The global population is projected to reach 9.7 billion in 2050 and 10.4 billion in 2100, leading to a doubling of agricultural demand by 2050. This increased demand, coupled with projected increases in climate variability due to global climate change, will strain sustainable and equitable food security. Drought, in its various forms, poses significant challenges to food systems and agricultural productivity. While projections show increases in drought frequency, severity, and spatial extent across many regions, uncertainties remain, particularly in regions like Southeast Asia's monsoon region. However, elevated drought risk is consistently expected across Central America, Europe, and the Amazon. Flash drought, characterized by rapid onset and intensification, presents a unique challenge due to the difficulty in implementing mitigation strategies with limited warning. The 2010 flash drought in western Russia serves as a stark example, causing widespread impacts including heatwaves, wildfires, air pollution, displacement, and significant wheat yield losses, leading to global price increases due to export bans. This study addresses the projected trends in global flash drought frequency in a warming climate and the changes in agricultural risk from flash drought in the future, using simulations from six Coupled Model Intercomparison Project Phase 6 (CMIP6) models.

Existing literature highlights the projected increase in drought frequency, severity, and spatial extent globally due to climate change. Studies using CMIP5 models indicate uncertainty and regional hotspots in future agricultural drought. While some areas show complicated changes in drought frequency due to precipitation uncertainties (e.g., Southeast Asia's monsoon region), consistent increases are projected for Central America, Europe, and the Amazon. Flash drought, with its rapid development, presents unique challenges for mitigation and early warning systems, as evidenced by the devastating impacts of the 2010 Russian flash drought. Prior studies using CMIP6 models have investigated future drought projections in specific regions, some finding increasing flash drought risk in the late 21st century (e.g., southeastern China), while others have shown that changes in flash drought frequency are complicated by monsoon variability (e.g., India). Previous work has also shown that flash droughts are projected to increase in duration and severity in a warmer climate.

This study uses six models from the Coupled Model Intercomparison Project Phase 6 (CMIP6) to assess flash drought projections. Historical simulations (1850–2014) and three future scenarios (SSP126, SSP245, and SSP585) spanning 2015–2100 were analyzed. Flash drought was quantified using evapotranspiration (ET), potential evapotranspiration (PET), and soil moisture from the CMIP6 models on daily timescales. The standardized evaporative stress ratio (SESR) was derived from ET and PET to quantify the rate of drought intensification. Soil moisture defined moisture thresholds in flash drought development. Four reanalysis datasets (MERRA, MERRA-2, ERA Interim, and ERA5) were used to evaluate the CMIP6 models' accuracy in capturing historical flash drought characteristics (1980–2014). CMIP6 models were assessed against reanalysis datasets to ensure accuracy before future projections. The seasonal cycle of flash drought occurrence was examined across 15 global hotspot regions. Future projections focused on changes in flash drought frequency under different scenarios and changes in agricultural risk from flash drought. A modified framework from a previous methodology was used to identify flash drought events, utilizing three criteria: ASESR (standardized change in SESR) at or below the 25th percentile, a minimum length of five pentad changes in SESR (30 days), and a final soil moisture value below the 20th percentile. Percentiles were derived from 1980-2014 data to maintain consistency. Flash droughts were identified based on latitude and season. An ensemble averaging approach was used for both spatial and temporal analysis. Data was interpolated to a 0.5° x 0.5° grid. Arid and cold regions were masked based on aridity index and PET thresholds. The study employed a multi-dataset, multi-model ensemble approach and a multivariate identification method to enhance the reliability and robustness of results. The relative contributions of precipitation and evaporative demand to changes in flash drought frequency were analyzed. Flash drought intensification rates were also investigated. Finally, data on land use from the Land-Use Harmonization (LUH2 v2f) project and future SSP land-use scenarios were used to understand the influence of land use changes on flash drought.

