectives, methodology, and results.
I apologize for the critical nature of this feedback; I hope it can help enhance the clarity of the article and that it may be resubmitted in a clearer and more suitable form.
Citation: https://doi.org/10.5194/egusphere-2025-584-RC1 -
AC2: 'Reply on RC1', Vempi Satriya Adi Hendrawan, 21 May 2026
RC1: Anonymous Referee #1
We thank Referee #1 for the careful review of the manuscript, and we apologise for the time it has taken to send this response. We hope the additional time for this revision was worth it.
The reviewer pointed out that the previous version was not clear about which parts were our own analysis, which came from the literature, and how everything fitted together. We have revised the manuscript to fix this. Point-by-point responses follow below, with line numbers and section references to the revised manuscript. Note that all line numbers refer to the clean version of the revised manuscript (without track changes).
---
The abstract is unclear. It does not specify what the authors actually did: it mentions a search of bibliographic sources and models but does not indicate a precise methodology.
Response: The abstract (lines 12–29) has been rewritten to state the methodology explicitly. It now names the four types of evidence used in this study: (1) a PRISMA-guided systematic literature review of 35 peer-reviewed papers; (2) analysis of 47,821 climate-related events from the national disaster database DIBI (2000–2023); (3) historical trend analysis using CRU TS and APHRODITE gridded observations over 1981–2010; and (4) ensemble climate projections from CMIP6 GCMs under SSP5–8.5 and CORDEX Southeast Asia RCMs under RCP8.5, evaluated for 2081–2100 against the 1981–2010 baseline. K-means clustering applied to projected CDD and RX5day changes to delineate three regional hazard zones is also stated. We hope the reader can now see exactly what was done and on what data.
Furthermore, a series of well-known issues are listed, but it is not explained how they were identified in this study.
Response: We thank the reviewer for raising this point. The abstract no longer lists well-known issues. It now reports three specific findings from this study and identifies the evidence source for each: (i) a north–south hazard contrast across the archipelago, evidenced by the national disaster dataset (DIBI) analysis and the historical climate trend analysis (Section 3.2.2); (ii) its projected amplification under SSP5–8.5/RCP8.5, evidenced by our CMIP6 and CORDEX-SEA ensemble analysis; and (iii) persistent evidence gaps for understudied hazards (storm surge, landslides, compound hazards) and underrepresented regions (eastern Indonesia), identified through bibliometric analysis of the 35 reviewed papers cross-referenced against DIBI records. Each claim in the abstract can now be traced to a specific section.
The first part of the introduction discusses the effects of natural disasters, mixing case studies and literature definitions, but it does not clearly connect to the work carried out by the authors.
Response: We appreciate this observation. The reviewer raised the gap between the opening discussion and the rest of the paper. The Introduction has been totally restructured into a clear three-step argument. Lines 30–51 in the current version of the manuscript establish the global context of intensifying climate-related disasters and their disproportionate impact on developing economies. Lines 52–72 focus on Indonesia and quantify its specific exposure (e.g., 5th globally in recorded climate disasters 2019–2023; 6.6 million people affected; 18.75 million severely affected by the 2023 El Niño drought). Lines 73–84 then identify the specific gap addressed by this work, which is the absence of an integrated synthesis that simultaneously links historical disaster records, observed climate trends, and ensemble projections across Indonesia. The objectives in lines 85–91 match the four analytical components in Section 2 (Method).
From lines 110 to 120, there is confusion: the definitions of the phenomena are approximate and inconsistent.
Response: Thank you for flagging this. We acknowledge the inconsistency the reviewer raised. We have revised the parts and the related paragraphs in the Method thoroughly. Specifically, Section 2.1 (lines 97–119) has been majorly rewritten to replace the unclear parts. The hazard framework now follows a single authoritative reference set: the EM-DAT classification (EM-DAT, 2024) and the IRDR Peril Classification (IRDR, 2014), with the climatic impact driver (CID) approach of IPCC (2023) linking hazards to their triggers. Figure 1 shows the trigger–hazard relationships, and Table 1 lists every hazard type examined in this study with its classification (hydrological/meteorological/climatological) and a one-line operational definition according to IRDR.
