Droughts with no agro-climatological extremes

Abstract

While droughts are primarily linked to climatic conditions, socio-economic factors, such as social vulnerability to drought, significantly influence the impact of drought events. In many vulnerable regions, droughts can occur even when there are minor or no significant deviations from typical agro-climatological conditions. This is often due to socio-economic factors like conflicts, migrations, and economic downturns. However, there has been limited exploration of these “Droughts with No Agro-Climatological Extremes (DNACE)” in terms of their spatio-temporal distribution and analysis. In this study, we aimed to fill this knowledge gap by identifying when, where, and how DNACE events occurred globally. We achieved this by integrating a sub-national geocoded disaster database (GDIS) and a combined drought indicator (CDI). Between 2001 and 2020, we identified 91 DNACE events globally, with the highest concentration in South, Central, and Southeastern Asia (35), followed by South and Eastern Africa (28), and South, Central, and Caribbean America (25). Compared to developed nations, developing countries accounted for 97% of these occurrences, impacting around 36 million people. Socio-economic factors played a significant role in these DNACE events. Political instability and internal conflicts were responsible for 27% of the events in Africa and Eastern Asia. Wars, refugees, and forced migrations contributed to 36% of the cases in South and Eastern Africa, Central Asia, and Eastern Europe. Migration stemming from economic crises represented 32% of events, affecting areas such as Central and Caribbean America, South Asia, and Africa. Additionally, human interventions played a role in 5% of the cases in Eastern Asia. This study underscores that droughts are not solely natural phenomena and emphasizes the critical need to consider socio-economic aspects when formulating drought mitigation and adaptation strategies.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-20617-2.

Introduction

Drought is recognized as one of the costliest and most far-reaching disasters, touching the lives of millions across the globe^1–4. According to the World Health Organization, drought affects approximately 55 million people yearly⁵. Projections for the end of the twenty-first century indicate a rise in drought occurrences, posing complex challenges to the global food supply chain and socio-economic well being^6–8. Given the anticipated severity of forthcoming droughts, comprehending global drought vulnerability and exposure to drought becomes crucial for the successful implementation of disaster preparedness, mitigation, and adaptation strategies.

While climatic^9,10, agricultural^3,11, and hydrological^12,13 factors are primary drivers of drought occurrences, it is the socio-economic aspects that determine how severely communities and populations are affected by drought events. For instance, the surge of refugees from conflict-ridden regions like South Sudan and Somalia has strained water resources in neighboring countries, heightening their vulnerability to water scarcity and drought¹⁴. In Syria¹⁵, Afghanistan¹⁶, and the Central African Republic¹⁷, etc., the prolonged civil wars have disrupted irrigation systems, damaged critical water infrastructure, weakened resource governance, and restricted access to agricultural land, significantly reducing water availability and crop productivity. These conflict-driven disruptions have contributed to drought-like circumstances. Furthermore, extensive deforestation in the Amazon rainforest¹⁸ has disrupted precipitation patterns, exacerbated dry conditions, and increased susceptibility to drought in the region. These examples highlight that droughts are not always natural; they occur even with minor (or no) anomalies of agro-climatological conditions due to socio-economic factors such as conflicts, migrations, and economic downturns. These “Droughts with No Agro-Climatological Extremes (DNACE)” present a unique challenge as they occur without the usual natural extreme conditions, and the onset or progression of the drought may be less predictable, making it difficult to manage and respond.

We define DNACE as drought events that are not associated with significant negative anomalies in conventional climatological (e.g., precipitation, temperature) or agricultural (e.g., vegetation health, soil moisture) indicators. These events primarily result from socio-economic, political, or anthropogenic factors, such as conflict, migration, poor governance, or abrupt land use changes, which can lead to drought-like impacts such as water scarcity, agricultural disruption, or food insecurity, despite the absence of agro-climatological extremes. DNACE events challenge traditional drought monitoring systems, as their onset and severity are governed by non-environmental drivers. This concept expands the conventional scope of drought assessment by recognizing that drought can occur independently of meteorological or agricultural conditions. It highlights the need to incorporate socio-economic dimensions into drought monitoring, early warning, and response frameworks.

Drought research has a long history; many studies^19–21 have examined the linkage between agro-climatological extremes and their propagation and impacts on socio-economic factors. However, the spatio-temporal distribution and the analysis of the DNACE have yet to be explored. We aimed to address this knowledge gap and reveal when, where, and how DNACE events occurred globally during 2001–2020. In this study, the socio-economic impacts of drought events were evaluated from the Geocoded Disaster (GDIS) dataset^1,2, and the agro-climatological drought hazards were identified by the combined drought indicator (CDI)^3,4. The associations and disparities between CDI and GDIS helped to introduce the DNACE events. In our ongoing study²², we validated the superiority of the CDI in correlating with the GDIS, compared to traditional drought indices such as the Standardized Precipitation Index (SPI), Standardized Soil Moisture Index (SSMI), Normalized Difference Vegetation Index (NDVI), and Temperature Condition Index (TCI). Our findings indicate that CDI has a greater potential to detect GDIS events²². Hence, CDI was used to identify agro-climatological droughts in this study.