The study found a global increase in flash drought occurrence by the end of the 21st century across all three future climate scenarios (SSP126, SSP245, and SSP585), with the largest increase observed under the SSP585 scenario (8.2%). Regionally, the most substantial increases were projected for Europe and the Amazon. In global flash drought hotspots, future projections showed increased frequency across most regions and scenarios, with statistically significant increasing trends in many regions. The SSP585 and SSP245 scenarios often showed elevated flash drought frequency compared to SSP126. Divergent trends were observed in some regions (northeastern China, India, the Great Rift Valley, and northern Australia), with SSP585 showing decreasing trends in some instances. Regarding cropland risk, all continents showed increases in flash drought-affected cropland area in all scenarios, with the most significant increases projected in North America and Europe under SSP585. SSP126 showed the highest projected risk to Australian croplands, while SSP245 projected the lowest risk. A ‘tipping point’ for flash drought risk was evident in several regions, where risk increases significantly between SSP245 and SSP585. Analysis of drivers revealed that increased potential evapotranspiration (PET) was a more significant contributor to increased flash drought frequency than decreased precipitation, particularly in the context of climate change. While precipitation increases were projected in some hotspot regions with minimal changes in flash drought frequency, decreases in precipitation, especially in SSP245 and SSP585, were associated with increased flash drought in regions like the Amazon, Iberian Peninsula, and Asia Minor. Higher PET anomalies in future scenarios corresponded with expected flash drought increases, especially in Europe and higher latitudes of North America. Regions with minimal PET increases showed negligible or small decreases in flash drought frequency. Across most hotspot regions, the percentage change in PET was larger than the negative percentage change in precipitation, indicating a larger influence of PET in increasing flash drought frequency. Although precipitation has historically been a dominant driver, future increases in PET are projected to be more impactful. Soil moisture changes were minimal compared to changes in precipitation and evaporative demand. The study also found a global increase in flash drought intensification rate across all scenarios, with the largest increases under SSP585. These increases were particularly notable in northern South America, the Sahel, and parts of India. Factors contributing to faster intensification include higher PET and differences in land cover. Deforestation in the Amazon exacerbated the increase in flash drought intensification rate. The increasing risk of flash drought to croplands will further strain food security, especially considering projected cropland expansion and intensification.

The findings highlight the significant increase in flash drought risk globally in a warming climate, regardless of the emission scenario. Increased PET emerges as a dominant driver of this rise, outweighing the impact of precipitation changes in many regions. The study's results align with regional studies on flash drought projections in specific areas such as southeastern China and India. The amplification of flash drought risk by the interplay of reduced precipitation and enhanced PET, particularly in already vulnerable regions like the Amazon, emphasizes the need for comprehensive adaptation and mitigation strategies. The observed ‘tipping point’ effect underscores the importance of even moderate emission reduction scenarios to minimize the drastic increase in agricultural risk. The large difference in projected agricultural risk between the SSP585 scenario and lower-emission scenarios highlights the substantial impact of climate change on food security. This risk is exacerbated by the rapid onset and intensification characteristic of flash droughts, significantly reducing response time. The significant role of deforestation in accelerating flash drought intensification in regions like the Amazon reinforces the importance of sustainable land management practices.

This study provides crucial projections of flash drought occurrence and agricultural risk under various climate change scenarios. The findings show a consistent global increase in flash drought frequency and intensified risk to croplands across all scenarios, with the most extreme scenario indicating dramatic increases in North America and Europe. Increased PET is identified as a primary driver. The study highlights the significant impact of flash drought on global food security and the need for proactive mitigation and adaptation strategies. Future research should focus on regional atmospheric and oceanic drivers of flash drought and on developing local-scale mitigation strategies.

The study relies on CMIP6 models, which have inherent uncertainties. While the study used an ensemble of models and incorporated reanalysis data for validation, inherent uncertainties associated with climate models may limit the precision of specific regional projections. The methodology for flash drought identification, while robust, may still have some sensitivity to the specific parameters used. The study focuses on flash drought occurrence and agricultural risk but does not thoroughly explore other potential impacts of flash droughts. Finally, the use of a fixed PET calculation method might not completely capture future changes in the impact of increased CO2 concentration on evapotranspiration.