We also reconcile the DIBI terminology used in our trend analysis with this framework explicitly (lines 150–155): the DIBI categories ”gelombang pasang” (tidal wave) and “abrasi” (abbration) are combined into the storm surge category, and the DIBI “cuaca ekstrem” (weather extreme) category, which mixes strong winds, tornadoes, and heavy rainfall and therefore does not fit into a single Table 1 hazard type, is kept as a separate category in the trend analysis but is explicitly excluded from the literature synthesis in Section 3.4. The reason for this exclusion is now stated clearly in the text.
Overall, the work is not well organized: the phenomena are listed in succession without providing unambiguous definitions and seem to change from one paragraph to another.
Response: This is a fair criticism of the previous version. To address it, we needed a check across and did a substantial revision of the whole manuscript. The manuscript now uses a single hazard classification from Table 1 throughout. Section 2.1 in the current manuscript defines each hazard type once; Section 3.2.1 (DIBI), Section 3.2.2 (historical climate extremes), Section 3.3 (future projections), and Section 3.4 (literature synthesis) all refer back to Table 1 explicitly and follow its ordering. Table 2 in Section 3.4 mirrors the same classification structure, so the reader sees the same hazards in the same order across the disaster records, the climate analysis, and the literature synthesis. Terms such as "flood", "flash flood", "landslide", "storm surge", and "T/R extremes" have been checked across the manuscript and now carry the meanings defined in Table 1 in every section in which they appear.
At line 136, a temporal trend analysis based on a certain number of events is mentioned, but the number of events is not specified.
Response: Thank you for catching this omission. The event count has been added at both places where the reviewer would expect to see it. Section 2.3 (line 155) now states that the analysis covers 47,821 recorded events spanning 2000–2023, and Section 3.2.1 (line 236) repeats the figure when the results are first presented.
In the third paragraph, while discussing the literature review, climatic extremes of the study area are introduced. It is unclear whether these data come from the authors' own analysis or from literature sources.
Response: The reviewer is correct that this distinction was previously unclear. The Results section has been split so that original analysis and literature-derived content are in separate, labelled sections: Section 3.1 reports the profile of the 35 reviewed papers (publication years, hazard coverage, methodology, geographic focus); Section 3.2.1 reports our DIBI analysis; Section 3.2.2 reports our historical climate extremes analysis from CRU TS and APHRODITE; Section 3.3 reports our CMIP6 and CORDEX-SEA projections; and Section 3.4 synthesises the reviewed literature by hazard type, with Table 2 listing every reviewed paper by author, year, scale, scenario, and key finding.
In addition, within the text, the source of each statement is now clearly marked. Our original analysis is introduced in the first person ("Our analysis reveals…", "Our DIBI analysis shows…", "Our ensemble projections indicate…"). Literature-derived findings are cited by author and year and framed as "the reviewed papers" or "reviewed literature". Studies discussed for context but not in the PRISMA-selected 35 are labelled "other related studies". The Discussion (Section 4) now has a dedicated paragraph comparing our results with previous work, so the reader can see at a glance where we confirm, extend, or differ from earlier findings.
The methodological boundary between systematic and supplementary literature is also stated in lines 139–141:
“It should be noted that, in addition to the systematically selected papers, other related studies are included in the synthesis where relevant, particularly for hazard types with limited coverage in the systematic search.”
Figure 4 is rather picturesque, but it does not seem useful for understanding what the authors actually did: it merely lists disasters that affected their country, without providing new insights.
Response: Thank you for this important point on the role of the figures. The spatial event map, which is now Figure 5. We revised the map to make it more seamless and to contain more substantial information, namely events with marker size scaled to fatalities.
The figure is to demonstrate the central spatial argument of the study with several selected major disaster events. We developed the accompanying text in Section 3.2.1 (lines 259–294) now states the analytical conclusions that the figures support: landslides concentrate in Java (76.2% of events and 56.2% of fatalities, with Sumatra contributing a further 20.5% of fatalities); drought impacts fall mostly on the southern dry belt from Java to Nusa Tenggara; wildfires cluster in the peatland-dominated regions of Kalimantan (38.8%) and Sumatra (32.4%); extreme weather events concentrate in Java (61.1%); and floods, while broadly distributed, account for 54.9% of fatalities and 97.0% of displaced persons. These spatial patterns are then clearly linked to the north–south climate divide found in the reviewed literature (Section 3.1) and tested against our own climate projections in Section 3.3, where the K-means hazard zones match the same north–south contrast.
