Exploratory Data Analysis of Long-Term Oak Ridge Reservation Meteorological Data for Extreme Weather Event Discovery

Exploratory Data Analysis of Long-Term Oak Ridge Reservation Meteorological Data for Extreme Weather Event Discovery | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Exploratory Data Analysis of Long-Term Oak Ridge Reservation Meteorological Data for Extreme Weather Event Discovery Ziwei Liu, Kevin Birdwell, Morgan Steckler, Xiao-Ying Yu This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8855721/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 7 You are reading this latest preprint version Abstract Big data challenges are commonly encountered when conducting radiological and chemical hazard analysis to support operations of nuclear facilities in the Department of Energy (DOE). Extreme weather significantly influences environmental conditions and site operations, underscoring the need to incorporate accurate weather characterization in hazard assessments and safety planning. We applied exploratory data analysis (EDA) to six years (2017–2022) of high-resolution meteorological observations from six towers at the Oak Ridge Reservation (ORR). In addition to statistical methods to capture data distributions, EDA were used to examine trends using the Mann–Kendall test and Sen’s slope estimation. The results reveal asymmetric seasonal trends, namely warming in summer and cooling in winter. Clustering analysis was employed to interpret underlying patterns, identifying frequent co-occurrence of high temperature and high humidity during summer. Extreme weather events were further defined using feature-specific thresholds (e.g., temperature–moisture hazards, wind chill events, high-wind conditions), informed by regulatory guidelines and clustering outcomes. Our results show that EDA approaches can effectively assess big, long-term meteorological datasets and extract actionable information for site operations, particularly in relation to potentially hazardous extreme events. This study demonstrates that EDA is important in extreme event classification before applying machine learning in modeling. Earth and environmental sciences/Climate sciences Physical sciences/Engineering Earth and environmental sciences/Environmental sciences Earth and environmental sciences/Natural hazards Exploratory Data Analysis (EDA) Extreme Weather Events Trend Analysis Unsupervised Clustering Multivariate Time Series Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Figure 11 Figure 12 Introduction Extreme weather events significantly impact environmental conditions and site operations. Oak Ridge National Laboratory (ORNL) accommodates large-scale scientific facilities and experimental programs that are at times highly sensitive to the impacts of extreme weather events. Understanding long-term trends and accurately identifying extreme events are crucial for effective risk mitigation strategies. Due to its geographical setting, ORNL experiences diverse and highly variable weather conditions, necessitating systematic meteorological trend assessment and extreme weather detection to address challenges in reliable forecasting and operational management. For instance, sudden temperature fluctuations, humidity variations, and wind extremes can affect experimental conditions, facility Heating, Ventilation, and Air Conditioning (HVAC) system performance, energy consumption, and overall infrastructure stability. An in-depth understanding of these meteorological factors and their long-term trends can enhance risk management practices, support infrastructure resilience, and contribute to the formulation of adaptive operational strategies 1 . To address these requirements, this work employs a detailed analysis of multi-year meteorological data at six distinct towers across multiple measurement heights. Through robust data quality assessments, comprehensive temporal and spatial trend analyses, and clustering methods, we aim to comprehensively characterize and interpret the underlying patterns in observed meteorological data. Specifically, the analysis uses statistical techniques, such as Sen's slope estimator and Mann-Kendall tests, and precise detection of extreme weather events through feature-based clustering. By synthesizing these methods, we provide actionable insights into meteorological phenomena relevant to ORNL's operational contexts. Our findings highlight crucial patterns of variability, both spatially and temporally, enabling refined risk assessments, targeted preventive measures, and informed decision-making. Ultimately, this analysis supports enhanced preparedness against extreme weather impacts, safeguards research integrity, and promotes operational resilience at the ORNL site. Data and Methods Site Description ORNL spans a varied topographic and forested landscape within the Ridge-and-Valley region of the Appalachian Mountains and plays a critical role in energy research, national security, and environmental sustainability. To support research and operational needs, ORNL and Y-12 maintain a high-resolution meteorological monitoring network consisting of eight meteorological towers and two wind profilers, strategically distributed across the site. The tower locations used for the purposes of this work are indicated by red markers in Fig. 1 . Most of the selected sites are equipped with multi-level sensors, with measurement heights ranging from 2 m to 60 m above ground level, capturing key meteorological variables such as air temperature, wind speed and direction, humidity, and solar radiation at 15-minute (min) intervals. Data Description This study utilizes six years (2017–2022) of meteorological data collected from these towers to characterize site-specific weather patterns and support trend analysis along with extreme event detection. The high-resolution data set consists of 15-min measurement intervals and includes data collected from ORNL and Y-12 consisting of air temperature, wind speed, wind direction, relative humidity, absolute humidity, solar radiation, and precipitation. The present study focuses on four key parameters: air temperature, wind speed, relative humidity, and absolute humidity, which were standardized to the following units: (1) temperature in degrees Celsius (°C), (2) wind speed in miles per hour (mph), (3) relative humidity in percent (%), and (4) absolute humidity in grams per cubic meter (g/m³). To ensure quality and consistency, raw 15-min time series were processed to address missing values, outliers, and sensor drift 2 . Beyond standard quality control, a multi-tower rolling window method was developed for robust anomaly detection. The approach compares concurrent sensor readings of the same variable across multiple towers within a specified temporal window. Appropriately cleaned data were then resampled or smoothed using running averages to enable analysis at broader temporal scales (e.g., hourly and daily). These quality-assured time series form the foundation for all subsequent exploratory data analysis, trend evaluation, and extreme event characterization. Summary statistics, including daily, weekly, and monthly maxima and minima, were extracted to support EDA and trend detection. The resulting processed features form the basis for future machine learning tasks aimed at understanding the temporal structure of meteorological phenomena and for evaluating their operational implications regarding building performance and site resilience at ORNL. Trend Analysis Two methods for trend analysis were adopted: namely Sen’s slope and the Mann–Kendall (MK) test, to determine the representative trend of the measured parameters. The MK test is a non-parametric statistical test that can be used for detecting time series trends which allow for identification of trends based primarily on ranks without the need for specifying linearity 3 . Hence, it offers robustness to non-normality and cleans data with missing values. The MK hypothesis includes the null hypothesis, where H0 refers to either a sample (i.e., measured meteorological parameter) or the independent random variables. The subsamples of each variable are independent and identically distributed over years 3 . In the null hypothesis (H0) of the MK test, data coming from a population with independent realizations are not significantly different (i.e., no trend). If a calculated p-value for a trend test is smaller than a significance level (e.g., 0.05), the null hypothesis is rejected (i.e., the trend is significant). Additionally, the MK method is well-known for assessing the significance of trends in hydroclimatic time series data, such