1. Introduction
Extreme heat and heavy rainfall events have already intensified in both magnitude and severity across many parts of the United States. The Southeast region is particularly vulnerable to extreme weather, including severe thunderstorms, floods, tropical cyclones, and coastal flooding. These threats are worsened by the region’s growing and expanding population [the U.S. Global Change Research Program (USGCRP 2023)]. Projections indicate that these extreme events will continue to escalate with climate change, with heavy rainfall events potentially becoming twice as frequent by the end of this century (Seneviratne et al. 2021). For locations in the southeastern United States, the number of days with high minimum temperatures—defined as nighttime temperatures that remain above 23°C—has been increasing (Carter et al. 2018). This trend is projected to intensify, with some areas experiencing more than 100 additional warm nights per year by the end of the century (USGCRP 2023). Other areas within this region, including Louisiana, which is associated with the highest land loss rates globally, remain vulnerable to severe weather events. Climate change, particularly rising sea levels and possible increases in hurricane intensity, is anticipated to impact the coastal ecosystems of such locations in the Southeast significantly (USGCRP 2023). Such changes will have significant socioeconomic costs, highlighting the urgent need for climate adaptation strategies in this region (Newman and Noy 2023).
Given these anticipated changes, climate change adaptation and disaster preparedness are critical in the southeastern United States. Proactive disaster preparedness is particularly challenging because policies are often informed by past experiences, which may no longer be relevant in a rapidly changing climate (DeLeo 2017). Extreme weather events and other disasters typically serve as catalysts for policy changes (Haque et al. 2019; Birkland 2009; Adger 2003), locking emergency preparedness into a reaction cycle rather than fostering a proactive approach to future crises. This is especially problematic in a changing climate where past experiences provide a poor predictor of future conditions. For instance, Miao (2019) found that state governments in the United States were more likely to engage in climate adaptation planning following recent extreme weather events. While this reactive approach can be beneficial, it often fails to integrate future climate projections, thereby excluding areas that have yet to experience significant disasters.
To emerge from this reactive cycle, the climate science community has proposed using “storylines” of extreme and unprecedented events to help people visualize the types of disasters that are possible in a changing climate. Storylines were introduced to focus not on quantifying the probabilistic aspects of climate change but rather to create descriptive narratives of possible future events or climate pathways (Shepherd et al. 2018; Baulenas et al. 2023). This is partly because traditional presentations of climate research results often focused on the changes to average variables, washing out focusing on tail-risk extremes. Conventional probabilistic models also, by their inherent nature, imply that the focus should be on the most likely events rather than on helping communities that consider events that, while they might not be highly probable, would be devastating if they took place. Events like the recent devastating Hurricane Milton in Florida in 2024 or the Lahaina wildfires in 2023 highlight the need to consider a wider range of tail risks. By not emphasizing probabilistic or average outcomes, storylines enable consideration of a fuller spectrum of plausible impacts (Ciullo et al. 2021).
The storyline approach has been used in recent years to identify storylines of weather-induced crop failures, designed to allow the disaster management community to carry out realistic stress-testing exercises (Sillmann et al. 2021) as well as on socioeconomic data (Young et al. 2021; van den Hurk et al. 2023). Others have used an “ensemble boosting” approach where the model is reinitialized with slightly altered atmospheric conditions starting from 5 days to about 3 weeks before the peak extreme event of interest to model how historical events could have been worse (e.g., Fischer et al. 2023) or different in the future with climate change (van der Wiel et al. 2021). These approaches can be useful in cases where there is high uncertainty in how climate change might affect atmospheric circulation, enabling physically consistent plausible storylines for the outcomes of different circulation changes (Zappa 2019). Ciullo et al. (2021) suggest that the storyline approach offers more flexibility for practitioners to stress-test systems by identifying plausible factors that could severely stress current systems. Shepherd et al. (2018) point out that the storyline approach improves risk awareness by triggering memory and knowledge that comes from the lived experience of climate events, which have already been experienced and to which individuals can relate directly. The storyline approach can also encourage community leaders, decision-makers, and policymakers to identify robust approaches by revealing a wider variety of variables that can lead to systematic failures, including social and economic factors that might disadvantage segments of the population. It has been demonstrated that the experience of unprecedented weather can create an opportunity for legislation to pass that otherwise would not have been implemented (Hui et al. 2022). By integrating past learned experiences of climate events with new scientific information of future risks delivered through a storyline approach, scenario planning can try to mimic the policy momentum that has been shown to come from experiences of climate change.
When used alongside scenario exercises, storylines enhance planning and community engagement. They encourage practitioners and communities to understand uncertainties and multiple possible futures simultaneously, aiding in understanding potential risks. van den Berg et al. (2021) note that scenarios have long been used in combination with storytelling to allow decision-makers to consider in a systematic fashion multiple plausible futures and imagine the consequences of different alternative decisions. Scenarios are particularly well suited for long-term planning when uncertainties are high and multiple outcomes are possible (Varum and Melo 2010) and when combined powerfully with storylines that tap into lived experience, provide important tools to address the policy development challenges of disaster preparedness and climate adaptation.