To make the regional reading of the data stronger, Supplementary Figure S1 now reports the full distribution of DIBI events by major island group for every hazard type, so the reader can measure the regional gaps that Figure 5 only shows. The figures, together with the text, are the evidence that the disaster record itself is organised along the same north–south axis that the climate projections amplify.
There is likely a lot of work behind this article, but the authors fail to clearly communicate what they actually did. It is not clear which statements are mere opinions, which results are drawn from the consulted literature, and which are the original findings derived from analyzing the disaster data of their country.
Response: We acknowledge this was the main weakness of the previous version, and it has been the main focus of this revision. The fix has two parts.
First, structural separation. The Results section is now organised so that each analytical component is in its own subsection: Section 3.1 (profile of the 35 reviewed papers), Section 3.2.1 (DIBI national disaster records), Section 3.2.2 (historical climate extremes from CRU TS and APHRODITE; our analysis), Section 3.3 (CMIP6 and CORDEX-SEA projections; our analysis), and Section 3.4 (literature synthesis by hazard type, with Table 2 listing every reviewed paper). Sections 3.2 and 3.3 now contain only original analysis; literature-derived content has been moved to Section 3.4 or the Discussion.
Second, clear attribution in the wording. Within each section, the source of every claim is now clearly marked: original results are introduced in the first person ("Our analysis reveals…", "Our DIBI analysis shows…", "Our ensemble projections indicate…"), while findings from the reviewed literature are attributed by author-year citations and framed as "the reviewed papers", "reviewed literature", or, for context studies outside the PRISMA-selected 35, "other related studies". The Discussion (Section 4) now contains a dedicated comparison paragraph that places our results next to prior work and states explicitly where we confirm, extend, or contrast previous findings, so the reader never has to guess whether a sentence is opinion, literature, or original result.
I apologize for the critical nature of this feedback.
Response: No apology is needed. We really appreciate the comments and feedback the reviewer has given. Thank you very much. The critical reading was exactly what the manuscript needed, and the restructuring of the Results section, the consistent use of definitions across the manuscript, and the clear attribution of every claim to its source are the direct result of this review. We are grateful for the time invested.
Finally, we would like to draw the attention of the editor and the reviewer to the overall scale of this revision. The comments from Referee #1 and Referee #2 pointed to fundamental issues in how the study was presented, and we have responded at a structural level, not just by changing a few sentences. The manuscript has been substantially reorganised: the Results section has been split into clearly separated subsections, the Introduction has been fully restructured, a single hazard classification framework has been introduced and applied consistently across all sections, and the source of every analytical result is now clearly stated.
In practical terms, most of the core sections of the revised manuscript, including the Discussion section, read and function differently from those in the originally submitted version. We believe that the depth and scope of these changes demonstrate how we have taken the reviewer's comments, and we hope that the revised version is a much stronger and clearer paper.
Citation: https://doi.org/10.5194/egusphere-2025-584-AC2
-
AC2: 'Reply on RC1', Vempi Satriya Adi Hendrawan, 21 May 2026
-
RC2: 'Comment on egusphere-2025-584', Anonymous Referee #2, 04 Feb 2026
An interesting paper but it does not live up to expectations. The topic addressed is appropriate, but the structure and content of the work need significant improvement. Please read below some specific comments:
Abstract: it contains a large amount of information but does not follow the traditional abstract structure, stating clearly aims, methods and results. Instead, there is a large description of results but some more details are needed regarding the objectives and methods.
Introduction: it is well-written but is unclear related to where? The first part (lines 25 to 57) do not describe clearly if it is a worldwide phenomenon or refers to effects at the continental level. Moreover, the Indonesia affectation description could be improved to be more clear for the reader.
Method: There are some issues in this section too:
Figure 1 is not a method description (line 98). The literature review ends in 2024 but when does start? It is not explained. Line 107 states that 25 papers were excluded but no further explanation regarding why. Then in line 109 another papers exclusion is indicated but again no reasons are given.
Section 2.2. The official data is compared with the literature review results?