Initially, we classified all drought occurrences within GDIS into two groups: ones consistent with CDI and ones inconsistent with CDI. To identify the consistency/ inconsistency of GDIS events, we superimposed all the polygons from the GDIS dataset on CDI maps based on their event occurrence period, as shown in Fig. 1.

Fig. 1.

Identification of DNACE Events through Interlinking GDIS and CDI Datasets. (a, d) Long-term timeseries in CDI and Year-wise GDIS Events for Cuba and Burundi respectively. (b, e) Consistent GDIS Events, where GDIS Polygons match Drought Hazards in CDI Maps over Cuba and Burundi. (c, f) Highlights the inconsistencies Between CDI and GDIS which also, signifies DNACE Events.

Figure 1 (a, b, and c) illustrates the association between the GDIS and the CDI over Cuba (Central America). Based on the GDIS dataset, two events occurred in Cuba: one from August 2004 to December 2004 and the other from October 2015 to February 2016. During the first event period, only in August 2004 to December 2004, the CDI trend line exceeded the predefined threshold of -1 (Fig. 1(a), 1(b)), indicating the presence of agro-climatological drought, and thus confirming the consistency between GDIS and CDI for that event. However, during the second event period (October 2015 to February 2016), positive anomalies of CDI were observed over Cuba (Fig. 1(a), 1(c)), showing the inconsistency between GDIS and CDI for that period. Similarly, In Burundi (Africa), three inconsistent GDIS events were examined, suggesting that agro-climatological extremes did not cause those events (Fig. 1d-f). These inconsistent GDIS drought events were recognized as DNACE events.

From GDIS, after eliminating short drought periods ( < = 2 months), we identified 91 DNACE events that were inconsistent with agro-climatological hazards, illustrated by CDI. The inconsistencies observed in the 91 events indicated that they did not occur primarily due to agro-climatological extremes. These discrepancies raised questions about the potential reasons behind these events; hence, contemporary literature and documentary surveys were carried out to understand the factors that might have initiated the inconsistent drought (DNACE) events. As pioneering research dealing with this aspect (DNACE) of drought, our study revolves around three primary key areas of focus: (1) Droughts are not always associated with agro-climatological/ agro-climatological extremes. (2) Developing regions are highly exposed to DNACE compared to developed countries. (3) DNACE and heightened local susceptibility to minor hazards are linked to factors like political instability, Wars, migrations, and human interventions.

Spatio-temporal distribution of droughts with no agro-climatological extremes

Globally, out of the 91 inconsistent events (Fig. 2), the most significant number of DNACE events occurred in South, Central, and Southeastern Asia (34), closely followed by South and Eastern Africa (28) and South, Central, and Caribbean America (25). Eastern Europe witnessed three DNACE events, while Oceania had one event reported in Papua New Guinea. Notably, among all the DNACE events, only 7% of these events occurred in developed nations (Italy, South Korea, Russia), which are typically classified by the World Bank as high-income countries²³, while the vast majority, constituting 93%, took place in developing nations, often categorized as low to middle-income countries²³ by the World Bank. This stark contrast underscores the heightened vulnerability of developing countries to DNACE.

Fig. 2.

Spatial distribution of DNACE events during 2001–2020. (a) Visualization of global GDIS events with identification of inconsistent GDIS Events (i.e., DNACE). (b–d) The detailed picture of subnational level DNACE event locations over Cuba and parts of South America, Southeastern Africa, and South Asia, respectively. (The Figure was generated using ArcGIS Desktop version 10.8 (Esri Inc., https://www.esri.com)).