The unprecedented simulated extremes using ensembles (UNSEEN) approach involves employing a vast ensemble of initialized climate model simulations to determine feasible climatic conditions that might have occurred in the recent past but did not occur (Thompson et al. 2017). Such large ensemble simulations have been established to estimate extreme value statistics (Diffenbaugh et al. 2017; Kelder et al. 2020; Gessner et al. 2021). Many studies using this UNSEEN approach investigate a historical event to understand how it could be worse (Sillman 2020) or create future impact analogs of historical events (Goulart et al. 2021).
Given this background, the main objective of this paper is to create a typology of heat wave and flooding storylines based on research conducted in five counties across the Southeast United States that are participating in the first round of the American Red Cross Community Adaptation Program whose primary aim is to develop scenario planning tools that people can use to prepare for extreme events. Our storylines created based on the UNSEEN approach will describe the historical risk of extreme heat and precipitation risk over the last 40 years (“recent memory”) and estimate the observed increase in the risk of these extreme events through the year 2021 for the same counties. Ultimately, our findings will provide valuable insights for community planners and decision-makers, enabling them to better understand the evolving risks and to develop effective adaptation strategies for their communities.
2. Data and methodology
Since we focus on investigating heat wave and flood risk, we use model and observational records of temperature and precipitation datasets. In this study, we employ the UNSEEN approach using temperature and precipitation archives of fifth generation seasonal forecasts (SEAS5), the long-range modeled seasonal forecast dataset provided by the European Centre for Medium-Range Weather Forecasts (ECMWF; Johnson et al. 2019). SEAS5 serves as an operational forecasting system, but its archived large ensemble data are particularly valuable for identifying unprecedented extreme events. This dataset includes a range of plausible scenarios that have not yet occurred, allowing researchers and practitioners to explore potential future extremes and better prepare for unforeseen circumstances. The model is initialized on the first of every month and runs for 7 months into the future, generating daily weather data for a 7-month time period. The model comprises 25 ensemble members from 1981 to 2016 and 51 from 2017 to the present. With 5 lead times, this large ensemble offers 125 alternative scenarios per year up to 2016 and 255 ensemble runs per year from 2017 onward.
The Climate Hazards Group Infrared Precipitation with Stations (CHIRPS; Funk et al. 2015) and Daymet (Thornton et al. 2020) were the observational datasets for rainfall and temperature, respectively. These datasets are natively available at a daily temporal resolution. SEAS5, CHIRPS, and Daymet datasets are available at a 1°, 0.25°, and 1 × 1 km2 spatial resolution, respectively, but for comparability purposes, before the analysis, all datasets were regridded to the spatial resolution of SEAS5. Despite the high resolution of Daymet as a result of numerous preprocessing, Sy et al. (2024) in their study compared the representativeness of widely used Earth system reanalyses (ERA5, ERA-Interim, MERRA-2, and NARR) and the Daymet in situ–based gridded datasets with in situ measurements from the National Oceanic and Atmospheric Administration U.S. Climate Reference Network for daily temperature and precipitation over the United States from 2006 to 2020. The results of their analysis revealed that the Daymet gridded datasets performed better than the ERA5 reanalysis in most regions of the United States. For this reason, we decided to use the Daymet dataset as our observational dataset.
As discussed in the previous section, storylines derived from large ensembles can significantly inform policy development and motivate communities to adopt more proactive, forward-thinking strategies/policies. We utilize the aforementioned model and observational datasets to create storylines for locations in the southeastern United States, aimed at assisting the disaster management community in envisioning and preparing for unprecedented synthetic extremes (Coughlan de Perez et al. 2023). Specifically, we analyze ensemble model simulations for each location of interest (details provided in the forthcoming paragraph) to assess how the risks of extreme heat and extreme rainfall have evolved over the past 40 years. We also investigate potential storylines of both actual and unprecedented synthetic extreme events that could serve to stress-test climate change adaptation and disaster risk management plans in each area. This process involves crafting detailed narratives of potential extreme events based on historical occurrences and hypothetical scenarios. The goal is to evaluate the robustness and effectiveness of current strategies while identifying areas needing improvement. Additionally, we pinpoint the most extreme event in the ensemble simulation, which can then be utilized as a benchmark for unprecedented synthetic extreme events in disaster scenario planning.
In this study, we define extreme events as the highest observed and potentially realized values of monthly maximum rainfall and temperature for each month based on the observed and SEAS5 datasets, respectively. Furthermore, we calculate the change in the risk of these extreme events between 1981 and 2021, a period we refer to as recent memory. To answer these questions, we apply extreme value statistics to determine return periods for 1981 and 2021. We fit a nonstationary generalized extreme value (GEV) distribution to both historical data and the UNSEEN ensemble based on the likelihood ratio test results to compare models (stationary GEV vs stationary Gumbel and nonstationary GEV vs stationary GEV) for each extreme event in each month. We linearly fit the location and scale parameters to time, allowing us to calculate the frequency and magnitude of each extreme event. In our analysis of the observed record, we document the most extreme event—either the highest daily precipitation total in the month or the highest daily maximum temperature in the month—during this recent memory period and assess the frequency with which such an event could occur under the climate conditions of 2021.