Section 2.4. Lines 166-167 explain that the used datasets "have been widely used in previous studies and have been proved for their performance" but there are not examples of such use.
Results. Section 3.1 include a comment about a trend of publications starting in 2014. Does that mean that this is the missing information of the previous section? If so, why did you start your research in 2014? There were no papers published before about hazards in Indonesia?
Section 3.2. Line 203 refers to figure 3 that illustrates the trend of disaster occurrence from 2000 to 2024. The starting data of 2000 is because the information begins that year or it is a decision of the research team?
Figure 4 includes major disasters affecting the country. How did you decide which ones should be in the map? Moreover, the map is difficult to read as a large amount of information is included there.
Sections 3.3.1 and 3.3.2 should be clearer when refering to historical and future projections, maybe using a) historical and b) projections but as it is written now it looks like a typing error.
Discussion: You wrote that your research reveal significant increases in the frequency and severity of hazards but such increase is related to your data. Are you confident that you have included in your paper all the available information regarding the climate hazards in Indonesia?
In line 503 you indicate that a K-means clustering method is used but it was not explained in your methods section. Even if you write later that details can be found in supplementary files, some previous comments would be appreciated.
References: the list is appropiated and includes significant references to the research topic.
In relation to the above comments, the paper, although it falls within the Journal's scope, needs to be improved as its scientific significance and overall quality does not allow acceptance as it is. I look forward to read an improved version of the manuscript.
Citation: https://doi.org/10.5194/egusphere-2025-584-RC2 -
AC3: 'Reply on RC2', Vempi Satriya Adi Hendrawan, 21 May 2026
RC2: Anonymous Referee #2
We thank Referee #2 for the careful review of the manuscript, and we apologise for the time it has taken to send this response. We hope the additional time for this revision was worth it.
The reviewer's comments covered many important areas, including the abstract structure, the geographic scope of the introduction, the literature search method, the dataset citations, the figure design, the separation of historical and future results, etc. We thank the reviewer for reading the manuscript carefully and for the helpful feedback. We have revised the manuscript to fix this. Point-by-point responses follow below, with line numbers and section references to the revised manuscript (clean version, without track changes).
---
Abstract: it contains a large amount of information but does not follow the traditional abstract structure, stating clearly aims, methods and results.
Response: Thank you for raising this point, as also pointed by Referee #1. The abstract has been fully rewritten to accommodate this substantial concern (lines 12–29). The abstract now follows an aims-methods-results-implications structure. The opening sentence (lines 12–13) now states the gap and study aim plainly: e.g., Indonesia is among the world's most hazard-prone nations, yet no single study brings together observed trends, projections, and disaster records. The methods sentence (lines 14–16) lists the four data sources clearly: a PRISMA-guided systematic review, the DIBI national disaster records, observational gridded climate data (CRU TS and APHRODITE), and ensemble projections from CMIP6 and CORDEX Southeast Asia under SSP5–8.5 and RCP8.5. The results then present the three main findings in order (lines 16–25): the north–south hazard contrast and how it is projected to become stronger, the quantitative archipelago-wide RX5day increase of ~11.5–23.3% and 9.3–11.6 additional CDD in Java by 2081–2100, and the three hazard zones identified using K-means clustering. The closing sentences (lines 26–29) identify the evidence gaps, coastal projections, compound hazard frameworks, eastern Indonesia, and state the implications for adaptation planning. We hope the structure now reads as the reviewer intended.
Introduction: it is well-written but is unclear related to where? The first part (lines 25 to 57) do not describe clearly if it is a worldwide phenomenon or refers to effects at the continental level.
Response: Thank you for this helpful observation. The geographic scope of each opening paragraph is now stated clearly. The first paragraph (lines 31–51) is explicitly global: EM-DAT data, dominating global impacts between 2000 and 2023 (lines 32–34), 4.40 billion people affected and USD 150.6 billion annual economic losses (line 34), and the finding that 90% of disaster fatalities over the past two decades occurred in low- and middle-income countries (lines 36–37). The shift to Indonesia is marked at line 56, "Indonesia exemplifies this intersection at an exceptional scale," so the reader knows exactly where the global framing ends. The Indonesia-specific evidence then sits in its own dedicated paragraph (lines 61–72): the World Risk Index placement (line 58), the 17,000-island archipelago (line 59), the 6.6 million people affected between 2019 and 2023, Indonesia's fifth-place global ranking in climate-related disaster events, and the 18.75 million people affected by the 2023 El Niño drought (lines 66–70). We believe that the shift from the global context to the national case study should now be clear.