as rainfall, temperature, and streamflow. Clustering Unsupervised clustering techniques were employed to identify desired patterns. Clustering is a form of exploratory data analysis that partitions observations into groups (clusters) such that data points within the same cluster are more like each other than those in other clusters. We used K-means clustering to explore inherent structures of multi-dimensional meteorological data 4 primarily to take advantage of its efficiency within large datasets and to obtain spherical clusters of similar size. Results Rolling Window A variable-specific outlier detection strategy was implemented to ensure data integrity across the six selected meteorological towers. For our key variables: air temperature (TempC), absolute humidity (AbsHum), and peak wind speed (PkWSpdMph), a multi-tower rolling-window approach was applied as shown in Table 1 . Table 1 Outlier detection by the multi-tower rolling window method. Variable Method Window Std Dev 1 Threshold TempC Multi-tower 1 day 3 AbsHum Multi-tower 6 hours 3 WSpdMph Multi-tower 6 hours 4 PkWSpdMph Multi-tower 6 hours 4 1 Std Dev: standard deviation This method identified data points that deviated beyond a specified number of standard deviations from the multi-tower mean within each rolling window. Air temperature anomalies were detected using a 1-day window with a 3-sigma threshold, absolute humidity outliers were identified using a 6-hour (h) window and a 3-sigma threshold, and peak wind speed anomalies were captured using a 6-h window with a 4-sigma threshold. By combining automated multi-tower anomaly detection with targeted manual inspection, the quality-control framework effectively enhanced data consistency and reliability across the six towers, removing sensor biases and spurious outliers as well as providing a robust foundation for the subsequent analyses of temporal trends and extreme weather events. Data Structure and Feature Characterization Figures S1 –S2 collectively illustrate the multi-year meteorological variability observed across six towers between 2017 and 2022. Temperatures generally ranged from − 25°C to 35°C with subtle vertical gradients and moderate interannual stability, although the most recent years suggested slightly higher medians and included some anomalous cold outliers. Tower A records highlighted seasonal cycles in temperature, absolute humidity, and relative humidity, alongside variable wind speed and episodic precipitation, reflecting both seasonal and synoptic-scale weather influences. Seasonal correlation matrices further reveal dynamic couplings among variables, for instance, stronger negative correlations between wind speed and temperature in winter and positive temperature–radiation links in spring and summer, which underscore the value of high-resolution, seasonally disaggregated data for characterizing microclimatic variability and extreme events. However, conventional statistical methods only reveal the overall distribution and do not capture temporal trends across years. The MK test and Sen’s slope analyses were applied at multiple time scales, as described in Section 3.3. MK Test and Sen’s Slope Analysis Figure 2 presents Sen’s slope estimates for minimum, maximum, mean, and median temperature across International Organization for Standardization (ISO) weeks, highlighting seasonal patterns in warming and cooling trends. ISO weeks follow the ISO-8601 calendar system, where weeks start on Monday and week 1 is defined as the week containing the first Thursday of the year 5 . Red boxes indicate the summer period (weeks 22–35), during which all four-temperature metrics generally exhibit positive slopes, reflecting a tendency toward summer warming. This effect is most consistent for mean and median temperatures, with smaller but still positive trends in minimum and maximum temperatures. In contrast, blue boxes highlight the winter period (weeks 1–8 and 48–52), which are dominated by negative slopes across all metrics, indicative of wintertime cooling trends. These opposing seasonal signals underscore the asymmetric nature of recent temperature changes: warmer summers and colder winters. Figure 3 presents Sen’s slope estimates for minimum, maximum, mean, and median absolute humidity (AbsHum) across weeks of a whole year, highlighting the seasonal variability of moisture trends. The red boxes denote the summer period (weeks 22–25), during which positive slopes dominate among most metrics, particularly for minimum and median absolute humidity, indicating a moistening trend in summer. Maximum absolute humidity also exhibits modest positive slopes in this period, suggesting more frequent or intense high-moisture events. In contrast, several weeks outside the summer period, especially in late autumn and winter, display near-zero or negative slopes, implying stable or declining moisture levels. The consistent summertime increases in absolute humidity, when coupled with concurrent temperature rises, highlight the potential for compounding heat–moisture stress during warm-season extremes. Figure 4 displays Sen’s slope estimates for minimum, maximum, mean, and median peak wind speed (PkWSpdMph) across weeks of a year, highlighting contrasting seasonal trends. The red boxes mark the summer period (weeks 22–35), during which Sen’s slopes are generally near zero or slightly positive for most metrics, suggesting relatively stable or modestly increasing peak wind speeds in mid-summer. In contrast, the blue boxes highlight the winter period (weeks 1–8 and 48–52), when positive slopes dominate among nearly all metrics, most notably for maximum and mean peak wind speeds, which indicate a tendency toward stronger winter winds. Seasonal asymmetry suggests that while peak wind speeds remain relatively steady in summer, the winter season is characterized by a strengthening of wind extremes, which, when coupled with corresponding cooling trends, can exacerbate wind chill effects, increasing operational and safety risks in exposed environments. Figure 5 (a) illustrates spatial maps of weekly temperature trends for a representative summer period (week 30), derived from the Mann–Kendall test and Sen’s slope estimator among all meteorological towers and including sensors at all heights. During this week, nearly all sensor locations display positive Sen’s slope values (red squares), indicating a warming trend. Larger marker sizes highlight locations with more pronounced increases at several towers and elevated measurement heights. Absolute humidity trends for the same period in Fig. 5 (b) reveal a consistent pattern of positive slopes among all reporting stations. This mid-summer moistening coupled with temperature increases suggests a heightened potential for heat–moisture stress. Such conditions may intensify thermal discomfort, reduce evaporative cooling efficiency, and elevate heat index values, particularly in above-canopy environments involving tall-equipment operations. Figure 6 (a) depicts spatial patterns of weekly temperature trends for a representative winter period (week 3), derived from the Mann–Kendall test and Sen’s slope estimator among all towers and including sensors at all heights. Nearly all sensors exhibit strongly negative Sen’s slope values (blue squares), indicating widespread cooling trends. In contrast, Fig. 6 (b) shows peak wind speed trends for the same week, which are positive (red–orange circles) across most stations. The corresponding patterns of intensified winds and declining temperatures are consistent with cold air outbreak conditions, where strong cold air advection and elevated winds enhance heat loss and exacerbate wind chill effects. Such conditions pose heightened thermal stress risks, especially in above-canopy environments involving tall-equipment operations and wintertime site safety planning. Weekly Sen’s slope trends for solar radiation, wind-direction variability, and vertical wind speed are also evaluated. The results show little evidence of sustained long-term trends. These results are provided in the Supplementary Information ( Figs. S3–S5 ). Monthly spatial maps for temperature are included in the Supplementary Information to provide a broader temporal perspective ( Fig. S6 ). Clustering Analysis Figure 7 illustrates the K-means clustering of temperature and absolute humidity. Among them, Cluster 2 (green) captures the hottest and most humid conditions; Cluster 1 (yellow) corresponds to the coldest and driest conditions; Cluster 0 (blue) represents cool–moderately humid conditions; Cluster 3 (orange) reflects warm–moderately humid conditions. The centroids and