These events have had a substantial impact on a global scale, affecting approximately 36 million people (Fig. 3). Notably, the region most affected by DNACE events (34 in total) was South, Central, and Southeastern Asia, where a population of around 30 million people experienced the consequences of these events. Out of the 34 DNACE events in the Asian region, the highest number of events occurred in Vietnam (16) and Thailand (12). China experienced two events, while Afghanistan, Kyrgyzstan, South Korea, and Sri Lanka each encountered one event. All DNACE events over Asia were observed between 2006 and 2016. The region of South and Eastern Africa encountered the second-highest number of DNACE events, totaling 28, impacting approximately 3 million people. Among the African countries, Burundi had the highest number of DNACE events, with ten occurrences observed between June 2001 and December 2001 and throughout the entire year of 2011. Tanzania, Uganda, and Swaziland each experienced three DNACE events, followed by Djibouti, Malawi, and Rwanda, which recorded two events each. The continent of America registered the third-highest number of DNACE events, totaling 25, with most of these events occurring between 2010 and 2016. Approximately 2 million people were affected by these events (Fig. 3). Among the 25 events in America, the highest number of occurrences were recorded in Cuba (9), followed by Haiti and Trinidad & Tobago, with four events each. Additionally, Peru and Colombia reported three events each. DNACE events in Europe were documented in Italy (2)²⁴, associated with the influx of irregular immigrants, and the eastern part of Russia²⁵ was linked to socio-political disruptions. In the Oceania region, a single event was recorded in Papua New Guinea from May 2015 to November 2015, affecting approximately 250 thousand people (Fig. 3). The comparative analysis between GDIS and CDI sheds light on the changing patterns of drought behavior and their implication for vulnerable regions. Identifying DNACE events and their significant impact on the population underscore the urgency for targeted drought management strategies and further investigation into the potential reasons and triggers behind these occurrences. This emphasizes the importance of understanding and addressing the factors contributing to DNACE events to enhance drought preparedness and resilience in the affected areas.

Fig. 3.

Year and region-specific affected population Due to DNACE Events: (a) Colored circles represent different geographical locations and affected population counts. Triangles indicate the occurrence of DNACE events in specific locations without available population data. (b) Region-wise count of DNACE events worldwide.

Potential reasons behind droughts with no agro-climatological extremes

The widespread impact of DNACE on the global population highlighted the crucial significance of understanding the underlying reasons behind these drought events. Therefore, we explored the potential triggers behind these DNACEs, incorporating contemporary literature and documentary surveys. Additionally, to achieve a comprehensive knowledge of the on-ground situation during the DNACE event period, we referred to news articles, data from Reliefweb²⁶, reports, and various other sources available from that specific timeframe. This survey indicated that all 91 DNACE events were linked to various single or multiple socio-economic factors (Fig. 4) (Table 1). The primary reasons identified were migration, economic crisis, wars, political instability, and human interventions.

Fig. 4.

Spatial distribution of DNACE events and their potential reasonings. (The Figure was generated using ArcGIS Desktop version 10.8 (Esri Inc., https://www.esri.com)).

Table 1.

Region-wise details of inconsistent events and their potential reasons.

No.	Continent	No. of events	Affected population	Countries	Reasons
1	America (South, Central, Caribbean)	25	24,629,000	Cuba, Guyana, Haití, Honduras, Paraguay, Perú, Trinidad And Tobago	Political conditions, Water Shortages, Recurrent climatic shocks
2	Asia (South, Central, Southeastern)	34	177,073,304	Afghanistan, China, Kyrgyzstan, South Korea, Sri Lanka, Thailand, Vietnam	Food crisis, War, Migration for food and water
3	Africa (South, Eastern)	28	50,690,702	Burundi, Djibouti, Kenya, Lesotho, Malawi, Rwanda, Senegal, Swaziland, Tanzania, Uganda	Food crisis, Migration
4	Europe (Europe, Eastern)	3		Italy, Russia	refugees and migrants, Recurrent climatic shocks
5	Oceania	1	2,520,000	Papua New Guinea	refugees and migrants

Migration: According to the survey reports, a significant portion of the population from African countries like Burundi^27,28, Djibouti²⁹, Kenya³⁰, Rwanda³¹, Eastern Russia²⁵, etc., chose to migrate to nearby nations. The reasons for this migration were attributed to the lack of socio-economic resources and unstable local conditions in their home regions. During the DNACE event period in 2001, around 55,000 individuals in Rwanda³¹ were identified as refugees and asylum seekers. On the other hand, in the 2011 DNACE event, Italy and neighboring regions witnessed a significant influx of illegal immigrants, establishing the area as one of the gateways to Europe through the Mediterranean route²⁴. During this period, Italy witnessed the arrival of thousands of refugees seeking shelter, leading to increased pressure on the country’s resources. These situations highlighted the challenges faced by both the migrants seeking better opportunities and the host communities struggling to accommodate the additional population and provide essential resources such as food, water, shelter, and employment opportunities.

Internal migration due to economic instability and unequal distribution of resources: In the year 2012 (at the same time as the GDIS event report), around 20% population of Peru³² (South America) faced internal migration. During the end of 2015 (at the same time as the GDIS event report), Cuba³³ witnessed a record-breaking influx of internal migrants from the poorest regions, leading to overpopulation in the capital, Havana. Consequently, many of these migrants were forced to seek temporary shelters and tenement buildings, exposing them to elevated risk and living challenges³³. Similarly, in Sri Lanka³⁴ (South Asia), Guyana^35–37 (South America), and Vietnam^38,39 (Southeast Asia), migration to more economically viable areas was noticed due to economic hardship. This coping mechanism of economics placed an additional burden and stress situation on both the origin and destination regions. In Colombia, from 2015 to 2016, around 40 thousand individuals experienced forced displacement, with minors comprising 45% of the affected population. This displacement not only restricted children’s access to education but also contributed to heightened social stress in the region. The forced evictions of the people resulted in increased demands for housing and settlements, leading to the conversion of natural land resources and the construction of unplanned infrastructures and contributing to the generation of DNACE in the region.