The locations of focus included in our study are as follows:
- 1)Montgomery County, Alabama, which contains a large urban area (the state capital). Previous research has indicated that historical experience of extreme flood events in Alabama is a good predictor of community preparedness (Shao et al. 2017).
- 2)Yazoo County, Mississippi, which is a rural county located on the Yazoo River, a tributary of the Mississippi River. The state expects increased precipitation and runoff in the lower Mississippi River (Dunne et al. 2022) and has already documented the negative impacts of extreme temperatures on crops (Anapalli et al. 2021). Mississippi also has one of the highest rates of workplace heat-health mortality in the United States (Gubernot et al. 2015).
- 3)Madison County, Tennessee, is a rural county in western Tennessee. The state has documented increased costs associated with extreme weather events, for example, in the Department of Transportation (Venner and Zamurs 2012).
- 4)Warren County, Kentucky, is a rural county on the northern edge of the southeastern United States.
- 5)Terrebonne Parish, Louisiana, is a coastal county on the Gulf of Mexico. Hydrological change and rising sea levels have increased Louisiana flooding (Getirana et al. 2023).
In the model simulations and observational datasets, county data were selected by identifying the grid points that either fall within the county boundaries or are the nearest grid points to the specified locations. The spatial mean was then calculated using these selected grid points. We assess heat wave risk using the maximum temperature in the month, while flooding risk is evaluated based on the highest daily precipitation total in the month.
For each variable of interest (monthly maximum temperature and precipitation), month, and county, we answer two questions: 1) What is the current return period of an event that would have a 1-in-100 return period in 1981? and 2) what is the current return period of the extreme event observed in recent memory? To do this, we fit an extreme value distribution to the data. We first fit a Gumbel distribution and then carry out a likelihood-ratio test to test whether a GEV fit is a significantly better fit for the data. In the case when this is true, we then test whether a nonstationary GEV is a significantly improved fit for the data, fitting the location and scale parameters linearly to time as a covariate (Kelder et al. 2020). We then use the nonstationary GEV results for the locations in which this is the best fit to articulate storylines for understanding climate change in locations with similar characteristics. (Results from likelihood-ratio tests are available in the online supplemental material.) First, we calculate how much the extreme has changed over 40 years between 1981 and 2021. To do this, we use the UNSEEN ensemble to estimate the 1-in-100-yr return period event in 1981 and calculate the return period of such an event in 2021. Return periods of less than 100 years indicate an increasing frequency of such extremes. Second, we identify the historical observed maximum in the 1981–2021 period and estimate the return period of such an event in 2021. If the highest event in the last 40 years has a high frequency (low return period) in the current climate, this indicates that people have little lived experience of the types of events that would be considered extreme in today’s climate.
3. Results
The climatology of monthly maximum temperatures in the regions of interest (Fig. 1a) indicates that all regions experience extremely high temperatures (>32.5°C) from May to September, posing a significant risk of heat waves. Similarly, Fig. 1b illustrates the variation in monthly total rainfall. Apart from Terrebonne County, which experiences peak rainfall (200 mm) in June, the other counties have diverse climatologies but generally experience peak rainfall (between 120 and 160 mm) from December to May. These months, characterized by relatively high precipitation totals, are associated with a high risk of flooding. Consequently, these months are the focus of this study due to their heightened risk of heat waves and flooding.
The climatological mean (derived from Daymet) of the highest monthly (a) maximum temperature and (b) total precipitation for each calendar month in the five counties of the southeastern United States.
Citation: Bulletin of the American Meteorological Society 106, 3; 10.1175/BAMS-D-23-0297.1
Results for all counties and the selected months of focus are included in Table A1 in the appendix. However, some of Terrebonne Parish, Louisiana’s temperature results failed multiple fidelity tests (as is shown in the supplemental material), indicating that the SEAS5 ensemble does not produce realistic weather in that location. Coastal interactions that influence temperature variations could also be poorly represented in the model. Therefore, we have not included results from Terrebonne Parish in the table. Fidelity test results for all locations are shown in the supplemental material. Other counties and months that failed multiple fidelity tests are indicated as “not calculated” in the table. Months that failed one (two) fidelity test but passed the others are indicated with one (double) asterisk(s) and should be interpreted cautiously; the observed value was within the 95th percentile of the SEAS5 ensemble for three of the four tests. Results from this table also show that extremely high daily rainfall is becoming more frequent in many locations for much of the year, with the most dramatic change during winter months.