ranges of temperature–humidity clusters are shown in Table 2 . Table 2 Cluster centroids and ranges of temperature–humidity clusters. Cluster Temp (°C) Min / Mean / Max AbsHum (g/m³) Min / Mean / Max 0 5.50 / 11.56 / 25.40 1.50 / 7.03 / 11.10 1 -16.00 / 2.39 / 10.40 0.80 / 4.05 / 7.20 2 17.90 / 24.25 / 34.60 10.50 / 17.04 / 25.60 3 11.80 / 18.93 / 32.50 3.70 / 11.68 / 15.60 Figure 8 presents that Cluster 2 is dominated by summer (69%), confirming that extreme hot–humid conditions occur in summer. Cluster 1 is winter-dominated (65%), representing cold–dry extremes. Clusters 0 and 3 serve as transitional regimes, linking winter to spring/fall and summer to spring/fall, respectively. The hot and humid conditions captured in Cluster 2 raise particular concern. The cluster centroid indicates a mean temperature of 24.3°C (maximum 34.6°C) and a mean absolute humidity of 17.4 g/m³ (maximum 25.6 g/m³). For comparison, the National Weather Service (NWS) Heat Index 6 shows that caution is warranted once air temperature reaches 80°F (26.7°C) with a relative humidity of 40%, which corresponds to an absolute humidity of about 10 g/m³ It is worth noting that these conditions are met for about 9 to 10% of all hours at ORNL. For instance, 857 such hours out of 8760 occurred in 2022. At the upper end of Cluster 2, the combination of 34.6°C and 25.6 g/m³ translates to a relative humidity of roughly 66%, yielding a heat index of about 116°F, a level categorized by the NWS as Danger, where the likelihood of heat-related disorders under prolonged exposure or strenuous activity becomes severe. Figure 9 shows the cluster results of temperature and windspeed. Cluster 1 (yellow) captures high-wind conditions across the full temperature range, with windspeed exceeding 20 mph and reaching up to 60 mph; Cluster 0 (blue) represents warm, low-to-moderate wind regimes; Cluster 2 (green) corresponds to hot, moderate-wind conditions; and Cluster 3 (orange) reflects cold, moderate-wind conditions. The clustering highlights that Cluster 1 is distinguished primarily by extreme windspeed rather than temperature. The centroids and ranges of temperature–windspeed clusters are shown in Table 3 . Table 3 Cluster centroids and ranges of temperature–windspeed clusters. Cluster Temp (°C) Min / Mean / Max Windspeed (mph) Min / Mean / Max 0 11.10 / 20.11 / 33.30 0.00 / 4.40 / 9.80 1 -10.00 / 13.00 / 32.80 12.70 / 20.26 / 75.00 2 13.10 / 24.00 / 34.60 5.10 / 11.44 / 23.70 3 -15.00 / 5.89 / 14.50 0.00 / 6.78 / 18.20 Figure 10 indicates that Cluster 1 is dominated by spring (43%) and winter (35%), presenting strong wind events occurring across a wide temperature range but more frequent in colder and transitional seasons. Cluster 3 is winter-dominated (52%), reflecting cold–moderate wind regimes. Cluster 0 occurs mainly in summer (44%) and fall (29%), while Cluster 2 peaks in summer (49%), representing hot–moderate-to-strong wind conditions. Cluster 1 represents cold and windy conditions that are prone to significant wind chill effects. The minimum values in this cluster reach − 10°C (14°F) for temperature and 12.7 mph for wind speed. According to the NWS Wind Chill Index 7 , this combination corresponds to a wind chill temperature of approximately − 33°C (–27°F). At this level, exposed skin can develop frostbite in 10–30 min. Such conditions highlight the elevated risk of cold-related health hazards during strong wind episodes in winter. Extreme Weather Events Extreme weather event thresholds were determined by combining long-term trend analysis with criteria established by the U.S. National Weather Service (USNWS). Five event types were defined to represent key meteorological hazards relevant to the study region (Table 4 ). Table 4 Extreme weather threshold and associated flags. Extreme Event Flag(s) Threshold Temperature–moisture hazard E1 Temp > 24°C and AbsHum > 20 g/m³ Wind chill event E2 Temp ≤ 4.8°C and PkWind ≥ 3 mph Low temperature event E3_1 Temp < 0°C E3_2 Temp < -5°C E3_3 Temp 25 mph Low wind speed event E5_1 PkWind < 2.0 mph E5_2 PkWind < 1.0 mph E5_3 PkWind 24°C) and high absolute humidity (> 20 g m⁻³), conditions known to elevate the heat index and increase the risk of heat-related illness. Wind chill events (E2) were identified when air temperature was ≤ 4.8°C and the wind speed was ≥ 3 mph, reflecting conditions under which cold stress and frostbite risk intensify. Low-temperature events (E3) included three severity levels (< 0°C, < -5°C, and < -10°C) to represent progressively stronger cold exposure. High-wind events (E4) were defined by peak wind speeds exceeding 25 mph, corresponding to the operational threshold for suspending outdoor activities at ORNL. Finally, low-wind events (E5) were characterized by wind speeds below 2.0 mph, with additional levels at 1.0 mph and 0.5 mph, conditions of particular concern for emergency release scenarios under pollutant stagnation conditions. This standardized framework ensures consistent identification and classification of meteorological conditions of concern within the tower network at ORR, with multiple thresholds for low temperature and low wind speed events capturing varying levels of severity. After determining thresholds for each type of extreme event, their occurrence frequency and duration were analyzed. Because the towers differed in measurement height and sensor configuration, tower-specific variable mappings were applied for event detection (Table 5 ). Table 5 Sensor variables and measurement heights used for event detection across towers. Tower Temperature Absolute Humidity Peak Wind Speed A 30m 15m 30m B 30m – 30m D 35m 15m 35m F 10m 10m 10m S 25m – 25m Y 33m – 33m For each tower, events were first detected independently using tower-specific variable mappings and then merged across towers using a union approach, ensuring that overlapping or adjacent intervals were not double-counted. Event days were counted on a calendar-day basis (i.e., multiple events or multiple towers on the same day were counted once). Figures 11 and 12 summarize the monthly occurrence of each event and highlight the variation in event duration through its mean, minimum, and maximum values. As shown in Fig. 11 , distinct seasonal patterns emerge. Temperature–moisture hazards (E1) are concentrated in the summer months (June – August), with an average of 6 − 14 days per month; the maximum single-event duration in July exceeded 1,000 min, highlighting potential heat stress during sustained hot and humid episodes. Wind chill events (E2) dominate the winter season (November–January), especially December, with more than 20 days per month and maximum durations surpassing 5,000 min, reflecting persistent cold and windy conditions. Low temperature events (E3) show a similar winter concentration, with extremes occasionally exceeding 10,000 minutes, posing risks to infrastructure and operations. In contrast, high wind events (E4) are distributed more broadly, peaking in spring and autumn (March–May). Although the monthly frequencies are lower (~ 15–19 days), their repeated occurrence implies potential operational disruptions. Finally, low wind speed events (E5) are most frequent in summer and early autumn (July–September), with up to 30 event days per month and durations exceeding 3,000 min, emphasizing the possibility of exacerbating poor air quality conditions. Overall, the results demonstrate that each event type exhibits clear seasonal peaks and distinctive persistence characteristics. These findings underscore the importance of tailoring risk mitigation and preparedness strategies to the seasonal hazard profiles observed across the tower network. Wind Patterns related to Extreme Events Wind pattern analysis was also conducted to further interpret the synoptic mechanisms behind observed extreme meteorological events in ORNL. Results show that low-wind conditions, often associated with high pollutant concentration potential, were highly correlated with daytime terrain-driven thermal winds (40%) and cross-valley winds from the west (27%). These conditions reflect the complex interactions of air flow with the Great Valley’s topography 8 , where weak thermally induced circulations and lateral deflection of airflow limit horizontal dispersion and promote pollutant buildup. However, unstable atmospheric conditions sometimes counteract these effects. Wind chill events were significantly associated with cross-valley winds from the northwest, north, and north-northeast sectors, accounting for 19–24% of cases overall and increasing to 35–50% during the cool season (October–March). This pattern suggests that cold air masses from higher