Wars/ Conflicts / Political instability: In 2006 (at the same time as the GDIS event report), Afghanistan witnessed the death of over 300 Taliban fighters^40,41, while several African countries such as Somalia, Kenya, Tanzania, and Burundi experienced prolonged conflicts that lasted for a decade or more⁴². Consequently, these wars and conflicts resulted in the devastation of infrastructure, disruption of agricultural activities, and the imposition of physical and psychological stress on the affected populations in these regions. Throughout the respective GDIS period, Trinidad and Tobago⁴³, Tanzania⁴², and Prague⁴⁴ also experienced an unstable political environment due to general elections and regional conflicts. These political instabilities often lead to poor decision-making and mismanagement of resources, leaving vulnerable populations at greater risk.

Human interventions: Over the past decade, Peru (South America) has undergone rapid land-use and land-cover (LULC) changes, with the year 2010 (at the same time as the GDIS event report) being particularly significant in this regard⁴⁵. Likewise, starting from 2005 (during the GDIS event reported), Thailand experienced rapid development with the establishment of power plants and the construction of large industrial areas^46,47. These human interventions, whether direct or indirect, have resulted in deforestation, over-extraction of water resources, and inappropriate land use changes. These actions intensify stress conditions and have a detrimental impact on natural ecosystems.

The study highlighted that most of the DNACE events were significantly influenced by socio-economic factors. Wars, internal distress, forced migration, and prevailing political instability subsequently accelerated the social stress and exacerbated the socio-economic drought challenges. Additionally, the economic crises and human interventions further compounded the impacts of DNACE on vulnerable communities. To effectively address DNACE events and build resilience in drought-prone regions, it is imperative to tackle these underlying socio-economic issues. Mitigating and responding to drought-induced humanitarian crises requires a comprehensive approach that acknowledges and tackles the interconnected socio-economic drivers.

To validate DNACE events using an independent socio-economic dataset, we used the Famine Early Warning Systems Network (FEWS NET)⁴⁸ food insecurity outlooks, available from July 2009 onward, and regionally focused on Africa. Out of 91 global DNACE events, 28 were located in Africa, and 10 occurred after mid-2009, making them eligible for comparison.

Figure 5 demonstrates this comparative assessment, showcasing 8 events, spanning Kenya (1), Djibouti (1), Burundi (5), and Uganda (1), based on the DNACE-identified time spans. Across these events, red DNACE polygons consistently align with yellow to orange zones of food insecurity classified by FEWS NET as “Stressed” (IPC Phase 2) to “Crisis” (IPC Phase 3). This spatial overlap illustrates strong visual agreement between DNACE events and FEWS NET-defined food insecurity stress conditions. Two additional DNACE events were observed in Lesotho and Senegal, but corresponding FEWS NET data were not available or reported for those areas. The observed spatial consistency supports the alignment between DNACE and an independent socio-economic impact assessment. Given that FEWS NET integrates a wide range of institutional, economic, and humanitarian indicators, its reasonable match with DNACE events suggests that the DNACE framework is capturing meaningful information and serves as a credible impact layer.

Fig. 5.

Validation of DNACE events based on FEWS NET food insecurity classifications across African countries: (a) DNACE event polygons derived from GDIS data for specific regions and time periods. (b) corresponding FEWS NET food insecurity maps for the same locations and periods. The spatial overlap between DNACE polygons and FEWS NET-classified “Stressed” and “Crisis” areas indicates the presence of socio-economic stress during these periods, thereby supporting the validity of the DNACE framework. (The Figure was generated using ArcGIS Desktop version 10.8 (Esri Inc., https://www.esri.com)).

For cross-checking, ideally, we would require additional localized socio-economic datasets that are not fully captured by existing global resources. However, spatially and temporally resolved independent socio-economic impact datasets remain largely unavailable on a global scale. As a result, this case study represents the extent to which such validation is presently feasible for DNACE.