An example of two of the UNSEEN results and extreme value distributions is presented in Fig. 2, in which historical observations are shown as blue crosses overlaid on the ensemble of simulations. Figure 2a depicts results for extreme temperatures in August in Yazoo County, Mississippi, and Fig. 2c. illustrates the change in return periods of extreme events. The observational and UNSEEN ensembles indicate an increase in the frequency of extreme hot events between 1981 and 2021. This is representative of the temperature results from many of the other locations in the study (see Table A1). Figure 2b depicts the UNSEEN ensemble and observational results for Madison County, Tennessee. Figure 2d plots the change in return periods for extreme precipitation, indicating that extreme precipitation has increased in frequency for Madison County, Tennessee. The return period of extreme precipitation has decreased over the last 40 years, although the most extreme event in the historical record (yellow line) remains an uncommon and extreme event, even in 2021. The selection of Yazoo County for maximum daily temperatures in August and Madison County for maximum daily precipitation in February was made to highlight two distinct types of climate extremes relevant to the respective regions. Yazoo County, located in Mississippi, is prone to experiencing extreme heat during the summer months, making it an ideal candidate for illustrating the impact of high temperatures. Conversely, Madison County in Tennessee often experiences heavy precipitation during the winter months, particularly in February, which provides an opportunity to examine precipitation extremes. These contrasting examples allow us to explore the diverse nature of climate risks across different locations and seasons, emphasizing the need for region-specific adaptation and disaster risk management strategies.
Example of UNSEEN results for two counties. (a),(c) Maximum daily temperatures in Yazoo County in the month of August. (b),(d) Maximum daily precipitation totals in Madison County in February. (a),(b) Historical data for temperature and precipitation (blue x) alongside the results of the large ensemble of simulations for each year (gray boxplots). The box depicts the median and interquartile range of the ensemble, and the gray semitransparent dots are any outliers lying outside 1.5 times the interquartile range. (c),(d) The return period GEV fits for the UNSEEN ensemble as a solid line with dark uncertainty margins. The observational data are plotted as dashed lines (no uncertainties plotted because they are so large). The black horizontal lines indicate the 1-in-100-yr event in 1981 for both locations, and the purple horizontal lines indicate the historical maximum in the 1981–2021 dataset for each location.
Citation: Bulletin of the American Meteorological Society 106, 3; 10.1175/BAMS-D-23-0297.1
Based on the results of this analysis, we propose a typology related to the severity of the most extreme event in recent memory and the rate of change of extreme events in a location. Figure 3 (y axis) presents the return period 2021 of the strongest events on record for precipitation and temperature in the month with the highest magnitude of these variables for each county. Counties with recent experiences of events considered extreme in today’s climate have high values on the y axis. These are few, and most recent temperature records have low return periods in today’s climate. The x axis of the same plot represents the magnitude of the change in the frequency of extremes for each location. For precipitation and temperature, current return periods are almost always lower than the return periods of 1981 (clustered to the left of the plot).
(a) A schematic showing the change in risk of an event in comparison to the worst event in recent memory, with quadrants labeled as to their implications. The x axis represents the current return period (in 2021) for an event that would have been a 1-in-100-yr event in 1981. Results to the left of the vertical line indicate an increased frequency of an event by 2021 relative to 1981. (b) Results from Table A1 (in the appendix) for temperature (triangles) and precipitation (circles) for the four analyzed counties, labeled by the month and year of the hottest or wettest day in the 1981–2021 record for each month–location combination. This uses a log scale on the x axis due to the clustering of the results.
Citation: Bulletin of the American Meteorological Society 106, 3; 10.1175/BAMS-D-23-0297.1
Therefore, results in the bottom-left quadrant have a high risk of surprise because extreme events are becoming more frequent, and the most extreme event in recent memory is no longer very extreme. We call this region “sitting ducks” because people might be unaware of the increasing risk. On the other hand, results in the top-right quadrant have a low risk of surprise because extreme events are becoming less frequent; yet, people have recent experience with unusual extreme events (“recent rarity”). Results in the top-left quadrant show a recent experience with an extreme event and an increased frequency of events, which we call “living memory.” Results in the bottom-right quadrant show a decreasing frequency of events but have no recent experience with a strong extreme, what we call “fading memory.”
Based on these results, we propose two distinct storylines for use in disaster management and scenario planning. In the first storyline, that of “high potential for surprise” and sitting ducks, the 1-in-100 event from 1981 is much more likely now, and the worst event in recent memory has a low return period in today’s climate, therefore not representative of an unusual extreme. Results in the bottom-left quadrant of Fig. 3 would qualify as high potential for surprise. For example, the highest recorded value in the Daymet dataset is for Yazoo County, Mississippi, average temperature is ∼41°C for August, corresponding to a 27-yr return period in today’s climate. The UNSEEN ensemble produces events exceeding ∼43°C for countywide temperatures in August (see Fig. 2). The rate of change is high, with the 1-in-100-yr event from 1981 now an event that happens approximately every 10 years.
A second example of this storyline would be early season heat extremes in Montgomery County, Alabama. High temperatures early in the summer season have been shown to have higher consequences on human health than similar temperatures later in the season, likely due to a lack of preparedness and acclimatization (Liss et al. 2017). Montgomery County recorded a maximum July temperature of ∼39°C in June 1985, which at the time was an event with a return period of 1-in-500 years. The return period of that same event is approximately 21 years, and the highest value produced by the UNSEEN ensemble for a June temperature in Montgomery is ∼43°C. Ultimately, what people might remember as an extreme event is no longer extreme, thus creating a situation that has a high potential for surprise.