latitudes often cross local terrain during winter, amplifying cold stress. Similarly, low-temperature events (below 0°C) were closely linked to NNE–NE and WNW–NNW winds, 18% and 17% overall (rising to 34–36% in the cool season), indicating an association with both cold-air drainage along the valley axis as well as synoptic-scale cold air advection episodes. In contrast, high temperature and moisture hazards were observed mainly during the warm months (April–September), comprising 14% of warm-season observations, and were dominated by southwest-to-northeast up-valley winds that transported warm, humid air from the Gulf of America/Mexico. These up-valley flows, often deflected from the west into the Great Valley, may enhance the accumulation of moisture within the Great Valley, contributing to thermal discomfort and elevated cooling demand, especially under low wind conditions that are typical of summer. Overall, wind pattern analysis highlights the critical role of valley-modulated airflow in shaping ORNL and ORR local climate, pollutant dispersion, and thermal risk. Integrating these terrain–wind–event relationships into predictive and optimization frameworks, such as HVAC control and environmental management, can improve resilience, comfort, and energy efficiency under site-specific meteorological conditions. Discussion The combination of high-risk temperature–moisture conditions in summer and significant wind chill effects in winter presents an environmental challenge to ORNL’s operational resilience. Our analyses indicate that the laboratory experiences frequent weather extremes at both ends of the temperature spectrum. In summer, higher temperatures coupled with greater humidity elevate the risks of heat-related illness, equipment overheating, and reduced worker productivity 9 . In winter, outbreaks of cold, windy weather increase risks of hypothermia, ice formation, and weather-related disruptions 10 . Effective management helps to mitigate the effects of environmental extremes that can endanger staff well-being and interrupt mission-critical operations. Our climatological records show trends consistent with broader patterns but also capture the recent intensification of extremes. For instance, while the mean annual temperature at Oak Ridge rose by roughly 1.3°C between the 1970s and 2000s 11 , our results highlight that the most consequential changes are occurring in the frequency and magnitude of extreme events—heat index surges and wind chill drops—rather than in gradual shifts of mean temperature. Our analysis provides site-specific evidence of these regional trends, providing more insight of data trends that were otherwise undiscovered using EDA approaches. Such findings are important to improve predictability of extreme weather events that may impact site operations. Our results validate that EDA can be a useful approach for assessing data trend indicative of conditions that may impair operation and potentially contribute to the DOE standards. Specifically, using statistical and EDA techniques in treating at least five years of meteorological data as a novel approach, which is required by DOE-STD-3009-2014 12 , will improve risk analysis efficiency significantly for safety designs, strategies, and operations and serve the DOE-STD-1189-2016 13 objectives by providing site relevant applications following DOE-HDBK-1224-2024 14 . The ability to analyze extreme weather patterns that have impact on site operation is important to conform to the DOE-STD-1020-2016 15 . Clustering analysis proved especially useful for identifying compound extreme events, such as joint high-temperature and high-humidity episodes or low-temperature and high-wind conditions, that are directly relevant to safety planning. Based on these findings, climate adaptation planning at ORNL should integrate both observed trends and projected changes into its risk management strategies. Practical actions may include updating heat-stress management protocols (for example, using heat index thresholds to guide work/rest cycles on hot days) and strengthening cooling infrastructure in critical facilities. Winter preparations should also be enhanced by insulating buildings and pipes against deep freezes, maintaining vegetation to limit wind damage, and ensuring contingency plans for essential operations during severe cold and wind events. These measures align with broader DOE resilience initiatives that address site-specific climate. For ORNL, the data clearly justify proactive measures: hot, humid summers and more anomalous winter events present tangible risks to a laboratory of this scale and importance. Overall, exploratory data analysis provides a detailed picture of how local climate conditions can affect site daily operations. The subtropical humid climate of Oak Ridge and the surrounding region results in high heat stress potential, while winters can exhibit highly variable conditions sometimes associated with strong winds that heighten cold stress. Thus, local climate sometimes produces high-risk weather conditions, dangerous heat levels in summer and significant wind chills in winter. When interpreted alongside regional climate data, the results suggest that ORNL is experiencing seasonal climate extremes that are typical of regional atmospheric dynamics on the northern fringe of the subtropical climate zone (Koppen Climate – Cfa), yielding differing extreme events in all seasons. The findings are consistent with other analyses in the Southeast and across DOE facilities, confirming that even within a short time span, statistically significant annual-scale climate shifts can be detected when high-quality data are used. Conclusion This study demonstrates the value of combining trend analysis, clustering, and threshold-based event definitions. Together, these methods allow us to quantify subtle changes, distinguish normal from extreme conditions, and identify threshold exceedances that have real operational impacts. For a research campus like ORNL, such information is essential for evidence-based decisions on safety and infrastructure. Continued high-resolution meteorological monitoring will be important to track how these trends evolve, whether winter cooling persists or reverses, and whether heat extremes worsen steadily or accelerate. These insights can inform ORNL’s climate adaptation and mitigation strategies, ensuring that the laboratory remains safe, resilient, and well prepared for future meteorological challenges. By situating ORNL’s data within the broader scientific context of climate variations, our findings highlight both local manifestations of weather extremes and their regional significance. The agreement between our findings and established meteorological trends strengthens confidence in these interpretations and reinforces the need for continued observation and analysis. In summary, EDA is a viable approach to capture key aspects of broader climate dynamics and provides actionable knowledge for improving environmental safety using a six-year meteorological record that is fully qualified to support ORNL’s operation. Declarations Competing interests The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Funding Research efforts were supported by the Nuclear Safety Research and Development (NSR&D) program sponsored by National Nuclear Security Administration (NNSA) Office of Environment, Health, Safety and Security (EHSS) of the U.S. Department of Energy (DOE). Author Contribution Conceptualization and design: X.Y.Y and Z.L.; Data curation: M.S., K.B. and X.Y.Y.; Investigation: Z.L., M.S. and X.Y.Y.; Formal analysis: Z.L.; Validation: M.S. and X.Y.Y.; Writing—original draft: Z.L., K.B., and X.Y.Y.; Writing—review and editing: Z.L., K.B., and X.Y.Y.; Visualization: Z.L. and M.S.; Supervision and funding acquisition: X.Y.Y... All authors have read and agreed to the published version of the manuscript. Acknowledgements This manuscript has been authored by UT-Battelle, LLC, under contract DE-AC05-00OR22725 with the US DOE. The US government retains and the publisher, by accepting the article for publication, acknowledges that the US government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript, or allow others to do so, for US government purposes. DOE will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan ( https://www.energy.gov/doe-public-access-plan ). Data Availability The meteorological dataset used in this study is openly available on Zenodo at: [https://zenodo.org/records/15171289](https:/zenodo.org/records/15171289) References Ayadi, R., Forouheshfar, Y. & Moghadas, O. Enhancing system resilience to climate change through artificial intelligence: a systematic literature review. Front Clim 7 , (2025). Steckler, M. R., Birdwell, K. R., Xu, H., YU, X. Y. & Descriptor High Temporal Resolution Meteorological Data at Oak Ridge Reservation (ORR-HiResMet). IEEE Data Descriptions . 