Discussion

DNACE challenges the conventional understanding of droughts and their underlying mechanisms. These events highlight the need to broaden our perspectives beyond meteorological and agrological variables and consider the complex interplay of various socio-economic and geopolitical factors contributing to drought vulnerability. The concept of DNACE itself is novel and has not been investigated to date. Previously, some studies examined the propagation of agro-climatological droughts to socio-economic conditions. A few studies estimated the drought threshold that triggered socio-economic distress in Germany⁴⁹ and the UK²⁰ using the European Drought Impact Inventory, whereas some^50,51 tried to compare the U. S. Drought Monitor and weekly drought severity maps to understand the socio-economic effects of droughts over a state of the USA. Efforts have also been taken to explore the propagation of agrometeorological hazards quantified from soil moisture to socio-economic impacts¹⁹. However, all these studies acknowledged that socio-economic repercussions were a result of agro-climatological drought severity, but none considered socio-economic factors as potential contributors to the causes of drought. This made the current DNACE work unique and set the foundation for its distinct identification and importance in drought assessment. Concepts in the drought literature, such as socioeconomic drought⁵², which generally describes the societal and economic consequences of climatological drought events, and creeping drought⁵³, which emphasizes the slow and imperceptible onset of environmental droughts, differ fundamentally from DNACE. In contrast, DNACE captures drought-like conditions that arise primarily from socio-economic, political, or anthropogenic stressors even in the absence of agro-climatological extremes. This distinction underscores the novelty of DNACE as a complementary framework in drought research and assessment.

To account for the possibility that drought-related socio-economic impacts may be influenced by prior conditions, we examined the effect of temporal lags on event classification. Specifically, we tested one, two, and three-month lag windows preceding each GDIS event to assess whether driver attribution was sensitive to antecedent conditions. When using the original GDIS-defined periods, 91 DNACE events were identified. This number decreased to 68, 54, and 52 when applying one, two, and three-month lags, respectively (Table S2). These results confirm that time-lagged effects can influence observed outcomes. However, in our main analysis, we retained the original GDIS event periods to ensure consistency. This decision was made due to the variable and sometimes imprecise nature of GDIS start and end dates, as well as the wide range in event durations, from single-month droughts to prolonged multi-season episodes. Applying lagged windows inconsistently across such varied events could introduce bias and compromise comparability. By sticking to the reported GDIS periods, we ensured a standardized and replicable approach for global-scale analysis.

Although we applied a consistent CDI threshold (≤ − 1) and minimum duration cutoff (≥ 2 months) for global analysis, our pixel-wise standardization ensures that droughts are defined relative to local climatological norms. Moreover, in a related study using Köppen climate zones⁵⁴, we found that CDI maintained strong performance across diverse climatic regions, supporting the applicability of this approach across different environmental contexts.

The comparative analysis between CDI and GDIS, along with the identification of DNACE events, has helped us better understand the vulnerability to drought in the region. We observed that the majority of DNACE events, specifically 88 out of 91, occurred in South and Central America, Sub-Saharan Africa, and parts of South Asia. In contrast, only three events were reported in developed regions like Europe and North America. This indicates that developing countries are at a higher risk of droughts. Previous studies^55,56 have also supported this notion by stating that countries with higher gross domestic product per capita, better infrastructure, and higher levels of education, such as North America, Japan, and European nations, tend to be less vulnerable to disasters. Developing nations may exhibit greater vulnerability to hazards due to their constrained resources, limited mitigation and adaptation strategies, higher poverty rates, and weaker governance and institutional structures^57,58. Our study underscored that migrations, wars, political instability, and human interventions play significant roles in the occurrence or exacerbation of DNACE events. These factors can significantly compound the challenges associated with droughts and have severe consequences on affected populations and ecosystems.

While numerous studies^59–61 have previously attributed migrations to environmental factors, our research takes a more comprehensive approach. We not only emphasize the environmental circumstances but also shed light on the fact that refugees and migrants, driven by economic crises, have further exacerbated socio-economic pressures, and placed additional stress on water resources and agricultural land. This heightened demand for resources, in turn, contributes significantly to water scarcity and increases vulnerability to DNACE events. For instance, decades of conflict in Afghanistan or Syria have resulted in widespread internal and international displacements. The movement of people into already stressed urban areas or neighboring nations like Lebanon, Jordan, and Turkey, in the case of Syria, has escalated water demand, resulting in socio-economic stress conditions.

Recent studies^62–64 showed that, in the pursuit of development, many developing countries, like Thailand, Vietnam, India, Peru, Nigeria, Ethiopia, etc., heavily invest in urbanization, industrialization, and dam construction. However, our study tries to understand the flip side of this development coin. Human interventions, such as converting natural landscapes like wetlands and grasslands into urban or agricultural areas, can disturb local ecosystems and diminish water retention capacity. Indonesia and Malaysia have extensively deforested natural forests in the region to make way for palm oil plantations, impacting regional climate patterns⁶⁵. In Haiti, deforestation for fuelwood extraction has resulted in the loss of natural forests⁶⁶, while in Africa, countries like Niger and Ethiopia have cleared forests for urbanization⁶⁷. Similarly, in California, USA, deforestation has occurred to accommodate urban development⁶⁸. These human interventions have disrupted natural ecosystems, reduced soil moisture, influenced local and regional climate patterns, and impacted regional water availability, intensifying DNACE events.