March rainfall in Montgomery County also falls into the high potential for surprise storyline. The highest recorded daily maximum rainfall was 131.9 mm in March 1990, which is a 1-in-45-yr event in the climate of 2021. The highest simulated daily maximum rainfall total in the UNSEEN ensemble for March in Montgomery is 210.8 mm, which can be used as an input to scenario planning for unprecedented events.
The second storyline applies to locations with recent rarity, where the worst event in recent memory has a high return period in today’s climate, even despite changes to the risk of extremes over time. Although we do not have any events in this quadrant, CMIP6 projections for the eastern United States shown in Fig. 4 that have been included for the purpose of showing a clearer overview of anticipated trends across the month of the year indicate a projected future decrease in the frequency of extreme precipitation during the hottest months (July and August), with a projected increase in the frequency of extreme precipitation in other months (Hoffman et al. 2023).
Projected change in maximum 1-day precipitation for all months of the year, expressed as a percent of precipitation in the preindustrial baseline of 1850–1900. Results are of the multimodel mean of 33 CMIP6 models for eastern North America, derived from the IPCC Atlas.
Citation: Bulletin of the American Meteorological Society 106, 3; 10.1175/BAMS-D-23-0297.1
While the risk of such a heavy rainfall event in July is decreasing, there is a nonzero probability of an extreme event greater than what was observed in this record. In fact, in the UNSEEN ensemble, the maximum daily simulated value for July precipitation in Warren County was 157.3 mm; and therefore, this ensemble could be used if people were interested in exploring and understanding unprecedented synthetic simulations of plausible record-breaking extremes.
4. Summary and discussion
To prepare for climate change, people cannot rely on recent experience. In the U.S. southeastern regions, we find evidence that temperature extremes have increased across four locations in the last 40 years, and the current risk is much higher than in 1981. As measured by the maximum daily total in each month, the risk of heavy precipitation has also increased in the last 40 years in most of the counties investigated in this study. The exception to this trend is the late summer months, which exhibit a drying trend in climate models (e.g., Hoffman et al. 2023).
We use a physics-based set of weather simulations to derive storylines from these changing frequencies of extreme events. In the context of four counties in the United States, we identify storylines for use in disaster preparedness and contingency planning, falling into the groups of “sitting ducks,” “recent rarity,” “living memory,” and “fading memory.” Unprecedented synthetic events from climate models can support contingency planning, particularly in the first scenario group, by providing alternative storylines of unprecedented synthetic events, and we offer examples from our case study counties in the U.S. southeastern region. Large UNSEEN ensembles of weather and climate models can be used to generate such unprecedented synthetic events as plausible extremes for use in disaster and climate change adaptation planning. In places without recent experience of extreme events, scenarios of perturbations around a historical extreme, such as the ensemble-boosting methods, could offer useful inputs to scenario planning.
Further research and improvements in model fidelity will be critical for applying these storylines in practice. Convection-permitting models are needed to resolve critical processes in many locations that are prone to extreme events, such as our Louisiana case study, which was excluded for a lack of fidelity with observational data. Convective-permitting models improve the resolution of extreme precipitation (Pal et al. 2019; Fosser et al. 2024) and can potentially capture dynamic processes that influence temperature variability. Model improvements can also reduce the need for bias correction; all bias-corrected results should be interrogated for physical plausibility before being used for operational decision-making.
Overall, we find that the risk of extreme heat is increasing rapidly in the southeastern United States, and maximum daily temperatures over the last 40 years do not indicate the types of extreme events that are now possible in the region. The locations we studied in this region fell into the sitting ducks category of risk. While we investigated only county-average temperatures, local variations in temperature within the county are likely to place certain residents at higher risk. For example, studies in Alabama and Kentucky highlight the urban heat island effect as a critical concern (Sabrin et al. 2022; Stone et al. 2023). The increased risk of extreme precipitation is also concerning across the case studies selected here.
In conclusion, while this study highlights shifts in return periods for temperature and precipitation extremes, a deeper analysis into the underlying mechanisms driving these changes lies outside the current scope. Future work could potentially address whether these shifts are primarily due to an increase in mean temperature, which could shift the entire distribution and reduce return periods, or if they result from a rise in the frequency of extreme anomalies exceeding climatological norms. Investigating these factors will provide a more detailed understanding of the dynamics influencing extreme event frequency, enabling a more comprehensive assessment of their implications for climate variability and risk management. Additionally, there is a potential value of exploring how the probability of exceeding extreme thresholds has evolved since 1981. Utilizing the UNSEEN ensemble could facilitate this analysis by establishing thresholds for extreme values based on observational data and examining changes in exceedance probabilities over time. This approach would not only enhance our understanding of climate variability and uncertainty but also allow for estimating the magnitude of exceedances.