Figure 1 is not a method description (line 98). The literature review ends in 2024 but when does start? It is not explained.
Response: Thank you for raising these important points, all three are well taken and have been addressed in the revised manuscript.
Figure 1: it has been repositioned and reframed. Figure 1 now sits inside Section 2.1 (of the current of the manuscript) "Hazard definition and classification" (lines 97–110), explicitly introduced as a classification framework, not a method description: "The primary weather and climate hazard triggers considered in this study… are illustrated in Figure 1" (lines 99–100). The caption (lines 113–116) makes the conceptual role explicit by grouping hazards by class (climatological / meteorological / hydrological).
Line 107 states that 25 papers were excluded but no further explanation regarding why. Then in line 109 another papers exclusion is indicated but again no reasons are given.
Search start date: Section 2.2 (lines 123–124) now states clearly that the search "was limited to English-language articles published up to 5 November 2024," with no lower date restriction applied. The earliest retrieved paper dates to 2013 (line 198), and we state at line 200 that this "reflects the relatively recent emergence of the climate hazard projection studies focused on Indonesia." This is a result of the search, not a design choice.
We now include a new Figure 2 to cover the entire process of the PRISMA systematic review. Exclusion reasons: the PRISMA flow (as seen Figure 2) and the accompanying text now name each exclusion explicitly. The 25 papers removed at screening (lines 132–134) were excluded because they did not address climate-related hazards in Indonesia or did not involve projections or trend analyses. The 33 papers removed at eligibility (lines 134–139) were excluded for four reasons: (i) hazard types outside the study scope, e.g. purely geophysical hazards; (ii) no quantitative hazard-trend or projection analysis; (iii) no specific Indonesian climate context; or (iv) not peer-reviewed. These criteria are now written in the main text rather than only shown in the diagram.
Section 2.2. The official data is compared with the literature review results?
Response: Thank you for this important question. It is important to explain how the two data sources relate to each other. DIBI (Section 2.3 of the current version of the manuscript) and the literature review (Section 2.2) are intentionally not compared statistically. They are treated as two complementary data sources that answer different questions. DIBI provides the observed record of disaster occurrence and impact across Indonesia from 2000–2023 (47,821 events at line 155), while the literature review summarises projected hazard trends from 35 peer-reviewed studies. They are combined qualitatively and spatially in Section 3.5 (lines 587–588): "These three zones spatially align with the hazard patterns identified in the DIBI disaster records (Section 3.2.1) and the reviewed literature (Section 3.1)."
The convergence is then carried into the Discussion (lines 605–607): "the DIBI disaster records and the reviewed literature converge on the same north–south spatial hazard pattern, strengthening confidence in the projected trends." The roles of the three data sources, that is, observed disasters as ground truth, projections as the forward-looking part, and K-means zones as the spatial link between them, are now stated clearly at both ends of the manuscript rather than left for the reader to figure out.
More broadly, we have significantly revised the structure of the manuscript to reflect this distinction at every stage. The Methods section has been reorganised to present each data source with a clearly defined role, and this logic is carried through consistently into the Results and Discussion. The separation between observed records and projected trends is now a structural feature of the paper.
Section 2.4. Lines 166-167 explain that the used datasets "have been widely used in previous studies and have been proved for their performance" but there are not examples of such use.
Response: Thank you for observing this point. We agree that the previous wording was not substantiated. The claim has been removed and replaced with specific dataset citations in Section 2.4 (lines 166–177). CRU TS v4.04 is now cited as Harris et al. (2020) at 0.5° spatial resolution for mean temperature and total rainfall (lines 166–167). APHRODITE is cited as Yatagai et al. (2012), and is now introduced as a "high-resolution precipitation product over the Asian domain" (line 167).