Previously, several studies^19,69 have utilized GDIS data to investigate the association between agro-climatological extremes and their socio-economic repercussions. A few regional studies^11,70,71 have also employed local or regional data and systems to explore this association. However, there is still a requirement for comprehensive data that includes detailed socio-economic information.

This study has several limitations and challenges that should be considered when interpreting the results. First, the GDIS, while useful for capturing subnational drought information, provides only around 60% coverage of the required drought events, leading to potential gaps in event identification. Second, both GDIS and EM-DAT often lack detailed metadata, such as precise location names or the extent of damage, hindering comprehensive event characterization. Third, many relevant local reports and news articles are published only in native languages or remain inaccessible online, posing significant language and data accessibility barriers. Lastly, spatial mismatches between the administrative polygons used in GDIS and the gridded pixel data of drought indices may introduce uncertainties in the spatial attribution of drought drivers. These limitations highlight the need for improved data integration and multilingual access to local drought impact information to better capture the complexity of DNACE events.

Looking ahead, future research could benefit from the integration of dynamic socio-economic indicators and models to better capture the evolving nature of drought vulnerability. Socio-economic drivers such as conflict, migration, and institutional resilience change over time and may influence how drought impacts are reported and perceived. Incorporating datasets like ACLED (conflict)⁷², UNHCR (displacement)⁷³, and the INFORM Risk Index (vulnerability)⁷⁴, or utilizing dynamic modeling approaches such as agent-based or system dynamics models, could help address these temporal complexities. Such approaches would allow for a more robust classification of DNACE events and improve our understanding of human-influenced drought outcomes.

These insights could also be important for policy considerations and risk management. DNACE highlights that drought risks are shaped not only by climatic and agricultural conditions but also by socio-economic pressures. Recognizing these factors in policy frameworks can help design more holistic drought risk management strategies that strengthen both environmental and social resilience. Early warning systems (EWS), for instance, could benefit from integrating migration data and other socio-economic indicators alongside agro-climatological variables. This would allow for the anticipation of stress hotspots where climatic extremes coincide with population pressures, enabling governments and humanitarian agencies to implement more proactive and targeted interventions. On a broader scale, acknowledging socio-economic drivers in drought planning and preparedness could enhance long-term adaptation strategies, improve resource allocation, and foster resilience in vulnerable regions.

Conclusion

In recent years, the global climate has witnessed a rise in the occurrence and severity of droughts, leading to more frequent and intense dry spells across the world. These alterations in the drought patterns, along with their societal impacts, are primarily attributed to the agro-meteorological vagaries. However, this study reveals that in some regions, droughts have occurred without extreme agro-climatological conditions. Therefore, here we aim to identify the reasons contributing to the genesis of droughts apart from agro-meteorological factors. The DNACE events were identified by comparing drought events derived from GDIS and CDI, and subsequently, the potential causes behind these DNACE events were determined through extensive literature surveys. The major findings from this study are as follows,

1. Droughts are not always natural; they can manifest even in the absence of significant agro-climatological anomalies or with only minor deviations from typical conditions.

2. During the study period from 2001 to 2020, globally, a total of 91 DNACE events were observed. Asia had the highest number of observed DNACE events, with Africa and America following closely on the list.

3. Among the total DNACE events, 88 were observed in developing regions, with only three recorded in developed countries. This disparity highlights the heightened vulnerability of developing countries to DNACEs compared to developed countries. The main driving factor behind this susceptibility is the fluctuating socio-economic scenarios prevalent in developing regions.

4. The fluctuating socio-economic scenarios, such as migration, economic crisis, wars, political instability, and human interventions, were the main underlying causes of DNACEs.

5. DNACE events have had a substantial impact on populations worldwide. Being the most populous regions, Asian countries like China, Vietnam, Thailand, and others have borne the brunt, with approximately 30 million people affected by DNACE events.

DNACE presents a complex challenge with far-reaching consequences. While they may not exhibit immediate impacts like extreme weather events, they still pose significant threats and can lead to long-term effects. Understanding their characteristics, causes, and impacts is crucial for effective mitigation and adaptation strategies to reduce their potential damage to communities and ecosystems. It is worth noting that Agro-climatological extreme-induced droughts are beyond our control, but DNACE events are solely caused by human interference, and it is within our capacity to manage and address them effectively to enhance resilience in susceptible regions.

To translate this understanding into practice, existing drought monitoring and early warning systems must evolve to include socio-economic vulnerability indicators, such as governance quality, migration trends, conflict zones, and land tenure changes, alongside traditional environmental triggers. Integrating DNACE awareness into regional drought plans and global frameworks (e.g., UNCCD⁷⁵, SDG targets⁷⁶) could help prioritize timely interventions in high-risk regions.