We recommend that policymakers, disaster managers, and community groups consider unprecedented events as inputs for preparedness and scenario planning across the southeastern United States. Policies addressing extreme heat at the state and local levels are currently limited in many regions of the United States. (Tietjen et al. 2024). This lack of policy, combined with the high risk of extreme heat found in this research, indicates a critical need for further development of heat policies and planning in the United States. By considering unprecedented events, policymakers can deliver improved strategies to manage the cascading effects of multiple, overlapping events such as a heat wave that might coincide in close time frame with another severe climatic event, such as extreme precipitation. By considering a wider range of plausible events and rising extremes within the boundaries of science, more resilient strategies that consider not only climate science but also socioeconomic variables can be developed, leveraging lived experiences of prior events.
Additionally, considering the humidity contributions, it is crucial to further investigate the risk/impact of high temperatures on the human body especially in the southeastern United States, where the relatively high humidity can significantly influence heat index calculations. Future work could explore incorporating heat index considerations into scenario planning, highlighting the importance of accounting for both temperature and humidity in preparing for extreme heat events. This integration will be vital for effective disaster management and public health planning in the region.
Acknowledgments.
The authors express their sincere appreciation for the support received from the NASA cooperative grant titled “Today’s Risk of Extreme Events” (Agreement 80NSSC22K1706). Additionally, the authors acknowledge the contribution of the Tufts University High-Performance Computer Cluster (https://it.tufts.edu/high-performance-computing), which was used in the data processing of the research discussed in this paper.
Data availability statement.
The datasets analysed in the current study are available in the following locations: CHIRPS: https://www.chc.ucsb.edu/data/chirps; Daymet: https://daymet.ornl.gov/getdata; SEAS5: https://cds.climate.copernicus.eu/cdsapp#!/dataset/seasonal-original-single-levels?tab=overview. The authors can provide the code or materials used in this study upon reasonable request.
APPENDIX Historical Records Appendix
In Table A1, we present a comprehensive analysis of the maximum historical records and UNSEEN results for the four designated case study counties during the warmest and wettest months of the year. The table also includes return periods estimated in 2021 for what would have been a 1-in-100-yr occurrence in 1981. The UNSEEN results for the months of the highest accumulated precipitation and maximum temperature are derived from the results in the supplemental material.
Historical records and UNSEEN results for the four case study counties for the hottest (pink shading) and wettest (white shading) months of the year. Return periods are estimated in 2021 for two events: What would have been the 1-in-100-yr event in 1981, as well as the most extreme heat and rainfall event on record. The presence of a single (double) asterisk denotes that the model successfully passed (failed) 3 (2) of the fidelity test measures. Months that do not have any asterisk are those in which the model passed all four fidelity test measures.
References
Adger, W. N., 2003: Social capital, collective action, and adaptation to climate change. Econ. Geogr., 79, 387–404, https://doi.org/10.1111/j.1944-8287.2003.tb00220.x.
Anapalli, S. S., S. R. Pinnamaneni, D. K. Fisher, and K. N. Reddy, 2021: Vulnerabilities of irrigated and rainfed corn to climate change in a humid climate in the Lower Mississippi Delta. Climatic Change, 164, 5, https://doi.org/10.1007/s10584-021-02999-0.
Baulenas, E., G. Versteeg, M. Terrado, J. Mindlin, and D. Bojovic, 2023: Assembling the climate story: Use of storyline approaches in climate-related science. Global Challenges, 7, 2200183, https://doi.org/10.1002/gch2.202200183.
Birkland, T. A., 2009: Disasters, lessons learned, and fantasy documents. J. Contingencies Crisis Manage., 17, 146–156, https://doi.org/10.1111/j.1468-5973.2009.00575.x.
Bradshaw, C. D., and Coauthors, 2022: Unprecedented climate extremes in South Africa and implications for maize production. Environ. Res. Lett., 17, 084028, https://doi.org/10.1088/1748-9326/ac816d.
Carter, L., and Coauthors, 2018: Southeast. Impacts, Risks, and Adaptation in the United States: Fourth National Climate Assessment, Volume II, D. R. Reidmiller et al., Eds., U.S. Global Change Research Program, 1–66, https://doi.org/10.7930/NCA4.2018.CH19.
Casanueva, A., S. Herrera, M. Iturbide, S. Lange, M. Jury, A. Dosio, D. Maraun, and J. M. Gutiérrez, 2020: Testing bias adjustment methods for regional climate change applications under observational uncertainty and resolution mismatch. Atmos. Sci. Lett., 21, e978, https://doi.org/10.1002/asl.978.
Ciullo, A., O. Martius, E. Strobl, and D. N. Bresch, 2021: A framework for building climate storylines based on downward counterfactuals: The case of the European Union Solidarity fund. Climate Risk Manage, 33, 100349, https://doi.org/10.1016/j.crm.2021.100349.