The CMIP6 ensemble (Supplementary Table S1) is referenced through Eyring et al. (2016) at line 171, with the IPCC Interactive Atlas as the source of access (Gutiérrez et al., 2021; Iturbide et al., 2022; lines 172, 176). The CORDEX Southeast Asia ensemble (Supplementary Table S2) is referenced through Tangang et al. (2020) at line 174, which documents its use in Southeast Asian climate studies. We believe that each of these citations therefore serves as a concrete example of previous use in the peer-reviewed literature, directly addressing the reviewer's concern.
Results Section 3.1 include a comment about a trend of publications starting in 2014. Does that mean that this is the missing information of the previous section? If so, why did you start your research in 2014? There were no papers published before about hazards in Indonesia?
Response: Thank you, it is a reasonable concern. The search had no lower date restriction; all years up to 5 November 2024 were eligible (Section 2.2, lines 123–124). As noted above, the earliest paper found dates to 2013 because of the methodological scope of the search query, not because of a date filter. The query (lines 126–130) requires papers to use CMIP-based climate model projections applied to climate-related hazards in Indonesia, which is a relatively new methodological approach. We have made this explicit in the revised Section 3.1 at lines 198–201: "The earliest reviewed papers date to 2013, reflecting the relatively recent emergence of CMIP-based climate hazard projection studies focused on Indonesia." So 2013 is a result of the search, not its starting point.
Section 3.2. Line 203 refers to figure 3 that illustrates the trend of disaster occurrence from 2000 to 2024. The starting date of 2000 is because the information begins that year or it is a decision of the research team?
Response: Thank you for the question. The 2000 starting year reflects both what DIBI makes publicly available and our deliberate choice based on record completeness. We have clarified this in the manuscript at two points. In Section 2.3 (lines 147–148) we now state that the analysis covers "climate-related hazard occurrences across Indonesia from 2000 to 2023," and in Section 3.2.1 at lines 236–237 we explain that this corresponds to "the period of publicly available national records, although records before 2008 may be less accurate due to reporting bias." We also note that the increasing trend in Figure 4 reflects "a combination of genuine increases in hazard frequency… as well as improved disaster reporting over time. This is also consistent with patterns documented in global disaster databases (CRED, 2024)" (lines 250–252). The reporting-bias caveat is also stated in the Figure 4 caption (lines 257–258).
In addition, we have incorporated a pie chart into Figure 4 to show the proportional breakdown of event types, giving the reader an immediate sense of which hazard categories dominate the record. A new Figure 6 has also been added to present additional impact parameters beyond event frequency, covering fatalities, affected people, and economic losses, so that the DIBI analysis now addresses both the occurrence and the human and material consequences of the recorded hazards.
Figure 4 includes major disasters affecting the country. How did you decide which ones should be in the map? Moreover, the map is difficult to read as a large amount of information is included there.
Response: We have addressed the underlying concern on both fronts. The selection criteria for Figure 5 are now stated explicitly in the caption (lines 273–276): events are selected using a "ranked impact criterion (fatalities, missing persons, injured, and houses damaged), with geographic coverage across all major regions of the archipelago." Marker size is scaled to the number of fatalities, and the top 10 ranked events for each hazard type are provided in Supplementary Table S4 (line 275). Regarding readability: the criticism is fair. We have reduced the number of labelled events from the earlier version and made the label font larger, while keeping coverage of Sumatra, Java, Kalimantan, Sulawesi, Nusa Tenggara, and Papua so the north–south and east–west contrasts are still visible. Specific reference events that point the reader to the deadliest recorded cases, Teluk Wondama/Wasior 2010 (170 flood fatalities) and Banjarnegara 2014 (99 landslide fatalities), are now identified directly in the text at lines 262–263.
Sections 3.3.1 and 3.3.2 should be clearer when referring to historical and future projections.
Response: We apologise that the original section structure made this difference harder to follow. Sections 3.2.2 and 3.3 of the current version of the manuscript have been renamed and rewritten to clearly separate the two. Section 3.2.2 is now titled "Historical climate extremes" (line 299) and covers only the 1981–2010 observational period from CRU TS and APHRODITE. Its opening sentence at line 300 sets the scope: "Figure 7 presents the spatial trend of mean temperature and total rainfall per decade across Indonesia over 1981–2010, derived from CRU TS and APHRODITE gridded observations."