Future research should focus on developing composite indicators that capture both environmental and socio-political dimensions of drought risk. This is especially relevant for regions such as Sub-Saharan Africa, South Asia, and conflict-affected areas in the Middle East, where human vulnerability overlaps with institutional fragility. By bridging science, policy, and community insight, we can move toward early warning systems and adaptation strategies that are not only climate-responsive but also socially informed and equity-oriented.

Data and method

GDIS: The Global Disaster Dataset (GDIS)^1,2 is a geocoded disaster dataset developed based on the EM-DAT database⁷⁷. GDIS provides spatial geometry of administrative units affected by disasters in the form of GIS polygons. In this study, the selection criteria for shortlisting GDIS events required the drought period to be equal to or longer than two months. The duration of drought events is incorporated into GDIS by utilizing event identifiers from the EM-DAT database. EM-DAT records a natural disaster if any of the following criteria are fulfilled: 10 or more people are reported dead, 100 or more people are affected, or there is a declaration of a state of emergency and a call for international assistance. This comprehensive dataset highlights the distribution and intensity of drought events across different regions, providing crucial insights into socio-economic drought vulnerability and its implications on a global scale. GDIS is constructed using ground-based information, making it a valuable resource for validating other drought indicators obtained from satellite data, reanalysis, and climate models.

During 2001 to 2020, 1641 drought events were identified using GDIS. The African region, including Kenya (108), Somalia (53), and Ethiopia (46), experienced the highest number of GDIS occurrences at 732 events. The American region, comprising Honduras (52), the United States (48), and Brazil (33), closely followed with 403 events. Asia underwent 398 GDIS events, with Thailand leading at 144, followed by China (76) and Vietnam (32). Additionally, Europe experienced 80 drought events, with the highest number occurring in Italy at 24, followed by Croatia with 18 incidents. In Oceania, 28 GDIS drought events were recorded, with Papua New Guinea having the highest number at 14.

CDI: The combined drought indicator (CDI)^3,4 is a newly developed drought index (Figure S1) that integrates satellite and model-based two agricultural variables (vegetation and soil moisture) and two meteorological variables (precipitation and land surface temperature)⁷⁸. In this study, precipitation data from CHIRPS⁷⁹, vegetation data from the Normalized Difference Vegetation Index (NDVI) of MODIS⁸⁰, and land surface temperature and soil moisture from ERA-5 Land datasets⁸¹ were used to construct the CDI. To ensure comparability across sources, all datasets used for the development of the CDI were standardized and resampled to a uniform spatial resolution of 0.1° × 0.1° using a spatial interpolation technique. Furthermore, all variables were aggregated to a consistent monthly temporal scale covering the period from 2001 to 2020. The specific details of each dataset, including temporal coverage, native resolution, and access sources, are provided in Table 2.

Table 2.

Datasets and their respective sources used in this study.

Variable	Dataset name	Temporal resolution	Spatial resolution	Data Source Link
Rainfall	CHIRPS	2001–2020 (monthly)	Original = 0.05°*0.05° (Resampled on 0.1°*0.1°)	https://www.chc.ucsb.edu/data/chirps
Soil moisture	ERA5-Land	2001–2020 (monthly)	0.1°*0.1°	https://cds.climate.copernicus.eu/cdsapp#!/dataset/reanalysis-era5-land-monthly-means
NDVI	MODIS	2001–2020 (monthly)	Original = 1 km (Resampled on 0.1°*0.1°)	https://modis.gsfc.nasa.gov/data/dataprod/mod13.php
Surface Temperature	ERA5-Land	2001–2020 (monthly)	0.1°*0.1°	https://cds.climate.copernicus.eu/cdsapp#!/dataset/reanalysis-era5-land-monthly-means
GDIS	SEDAC NASA (Based on—EMDAT),	event-wise	subnational level	https://sedac.ciesin.columbia.edu/data/set/pend-gdis-1960-2018/data

In the further step, the principal component method was employed to combine all the variables and determine their respective weights. The following formula was used to compute the CDI.

Where Wt SPI, Wt LST, Wt NDVI, and WtSM stand for the PCA-based weight (Figure S2) (Table S1) values for individual parameters, i.e., Standardized Precipitation Index (SPI), Land Surface Temperature (LST), NDVI, and Soil Moisture (SM), respectively. Here, ‘z’ represents a specific year, and ‘i’ indicates the corresponding month (January to December) for the standardized input data. After normalization, all the CDI results were visualized on a monthly scale. The CDI ranges from highly negative values, indicating extremely dry conditions (-2.00 or lower), to highly positive values, representing extremely wet situations (+ 2.00 or higher).