Coughlan de Perez, E., H. Ganapathi, G. I. T. Masukwedza, G. Timothy, and T. Kelder, 2023: Potential for surprising heat and drought events in wheat-producing regions of USA and China. npj Climate Atmos. Sci., 6, 56, https://doi.org/10.1038/s41612-023-00361-y.
DeLeo, R. A., 2017: Anticipatory policymaking in global venues: Policy change, adaptation, and the UNFCCC. Futures, 92, 39–47, https://doi.org/10.1016/j.futures.2016.09.001.
Diffenbaugh, N. S., and Coauthors, 2017. Quantifying the influence of global warming on unprecedented extreme climate events. Proc. Natl. Acad. Sci. USA, 114, 4881–4886, https://doi.org/10.1073/pnas.1618082114.
Dunne, K. B. J., S. G. Dee, J. Reinders, S. E. Muñoz, and J. A. Nittrouer, 2022: Examining the impact of emissions scenario on lower Mississippi River flood hazard projections. Environ. Res. Commun., 4, 091001, https://doi.org/10.1088/2515-7620/ac8d53.
Fischer, E. M., and Coauthors, 2023: Storylines for unprecedented heatwaves based on ensemble boosting. Nat. Commun., 14, 4643, https://doi.org/10.1038/s41467-023-40112-4.
Fosser, G., and Coauthors, 2024: Convection-permitting climate models offer more certain extreme rainfall projections. npj Climate Atmos. Sci., 7, 51, https://doi.org/10.1038/s41612-024-00600-w.
Funk, C., and Coauthors, 2015: The climate hazards infrared precipitation with stations—A new environmental record for monitoring extremes. Sci. Data, 2, 150066, https://doi.org/10.1038/sdata.2015.66.
Gessner, C., E. M. Fischer, U. Beyerle, and R. Knutti, 2021: Very rare heat extremes: Quantifying and understanding using ensemble reinitialization. J. Climate, 34, 6619–6634, https://doi.org/10.1175/JCLI-D-20-0916.1.
Getirana, A. , and Coauthors, 2023: Climate and human impacts on hydrological processes and flood risk in southern Louisiana. Water Resour. Res., 59, e2022WR033238, https://doi.org/10.1029/2022WR033238.
Goulart, H. M. D., K. Van Der Wiel, C. Folberth, J. Balkovic, and B. Hurk, 2021: Storylines of weather-induced crop failure events under climate change. Earth Syst. Dyn., 12, 1503–1527, https://doi.org/10.5194/esd-12-1503-2021.
Gubernot, D. M., G. B. Anderson, and K. L. Hunting, 2015: Characterizing occupational heat-related mortality in the United States, 2000–2010: An analysis using the census of fatal occupational injuries database. Amer. J. Ind. Med., 58, 203–211, https://doi.org/10.1002/ajim.22381.
Haque, C. E., M.-U.-I. Choudhury, and M. S. Sikder, 2019: “Events and failures are our only means for making policy changes”: Learning in disaster and emergency management policies in Manitoba, Canada. Nat. Hazards, 98, 137–162, https://doi.org/10.1007/s11069-018-3485-7.
Hoffman, J. S., and Coauthors, 2023: Southeast. Fifth National Climate Assessment, A. R. Crimmins et al., Eds., U.S. Global Change Research Program, 1–65, https://doi.org/10.7930/NCA5.2023.CH22.
Hui, I., A. Zhao, B. E. Cain, and A. M. Driscoll, 2022: Baptism by wildfire? Wildfire experiences and public support for wildfire adaptation policies. Amer. Polit. Res., 50, 108–116, https://doi.org/10.1177/1532673X211023926.
Johnson, S. J., and Coauthors, 2019: SEAS5: The new ECMWF seasonal forecast system. Geosci. Model Dev., 12, 1087–1117, https://doi.org/10.5194/gmd-12-1087-2019.
Kelder, T., and Coauthors, 2020: Using UNSEEN trends to detect decadal changes in 100-year precipitation extremes. npj Climate Atmos. Sci., 3, 47, https://doi.org/10.1038/s41612-020-00149-4.
Kelder, T., N. Wanders, K. van der Wiel, T. I. Marjoribanks, L. J. Slater, R. I. Wilby, and C. Prudhomme, 2022: Interpreting extreme climate impacts from large ensemble simulations—Are they unseen or unrealistic? Environ. Res. Lett., 17, 044052, https://doi.org/10.1088/1748-9326/ac5cf4.
Liss, A., R. Wu, K. K. H. Chui, and E. N. Naumova, 2017: Heat-related hospitalizations in older adults: An amplified effect of the first seasonal heatwave. Sci. Rep., 7, 39581, https://doi.org/10.1038/srep39581.
Miao, Q., 2019: What affects government planning for climate change adaptation: Evidence from the U.S. states. Environ. Policy Governance, 29, 376–394, https://doi.org/10.1002/eet.1866.
Newman, R., and I. Noy, 2023: The global costs of extreme weather that are attributable to climate change. Nat. Commun., 14, 6103, https://doi.org/10.1038/s41467-023-41888-1.