Section 3.3 is now titled "Future climate projections" (line 329) and covers only the 2081–2100 projection period from the CMIP6 (SSP5–8.5) and CORDEX SEA (RCP8.5) ensembles. Its opening (line 330) sets the scope: "The ensemble projections indicate that the north–south contrast identified in the historical period is projected to intensify under future warming." The two periods, 1981–2010 observations versus 2081–2100 projections, are now marked at every opening paragraph and within each finding.
Discussion: You wrote that your research reveal significant increases in the frequency and severity of hazards but such increase is related to your data. Are you confident that you have included in your paper all the available information regarding the climate hazards in Indonesia?
Response: Thank you for raising these important issues. We do not claim to have covered all available information on climate hazards in Indonesia. The literature review used Scopus as the sole database with English-language restrictions, and we now state this as an explicit limitation at lines 687–689: "The systematic review was limited to Scopus and English-language publications, so other language studies and grey literature, as well as papers indexed such as in Web of Science or Google Scholar, may contain relevant findings not captured here."
For this reason, we are careful in the revised manuscript to present our conclusions as patterns shown in the DIBI records and the reviewed literature, rather than as final statements about all hazards across the country. We also acknowledge directly in Section 3.2.1 (lines 250–252) that the increasing DIBI trend "reflects a combination of genuine increases in hazard frequency, exposure, and vulnerability… as well as improved disaster reporting over time." The conclusions are therefore based on the evidence we looked at, with the limits of that evidence stated clearly, rather than on a general claim.
We also address this more fully in a dedicated limitations paragraph in the Discussion, which acknowledges four specific constraints: (i) the restriction of the systematic search to Scopus and English-language publications, which likely underrepresents Indonesian-language research; (ii) the K-means zones being based on two climate indices only, so they show hazard potential rather than full risk; (iii) the reviewed literature being concentrated on Java and western Indonesia, which limits conclusions for eastern Indonesia; and (iv) the use of the high-emission scenario as an upper-bound planning perspective rather than the most likely future.
In line 503 you indicate that a K-means clustering method is used but it was not explained in your methods section. Even if you write later that details can be found in supplementary files, some previous comments would be appreciated.
Response: The reviewer is right that the K-means method needed a proper introduction in the main text rather than being left to the supplementary material, and Section 2.5 (lines 184–193) has been expanded accordingly. The revised section now covers four points: (1) the choice of CDD and RX5day as the clustering variables, because they represent the two opposite ends of the rainfall extreme spectrum, that is, long dry periods and intense short-duration rainfall (lines 187–189); (2) the goal of the algorithm, which is to minimise within-cluster variance to group grid cells with similar hazard patterns; (3) the selection of k=3 clusters, which produces the physically meaningful dry, wet, and transitional zones used in Section 3.5 (line 190); and (4) how the resulting cluster boundaries are then compared with the DIBI records and the literature review (lines 192–193).
The Supplementary Method still contains all the technical details, including silhouette curves, sensitivity to initialisation, and the other k values tested, but a reader of the main text now has enough background to follow what was done. In the revised manuscript, the method is now introduced in Section 2.5 (lines 184–193) before any results section refers to it.
References: the list is appropriate and includes significant references to the research topic.
Response: Thank you for the kind assessment.
Finally, we want to highlight the overall scale of this revision. Both referees raised fundamental issues with how the study was presented, and we have tried to fix these at the level of the whole paper, not just in the individual responses. The Introduction has been fully rewritten, a common hazard classification is now used throughout, and every result is clearly linked to its data source. Some of the problems, such as mixing observed data with future projections, and not clearly explaining how the three data sources relate to each other, ran through the whole paper. Fixing them meant changing the Results and Discussion as well. The Results section now has clearly separated parts for past disaster records, historical climate trends, future projections, and spatial zones. The Discussion follows the same order and includes a new paragraph on the study's limitations. We hope the revised paper is clearer and stronger as a result.
Citation: https://doi.org/10.5194/egusphere-2025-584-AC3
-
AC3: 'Reply on RC2', Vempi Satriya Adi Hendrawan, 21 May 2026
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 18,342 | 11,280 | 276 | 29,898 | 530 | 448 | 515 |
- HTML: 18,342
- PDF: 11,280
- XML: 276
- Total: 29,898
- Supplement: 530
- BibTeX: 448
- EndNote: 515
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