Comparative analysis between GDIS and CDI to get DNACE

To identify DNACE, a comparative study between GDIS and CDI was performed. The first step involved superimposing all GDIS event polygons onto the corresponding CDI results for their respective periods. Further, all the CDI values within the specific GDIS event period were recorded and examined. During the analysis, if the CDI values were below the − 1 threshold during the particular event period recorded in GDIS, then that the GDIS event was categorized as a consistent drought event. Conversely, if any CDI value was greater than − 1 during the examination, the event was marked as an inconsistent GDIS event. Note that our detection criterion for DNACE is conservative, maximizing the number of consistent drought events. Even though there is only one pixel with CDI less than − 1, it is identified as a consistent event (non-DNACE event).

To evaluate the performance of detecting GDIS events using different drought thresholds, we considered multiple threshold values of the CDI (0, -1, -1.5) and analyzed their association with GDIS (Table S2). Since droughts can result from cumulative conditions or delayed impacts, we also examined CDI values one month and two months prior to each reported event. The analysis was performed on a pixel-wise basis, using each pixel’s long-term historical data to account for local climatological variability. The − 1 threshold was selected as a reference point due to its widespread use in standardized drought indices representing moderate to severe drought conditions. A detailed explanation of the methodology and extended results can be found in our previous paper²², which serves as a methodological foundation for the current work. Similarly, we also considered temporal lags of one, two, and three months prior to the exact GDIS event period (Table S2), which resulted in different sets of observed events. However, we ultimately used the original GDIS-defined period to ensure uniformity and compatibility across all analyses.

Identification of potential reasoning behind DNACE

A comparative analysis was conducted between CDI and GDIS, leading to the derivation of DNACE. To gain a deeper understanding of DNACE and its underlying causes, a thorough investigation was carried out for each event. This investigation included referencing online literature, news articles, reports, and relief web data pertaining to the time when the events occurred (Table S3). Various potential factors, such as wars, migration, political instability, and others, were considered as reasons for each event. In instances where multiple factors were present, the events were classified by combining the relevant classes that corresponded to the observed factors in their respective regions. One of the limitations of this approach was the reliance on online sources for references. While a majority of the records in English were readily accessible on the internet, there were instances where local news reports might have documented events and their specifics in local languages, making them unavailable on online platforms.

Stepwise methodology for assigning drought impact drivers

To establish a consistent and literature-supported classification of drought impact drivers, we first conducted a scoping review of foundational studies^72,82–86. These works consistently identified migration, conflict, land use change, economic distress, and political instability as common pathways through which drought impacts unfold. Based on this, we defined five driver categories:

• Migration (Ecological or Economic).

• War and Conflicts.

• Land Use and Land Cover Change.

• Economic Crisis.

• Political Instability.

To validate and contextualize these categories, we conducted a systematic search in the Web of Science (2001–2023) using the query:

Topic Search = (drought AND (migration OR conflict OR “land use change” OR “economic crisis” OR “political instability”)).

From ~ 2,000 returned articles, we reviewed a relevant subset with clear spatiotemporal alignment to GDIS events. Later, for each event, we used its reported start and end dates from GDIS, to conduct additional searches via Google, ReliefWeb, FAO/GIEWS, and UN OCHA platforms. Only sources published within the event timeframe and explicitly linking drought to socio-economic or political impacts were considered. We prioritized institutional reports, peer-reviewed publications, and reputable news sources (BBC, Reuters, The Guardian, Al Jazeera).

A driver was assigned to a GDIS event if the source clearly described an impact matching one or more of the predefined categories. Multiple drivers could be assigned where supported. In cases of ambiguity, assignments were made only if confirmed by multiple sources or high-confidence institutional reports.

For example, a 2001 ReliefWeb report on Burundi described population movements due to insecurity and economic hardship, leading us to assign Migration and Economic Crisis as drivers. Similarly, a 2011 article in The Guardian highlighted food and water shortages in Djibouti, worsened by weak infrastructure and limited aid, supporting the same driver pair. In both cases, socio-economic pressures, rather than agro-climatological stress, were the dominant triggers.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

S.K. and Y.S. designed the study, with S.K. executing the work. S.K. prepared the manuscript with contributions from Y.S. Y.S. secured the funding. S.K. and Y.S. reviewed and revised the manuscript.

Funding

This study was supported by the KAKENHI grants (Grant No. 21H01430, 25H00760 and 24K17352) and the JAXA grants (Grant No. ER3AMF106 and ER4AMF101).

Data availability

All the original datasets used to compute CDI are publicly available online. The Python code to compute CDI weights is available in the Zenodo repository at [https://zenodo.org/doi/10.5281/zenodo.10112546](https:/zenodo.org/doi/10.5281/zenodo.10112546) .

Code availability

Python script (Python v.3.9.12) was used to compute CDI weights. The code is available in the Zenodo repository at https://zenodo.org/doi/10.5281/zenodo.10112546.

Declarations

Competing interests

The authors declare no competing interests.