2 , 118–124 (2025). Zhou, H., Ren, H., Royer, P., Hou, H. & Yu, X. Y. Big Data Analytics for Long-Term Meteorological Observations at Hanford Site. Atmosphere 13 , 136 (2022). Sanhudo, L., Rodrigues, J. & Filho, Ê. V. Multivariate time series clustering and forecasting for building energy analysis: Application to weather data quality control. J. Building Eng. 35 , 101996 (2021). ISO - ISO 8601 — Date. and time format. ISO (2017). https://www.iso.org/iso-8601-date-and-time-format.html US Department of Commerce, N. What is the heat index? https://www.weather.gov/ama/heatindex US Department of Commerce, N. Understanding Wind Chill. https://www.weather.gov/safety/cold-wind-chill-chart Energy, U.-B. L. for the U. D. of. Wind Regimes in Complex Terrain of the Great Valley of Eastern Tennessee… ORNL. https://www.ornl.gov/publication/wind-regimes-complex-terrain-great-valley-eastern-tennessee(2011). Habibi, P. et al. Climate change and heat stress resilient outdoor workers: findings from systematic literature review. BMC Public. Health . 24 , 1711 (2024). CDC & Preventing Hypothermia Winter Weather (2024). https://www.cdc.gov/winter-weather/prevention/index.html Appendix B_Climate. Overview of the Oak Ridge Area.pdf. Preparation of Nonreactor Nuclear Facility Documented Safety Analysis. (Invoked). https://www.standards.doe.gov/standards-documents/3000/3009-astd-2014 Integration of Safety into the Design Process. (Invoked). https://www.standards.doe.gov/standards-documents/1100/1189-astd-2016 Hazard and Accident Analysis Handbook. https://www.standards.doe.gov/standards-documents/1200/1224-bhdbk-2024 Natural Phenomena Hazards Analysis and Design Criteria for DOE Facilities. (Invoked). https://www.standards.doe.gov/standards-documents/1000/1020-astd-2016 Additional Declarations No competing interests reported. Supplementary Files EDAORRSI02102026.docx Cite Share Download PDF Status: Under Review Version 1 posted Reviewers agreed at journal 20 Mar, 2026 Reviewers agreed at journal 09 Mar, 2026 Reviewers invited by journal 09 Mar, 2026 Editor invited by journal 17 Feb, 2026 Editor assigned by journal 12 Feb, 2026 Submission checks completed at journal 12 Feb, 2026 First submitted to journal 11 Feb, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8855721","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":603314746,"identity":"9b1da360-9795-40e9-8cc0-3d325995e3f3","order_by":0,"name":"Ziwei Liu","email":"","orcid":"","institution":"Oak Ridge National Laboratory","correspondingAuthor":false,"prefix":"","firstName":"Ziwei","middleName":"","lastName":"Liu","suffix":""},{"id":603314747,"identity":"b6a6c9c7-465c-45b1-992d-a44efcb3bf33","order_by":1,"name":"Kevin Birdwell","email":"","orcid":"","institution":"Oak Ridge National Laboratory","correspondingAuthor":false,"prefix":"","firstName":"Kevin","middleName":"","lastName":"Birdwell","suffix":""},{"id":603314752,"identity":"05b69502-d7b4-4449-a3e2-6e74544122bb","order_by":2,"name":"Morgan Steckler","email":"","orcid":"","institution":"Oak Ridge National Laboratory","correspondingAuthor":false,"prefix":"","firstName":"Morgan","middleName":"","lastName":"Steckler","suffix":""},{"id":603314753,"identity":"4a42d513-88b5-42e5-8391-3dd72ab2d424","order_by":3,"name":"Xiao-Ying Yu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA2klEQVRIiWNgGAWjYJCCA0DMw8DADKItSNLClgCkJUiyjMeAOC0GN3IMD/yoOCxjzr/m4+PKHRIM8u49BgS0pCUc7DlzmMdyxtvNhmfPSDAYnjlDSEvygQO8bYd5DG6c3SbZ2AbUMiMtgYCWxIaDf8FazjwjVkvygcNgW873sIG1yEskH8CrRfLMs4TDMmfSgbawGRsCtfAY8BzGr4XveI7xxzcV1vYG5w8/fNjYZiMn397YgFeLAsTEZmCMQHzAY4DfDgYGeYiJdQwM/AdQREbBKBgFo2AUwAEA7pRNsjtgAHAAAAAASUVORK5CYII=","orcid":"","institution":"Oak Ridge National Laboratory","correspondingAuthor":true,"prefix":"","firstName":"Xiao-Ying","middleName":"","lastName":"Yu","suffix":""}],"badges":[],"createdAt":"2026-02-11 21:38:37","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8855721/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8855721/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":104518442,"identity":"b8d5892c-3ad3-42ea-a34c-09a778bda711","added_by":"auto","created_at":"2026-03-12 18:30:22","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":3385252,"visible":true,"origin":"","legend":"\u003cp\u003eSelected tower locations (°N, °W) within ORR.\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-8855721/v1/0e0025d278dfbd57a0c34dd5.png"},{"id":104518445,"identity":"87d3686e-0823-4b99-a6c0-a6c394fd32b9","added_by":"auto","created_at":"2026-03-12 18:30:22","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":1318135,"visible":true,"origin":"","legend":"\u003cp\u003eTemperature boxplots of Sen’s slope results of weekly data of all towers within six years (2017 – 2022).\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-8855721/v1/e836e719300cde2a22265403.png"},{"id":104780858,"identity":"f194e74b-f867-4aef-9441-e3fb081f6b74","added_by":"auto","created_at":"2026-03-17 07:54:07","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":1296340,"visible":true,"origin":"","legend":"\u003cp\u003eHumidity boxplots of Sen’s slope results of weekly data of all towers within six years (2017 – 2022).\u003c/p\u003e","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-8855721/v1/b0065ad17b0462031368347b.png"},{"id":104518444,"identity":"ab207463-486c-4dc5-9ccf-d6ac9c1114de","added_by":"auto","created_at":"2026-03-12 18:30:22","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":1442591,"visible":true,"origin":"","legend":"\u003cp\u003eWind speed boxplots of Sen’s slope results of weekly data of all towers within six years (2017 – 2022).\u003c/p\u003e","description":"","filename":"Figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-8855721/v1/df90bf62df1ad7a2b117479c.png"},{"id":104518451,"identity":"d4ebd02f-a99f-4bad-8b1f-8c61da10c5c0","added_by":"auto","created_at":"2026-03-12 18:30:22","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":10459435,"visible":true,"origin":"","legend":"\u003cp\u003eSpatial maps of a summer week (Jul. 20 - Jul. 26) based on Mann–Kendall test results, (a) temperature and (b) absolute humidity.\u003c/p\u003e","description":"","filename":"Figure5.png","url":"https://assets-eu.researchsquare.com/files/rs-8855721/v1/2007679273b9b861e0e06568.png"},{"id":104518453,"identity":"b82a0142-1e9b-490a-9aa8-0cb6a32ce4de","added_by":"auto","created_at":"2026-03-12 18:30:22","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":11973131,"visible":true,"origin":"","legend":"\u003cp\u003eSpatial map of a winter week (Jan. 13 – Jan. 19) based on Mann–Kendall test results, (a) temperature and (b) peak wind speed.\u003c/p\u003e","description":"","filename":"Figure6.png","url":"https://assets-eu.researchsquare.com/files/rs-8855721/v1/35ec7661d6be1e19c26b5888.png"},{"id":104781238,"identity":"b4b1385c-2dc0-4a01-aa99-22a7ab2be645","added_by":"auto","created_at":"2026-03-17 07:55:11","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":324366,"visible":true,"origin":"","legend":"\u003cp\u003eScatterplots of temperature and humidity based on clustering analysis.\u003c/p\u003e","description":"","filename":"Figure7.png","url":"https://assets-eu.researchsquare.com/files/rs-8855721/v1/a9adf5cb00e056c403180516.png"},{"id":104781211,"identity":"cd0200a0-0d74-4ca8-9b32-ba04b5b81571","added_by":"auto","created_at":"2026-03-17 07:55:09","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":159497,"visible":true,"origin":"","legend":"\u003cp\u003eClustering results of seasonal composition of temperature–humidity clusters.\u003c/p\u003e","description":"","filename":"Figure8.png","url":"https://assets-eu.researchsquare.com/files/rs-8855721/v1/4afcd66630a6a5b305b57480.png"},{"id":104781178,"identity":"ed96ae36-ff19-46c8-b846-b2d903ef922a","added_by":"auto","created_at":"2026-03-17 07:55:03","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":375573,"visible":true,"origin":"","legend":"\u003cp\u003eScatterplots of temperature and wind speed.\u003c/p\u003e","description":"","filename":"Figure9.png","url":"https://assets-eu.researchsquare.com/files/rs-8855721/v1/43e15d98f2890adc0d10272b.png"},{"id":104518448,"identity":"0a9caee5-7010-475a-ad67-7c5840c350c7","added_by":"auto","created_at":"2026-03-12 18:30:22","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":158043,"visible":true,"origin":"","legend":"\u003cp\u003eClustering seasonal composition of temperature–windspeed clusters.