Pal, S., H.-I. Chang, C. L. Castro, and F. Dominguez, 2019: Credibility of convection-permitting modeling to improve seasonal precipitation forecasting in the Southwestern United States. Front. Earth Sci., 7, 11, https://doi.org/10.3389/feart.2019.00011.
Sabrin, S., M. Karimi, and R. Nazari, 2022: Modeling heat island exposure and vulnerability utilizing earth observations and social drivers: A case study for Alabama, USA. Build. Environ., 226, 109686, https://doi.org/10.1016/j.buildenv.2022.109686.
Seneviratne, S. I., and Coauthors, 2021: Weather and climate extreme events in a changing climate. Climate Change 2021: The Physical Science Basis, V. Masson-Delmotte et al., Eds., Cambridge University Press, 1513–1766, https://doi.org/10.1017/9781009157896.013.
Shao, W., S. Xian, N. Lin, H. Kunreuther, N. Jackson, and K. Goidel, 2017: Understanding the effects of past flood events and perceived and estimated flood risks on individuals’ voluntary flood insurance purchase behavior. Water Res., 108, 391–400, https://doi.org/10.1016/j.watres.2016.11.021.
Shepherd, T. G., and Coauthors, 2018: Storylines: An alternative approach to representing uncertainty in physical aspects of climate change. Climatic Change, 151, 555–571, https://doi.org/10.1007/s10584-018-2317-9.
Sillmann, J., T. G. Shepherd, B. van den Hurk, W. Hazeleger, O. Martius, J. Slingo, and J. Zscheischler, 2021: Event‐based storylines to address climate risk. Earth’s Future, 9, e2020EF001783, https://doi.org/10.1029/2020EF001783.
Stone, B., Jr., K. Lanza, E. Mallen, J. Vargo, and A. Russell, 2023: Urban heat management in Louisville, Kentucky: A framework for climate adaptation planning. J. Plann. Educ. Res., 43, 346–358, https://doi.org/10.1177/0739456X19879214.
Sy, S., F. Madonna, F. Serva, I. Diallo, and B. Quesada, 2024: Assessment of NA-CORDEX regional climate models, reanalysis and in situ gridded-observational data sets against the U.S. Climate Reference Network. Int. J. Climatol., 44, 305–327, https://doi.org/10.1002/joc.8331.
Thompson, V., N. J. Dunstone, A. A. Scaife, D. M. Smith, J. M. Slingo, S. Brown, and S. E. Belcher, 2017: High risk of unprecedented UK rainfall in the current climate. Nat. Commun., 8, 107, https://doi.org/10.1038/s41467-017-00275-3.
Thornton, M. M., R. Shrestha, Y. Wei, P. E. Thornton, and S.-C. Kao, 2020: Daymet: Daily surface weather data on a 1-km grid for North America, version 4. ORNL Distributed Active Archive Center, accessed 26 March 2023, https://doi.org/10.3334/ORNLDAAC/1840.
Tietjen, B., J. Clark, and E. Coughlan de Perez, 2024: Progress and gaps in U.S. Adaptation policy at the local level. Global Environ. Change, 87, 102882, https://doi.org/10.1016/j.gloenvcha.2024.102882.
USGCRP, 2023: Fifth National Climate Assessment. A. R. Crimmins et al., Eds., U.S. Global Change Research Program, 1834 pp., https://doi.org/10.7930/NCA5.2023.
van den Berg, P., D. Scholten, J. Schachter, and K. Blok, 2021: Updating scenarios: A multi-layer framework for structurally incorporating new information and uncertainties into scenarios. Futures, 130, 102751, https://doi.org/10.1016/j.futures.2021.102751.
van den Hurk, B. J. J. M., and Coauthors, 2023: Climate impact storylines for assessing socio-economic responses to remote events. Climate Risk Manage., 40, 100500, https://doi.org/10.1016/j.crm.2023.100500.
van der Wiel, K., G. Lenderink, and H. de Vries, 2021: Physical storylines of future European drought events like 2018 based on ensemble climate modelling. Wea. Climate Extremes, 33, 100350, https://doi.org/10.1016/j.wace.2021.100350.
Varum, C. A., and C. Melo, 2010: Directions in scenario planning literature—A review of the past decades. Futures, 42, 355–369, https://doi.org/10.1016/j.futures.2009.11.021.
Venner, M., and J. Zamurs, 2012: Increased maintenance costs of extreme weather events: Preparing for climate change adaptation. Transp. Res. Rec., 2292, 20–28, https://doi.org/10.3141/2292-03.
Young, H. R., T. G. Shepherd, J. Acidri, R. J. Cornforth, C. Petty, J. Seaman, and L. C. Todman, 2021: Storylines for decision-making: Climate and food security in Namibia. Climate Dev., 13, 515–528, https://doi.org/10.1080/17565529.2020.1808438.
Zappa, G., 2019: Regional climate impacts of future changes in the mid–latitude atmospheric circulation: A storyline view. Curr. Climate Change Rep., 5, 358–371, https://doi.org/10.1007/s40641-019-00146-7.