\u003c/p\u003e","description":"","filename":"Figure10.png","url":"https://assets-eu.researchsquare.com/files/rs-8855721/v1/2ddd33429d0082dee49c1fa5.png"},{"id":104518452,"identity":"09064700-89a7-4d36-a4c3-99d9cd8fa765","added_by":"auto","created_at":"2026-03-12 18:30:22","extension":"png","order_by":11,"title":"Figure 11","display":"","copyAsset":false,"role":"figure","size":562218,"visible":true,"origin":"","legend":"\u003cp\u003eHeatmap of total days (mean) by event and month.\u003c/p\u003e","description":"","filename":"Figure11.png","url":"https://assets-eu.researchsquare.com/files/rs-8855721/v1/10d8d5fb2793eb70f93b2252.png"},{"id":104780783,"identity":"6918737a-dba2-4a80-b8a6-54c23c394eec","added_by":"auto","created_at":"2026-03-17 07:53:57","extension":"png","order_by":12,"title":"Figure 12","display":"","copyAsset":false,"role":"figure","size":182375,"visible":true,"origin":"","legend":"\u003cp\u003eDistribution of duration by events (min/mean/max).\u003c/p\u003e","description":"","filename":"Figure12.png","url":"https://assets-eu.researchsquare.com/files/rs-8855721/v1/36c6ca0f2b83f20ce511afd1.png"},{"id":104784520,"identity":"818bd5ec-fbaa-474c-aea4-0aae94c58440","added_by":"auto","created_at":"2026-03-17 08:08:05","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":29363991,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8855721/v1/d3b4d660-a942-4040-add9-8992fe29d909.pdf"},{"id":104518454,"identity":"94f415cf-1d4b-42ba-97bc-c975d51c59f2","added_by":"auto","created_at":"2026-03-12 18:30:22","extension":"docx","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":6072529,"visible":true,"origin":"","legend":"","description":"","filename":"EDAORRSI02102026.docx","url":"https://assets-eu.researchsquare.com/files/rs-8855721/v1/daa2f94508d92e01f2dae1b0.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Exploratory Data Analysis of Long-Term Oak Ridge Reservation Meteorological Data for Extreme Weather Event Discovery","fulltext":[{"header":"Introduction","content":"\u003cp\u003eExtreme weather events significantly impact environmental conditions and site operations. Oak Ridge National Laboratory (ORNL) accommodates large-scale scientific facilities and experimental programs that are at times highly sensitive to the impacts of extreme weather events. Understanding long-term trends and accurately identifying extreme events are crucial for effective risk mitigation strategies.\u003c/p\u003e \u003cp\u003eDue to its geographical setting, ORNL experiences diverse and highly variable weather conditions, necessitating systematic meteorological trend assessment and extreme weather detection to address challenges in reliable forecasting and operational management. For instance, sudden temperature fluctuations, humidity variations, and wind extremes can affect experimental conditions, facility Heating, Ventilation, and Air Conditioning (HVAC) system performance, energy consumption, and overall infrastructure stability. An in-depth understanding of these meteorological factors and their long-term trends can enhance risk management practices, support infrastructure resilience, and contribute to the formulation of adaptive operational strategies \u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eTo address these requirements, this work employs a detailed analysis of multi-year meteorological data at six distinct towers across multiple measurement heights. Through robust data quality assessments, comprehensive temporal and spatial trend analyses, and clustering methods, we aim to comprehensively characterize and interpret the underlying patterns in observed meteorological data. Specifically, the analysis uses statistical techniques, such as Sen's slope estimator and Mann-Kendall tests, and precise detection of extreme weather events through feature-based clustering. By synthesizing these methods, we provide actionable insights into meteorological phenomena relevant to ORNL's operational contexts. Our findings highlight crucial patterns of variability, both spatially and temporally, enabling refined risk assessments, targeted preventive measures, and informed decision-making. Ultimately, this analysis supports enhanced preparedness against extreme weather impacts, safeguards research integrity, and promotes operational resilience at the ORNL site.\u003c/p\u003e"},{"header":"Data and Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eSite Description\u003c/h2\u003e \u003cp\u003eORNL spans a varied topographic and forested landscape within the Ridge-and-Valley region of the Appalachian Mountains and plays a critical role in energy research, national security, and environmental sustainability. To support research and operational needs, ORNL and Y-12 maintain a high-resolution meteorological monitoring network consisting of eight meteorological towers and two wind profilers, strategically distributed across the site. The tower locations used for the purposes of this work are indicated by red markers in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e1\u003c/span\u003e. Most of the selected sites are equipped with multi-level sensors, with measurement heights ranging from 2 m to 60 m above ground level, capturing key meteorological variables such as air temperature, wind speed and direction, humidity, and solar radiation at 15-minute (min) intervals.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eData Description\u003c/h3\u003e\n\u003cp\u003eThis study utilizes six years (2017\u0026ndash;2022) of meteorological data collected from these towers to characterize site-specific weather patterns and support trend analysis along with extreme event detection. The high-resolution data set consists of 15-min measurement intervals and includes data collected from ORNL and Y-12 consisting of air temperature, wind speed, wind direction, relative humidity, absolute humidity, solar radiation, and precipitation. The present study focuses on four key parameters: air temperature, wind speed, relative humidity, and absolute humidity, which were standardized to the following units: (1) temperature in degrees Celsius (\u0026deg;C), (2) wind speed in miles per hour (mph), (3) relative humidity in percent (%), and (4) absolute humidity in grams per cubic meter (g/m\u0026sup3;).\u003c/p\u003e \u003cp\u003eTo ensure quality and consistency, raw 15-min time series were processed to address missing values, outliers, and sensor drift \u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e. Beyond standard quality control, a multi-tower rolling window method was developed for robust anomaly detection. The approach compares concurrent sensor readings of the same variable across multiple towers within a specified temporal window. Appropriately cleaned data were then resampled or smoothed using running averages to enable analysis at broader temporal scales (e.g., hourly and daily). These quality-assured time series form the foundation for all subsequent exploratory data analysis, trend evaluation, and extreme event characterization.\u003c/p\u003e \u003cp\u003eSummary statistics, including daily, weekly, and monthly maxima and minima, were extracted to support EDA and trend detection. The resulting processed features form the basis for future machine learning tasks aimed at understanding the temporal structure of meteorological phenomena and for evaluating their operational implications regarding building performance and site resilience at ORNL.\u003c/p\u003e\n\u003ch3\u003eTrend Analysis\u003c/h3\u003e\n\u003cp\u003eTwo methods for trend analysis were adopted: namely Sen\u0026rsquo;s slope and the Mann\u0026ndash;Kendall (MK) test, to determine the representative trend of the measured parameters. The MK test is a non-parametric statistical test that can be used for detecting time series trends which allow for identification of trends based primarily on ranks without the need for specifying linearity \u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e. Hence, it offers robustness to non-normality and cleans data with missing values. The MK hypothesis includes the null hypothesis, where H0 refers to either a sample (i.e., measured meteorological parameter) or the independent random variables. The subsamples of each variable are independent and identically distributed over years \u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eIn the null hypothesis (H0) of the MK test, data coming from a population with independent realizations are not significantly different (i.e., no trend). If a calculated p-value for a trend test is smaller than a significance level (e.g., 0.05), the null hypothesis is rejected (i.e., the trend is significant). Additionally, the MK method is well-known for assessing the significance of trends in hydroclimatic time series data, such as rainfall, temperature, and streamflow.\u003c/p\u003e\n\u003ch3\u003eClustering\u003c/h3\u003e\n\u003cp\u003eUnsupervised clustering techniques were employed to identify desired patterns. Clustering is a form of exploratory data analysis that partitions observations into groups (clusters) such that data points within the same cluster are more like each other than those in other clusters. We used K-means clustering to explore inherent structures of multi-dimensional meteorological data\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e primarily to take advantage of its efficiency within large datasets and to obtain spherical clusters of similar size.\u003c/p\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eRolling Window\u003c/h2\u003e \u003cp\u003eA variable-specific outlier detection strategy was implemented to ensure data integrity across the six selected meteorological towers. For our key variables: air temperature (TempC), absolute humidity (AbsHum), and peak wind speed (PkWSpdMph), a multi-tower rolling-window approach was applied as shown in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eOutlier detection by the multi-tower rolling window method.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVariable\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMethod\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eWindow\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eStd Dev\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e Threshold\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTempC\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMulti-tower\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1 day\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAbsHum\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMulti-tower\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e6 hours\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWSpdMph\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMulti-tower\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e6 hours\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePkWSpdMph\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMulti-tower\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e6 hours\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"4\"\u003e\u003csup\u003e1\u003c/sup\u003eStd Dev: standard deviation\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThis method identified data points that deviated beyond a specified number of standard deviations from the multi-tower mean within each rolling window. Air temperature anomalies were detected using a 1-day window with a 3-sigma threshold, absolute humidity outliers were identified using a 6-hour (h) window and a 3-sigma threshold, and peak wind speed anomalies were captured using a 6-h window with a 4-sigma threshold.\u003c/p\u003e \u003cp\u003eBy combining automated multi-tower anomaly detection with targeted manual inspection, the quality-control framework effectively enhanced data consistency and reliability across the six towers, removing sensor biases and spurious outliers as well as providing a robust foundation for the subsequent analyses of temporal trends and extreme weather events.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eData Structure and Feature Characterization\u003c/h3\u003e\n\u003cp\u003e \u003cb\u003eFigures \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e\u0026ndash;S2\u003c/b\u003e collectively illustrate the multi-year meteorological variability observed across six towers between 2017 and 2022. Temperatures generally ranged from \u0026minus;\u0026thinsp;25\u0026deg;C to 35\u0026deg;C with subtle vertical gradients and moderate interannual stability, although the most recent years suggested slightly higher medians and included some anomalous cold outliers. Tower A records highlighted seasonal cycles in temperature, absolute humidity, and relative humidity, alongside variable wind speed and episodic precipitation, reflecting both seasonal and synoptic-scale weather influences. Seasonal correlation matrices further reveal dynamic couplings among variables, for instance, stronger negative correlations between wind speed and temperature in winter and positive temperature\u0026ndash;radiation links in spring and summer, which underscore the value of high-resolution, seasonally disaggregated data for characterizing microclimatic variability and extreme events. However, conventional statistical methods only reveal the overall distribution and do not capture temporal trends across years. The MK test and Sen\u0026rsquo;s slope analyses were applied at multiple time scales, as described in Section 3.3.\u003c/p\u003e\n\u003ch3\u003eMK Test and Sen’s Slope Analysis\u003c/h3\u003e\n\u003cp\u003eFigure \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e2\u003c/span\u003e presents Sen\u0026rsquo;s slope estimates for minimum, maximum, mean, and median temperature across International Organization for Standardization (ISO) weeks, highlighting seasonal patterns in warming and cooling trends. ISO weeks follow the ISO-8601 calendar system, where weeks start on Monday and week 1 is defined as the week containing the first Thursday of the year\u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eRed boxes indicate the summer period (weeks 22\u0026ndash;35), during which all four-temperature metrics generally exhibit positive slopes, reflecting a tendency toward summer warming. This effect is most consistent for mean and median temperatures, with smaller but still positive trends in minimum and maximum temperatures. In contrast, blue boxes highlight the winter period (weeks 1\u0026ndash;8 and 48\u0026ndash;52), which are dominated by negative slopes across all metrics, indicative of wintertime cooling trends. These opposing seasonal signals underscore the asymmetric nature of recent temperature changes: warmer summers and colder winters.\u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e3\u003c/span\u003e presents Sen\u0026rsquo;s slope estimates for minimum, maximum, mean, and median absolute humidity (AbsHum) across weeks of a whole year, highlighting the seasonal variability of moisture trends. The red boxes denote the summer period (weeks 22\u0026ndash;25), during which positive slopes dominate among most metrics, particularly for minimum and median absolute humidity, indicating a moistening trend in summer. Maximum absolute humidity also exhibits modest positive slopes in this period, suggesting more frequent or intense high-moisture events. In contrast, several weeks outside the summer period, especially in late autumn and winter, display near-zero or negative slopes, implying stable or declining moisture levels. The consistent summertime increases in absolute humidity, when coupled with concurrent temperature rises, highlight the potential for compounding heat\u0026ndash;moisture stress during warm-season extremes.\u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e4\u003c/span\u003e displays Sen\u0026rsquo;s slope estimates for minimum, maximum, mean, and median peak wind speed (PkWSpdMph) across weeks of a year, highlighting contrasting seasonal trends. The red boxes mark the summer period (weeks 22\u0026ndash;35), during which Sen\u0026rsquo;s slopes are generally near zero or slightly positive for most metrics, suggesting relatively stable or modestly increasing peak wind speeds in mid-summer. In contrast, the blue boxes highlight the winter period (weeks 1\u0026ndash;8 and 48\u0026ndash;52), when positive slopes dominate among nearly all metrics, most notably for maximum and mean peak wind speeds, which indicate a tendency toward stronger winter winds. Seasonal asymmetry suggests that while peak wind speeds remain relatively steady in summer, the winter season is characterized by a strengthening of wind extremes, which, when coupled with corresponding cooling trends, can exacerbate wind chill effects, increasing operational and safety risks in exposed environments.\u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e5\u003c/span\u003e \u003cb\u003e(a)\u003c/b\u003e illustrates spatial maps of weekly temperature trends for a representative summer period (week 30), derived from the Mann\u0026ndash;Kendall test and Sen\u0026rsquo;s slope estimator among all meteorological towers and including sensors at all heights. During this week, nearly all sensor locations display positive Sen\u0026rsquo;s slope values (red squares), indicating a warming trend. Larger marker sizes highlight locations with more pronounced increases at several towers and elevated measurement heights. Absolute humidity trends for the same period in Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e5\u003c/span\u003e \u003cb\u003e(b)\u003c/b\u003e reveal a consistent pattern of positive slopes among all reporting stations. This mid-summer moistening coupled with temperature increases suggests a heightened potential for heat\u0026ndash;moisture stress. Such conditions may intensify thermal discomfort, reduce evaporative cooling efficiency, and elevate heat index values, particularly in above-canopy environments involving tall-equipment operations.\u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e6\u003c/span\u003e \u003cb\u003e(a)\u003c/b\u003e depicts spatial patterns of weekly temperature trends for a representative winter period (week 3), derived from the Mann\u0026ndash;Kendall test and Sen\u0026rsquo;s slope estimator among all towers and including sensors at all heights. Nearly all sensors exhibit strongly negative Sen\u0026rsquo;s slope values (blue squares), indicating widespread cooling trends. In contrast, Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e6\u003c/span\u003e \u003cb\u003e(b)\u003c/b\u003e shows peak wind speed trends for the same week, which are positive (red\u0026ndash;orange circles) across most stations. The corresponding patterns of intensified winds and declining temperatures are consistent with cold air outbreak conditions, where strong cold air advection and elevated winds enhance heat loss and exacerbate wind chill effects. Such conditions pose heightened thermal stress risks, especially in above-canopy environments involving tall-equipment operations and wintertime site safety planning.\u003c/p\u003e \u003cp\u003eWeekly Sen\u0026rsquo;s slope trends for solar radiation, wind-direction variability, and vertical wind speed are also evaluated. The results show little evidence of sustained long-term trends. These results are provided in the Supplementary Information (\u003cb\u003eFigs. S3\u0026ndash;S5\u003c/b\u003e). Monthly spatial maps for temperature are included in the Supplementary Information to provide a broader temporal perspective (\u003cb\u003eFig. S6\u003c/b\u003e).\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eClustering Analysis\u003c/h2\u003e \u003cp\u003eFigure\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e7\u003c/span\u003e illustrates the K-means clustering of temperature and absolute humidity. Among them, Cluster 2 (green) captures the hottest and most humid conditions; Cluster 1 (yellow) corresponds to the coldest and driest conditions; Cluster 0 (blue) represents cool\u0026ndash;moderately humid conditions; Cluster 3 (orange) reflects warm\u0026ndash;moderately humid conditions. The centroids and ranges of temperature\u0026ndash;humidity clusters are shown in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eCluster centroids and ranges of temperature\u0026ndash;humidity clusters.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCluster\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTemp (\u0026deg;C)\u003c/p\u003e \u003cp\u003eMin / Mean / Max\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAbsHum (g/m\u0026sup3;)\u003c/p\u003e \u003cp\u003eMin / Mean / Max\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e5.50 / 11.56 / 25.40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.50 / 7.03 / 11.10\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-16.00 / 2.39 / 10.40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.80 / 4.05 / 7.20\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e17.90 / 24.25 / 34.60\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e10.50 / 17.04 / 25.60\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e11.80 / 18.93 / 32.50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e3.70 / 11.68 / 15.60\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eFigure\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e8\u003c/span\u003e presents that Cluster 2 is dominated by summer (69%), confirming that extreme hot\u0026ndash;humid conditions occur in summer. Cluster 1 is winter-dominated (65%), representing cold\u0026ndash;dry extremes. Clusters 0 and 3 serve as transitional regimes, linking winter to spring/fall and summer to spring/fall, respectively.\u003c/p\u003e \u003cp\u003eThe hot and humid conditions captured in Cluster 2 raise particular concern. The cluster centroid indicates a mean temperature of 24.3\u0026deg;C (maximum 34.6\u0026deg;C) and a mean absolute humidity of 17.4 g/m\u0026sup3; (maximum 25.6 g/m\u0026sup3;). For comparison, the National Weather Service (NWS) Heat Index\u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e shows that caution is warranted once air temperature reaches 80\u0026deg;F (26.7\u0026deg;C) with a relative humidity of 40%, which corresponds to an absolute humidity of about 10 g/m\u0026sup3; It is worth noting that these conditions are met for about 9 to 10% of all hours at ORNL. For instance, 857 such hours out of 8760 occurred in 2022. At the upper end of Cluster 2, the combination of 34.6\u0026deg;C and 25.6 g/m\u0026sup3; translates to a relative humidity of roughly 66%, yielding a heat index of about 116\u0026deg;F, a level categorized by the NWS as Danger, where the likelihood of heat-related disorders under prolonged exposure or strenuous activity becomes severe.\u003c/p\u003e \u003cp\u003eFigure\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e9\u003c/span\u003e shows the cluster results of temperature and windspeed. Cluster 1 (yellow) captures high-wind conditions across the full temperature range, with windspeed exceeding 20 mph and reaching up to 60 mph; Cluster 0 (blue) represents warm, low-to-moderate wind regimes; Cluster 2 (green) corresponds to hot, moderate-wind conditions; and Cluster 3 (orange) reflects cold, moderate-wind conditions. The clustering highlights that Cluster 1 is distinguished primarily by extreme windspeed rather than temperature. The centroids and ranges of temperature\u0026ndash;windspeed clusters are shown in Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eCluster centroids and ranges of temperature\u0026ndash;windspeed clusters.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCluster\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTemp (\u0026deg;C)\u003c/p\u003e \u003cp\u003eMin / Mean / Max\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eWindspeed (mph)\u003c/p\u003e \u003cp\u003eMin / Mean / Max\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e11.10 / 20.11 / 33.30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.00 / 4.40 / 9.80\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-10.00 / 13.00 / 32.80\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e12.70 / 20.26 / 75.00\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e13.10 / 24.00 / 34.60\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e5.10 / 11.44 / 23.70\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-15.00 / 5.89 / 14.50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.00 / 6.78 / 18.20\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eFigure\u0026nbsp;\u003cspan refid=\"Fig12\" class=\"InternalRef\"\u003e10\u003c/span\u003e indicates that Cluster 1 is dominated by spring (43%) and winter (35%), presenting strong wind events occurring across a wide temperature range but more frequent in colder and transitional seasons. Cluster 3 is winter-dominated (52%), reflecting cold\u0026ndash;moderate wind regimes. Cluster 0 occurs mainly in summer (44%) and fall (29%), while Cluster 2 peaks in summer (49%), representing hot\u0026ndash;moderate-to-strong wind conditions.\u003c/p\u003e \u003cp\u003eCluster 1 represents cold and windy conditions that are prone to significant wind chill effects. The minimum values in this cluster reach \u0026minus;\u0026thinsp;10\u0026deg;C (14\u0026deg;F) for temperature and 12.7 mph for wind speed. According to the NWS Wind Chill Index \u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e, this combination corresponds to a wind chill temperature of approximately \u0026minus;\u0026thinsp;33\u0026deg;C (\u0026ndash;27\u0026deg;F). At this level, exposed skin can develop frostbite in 10\u0026ndash;30 min. Such conditions highlight the elevated risk of cold-related health hazards during strong wind episodes in winter.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eExtreme Weather Events\u003c/h2\u003e \u003cp\u003eExtreme weather event thresholds were determined by combining long-term trend analysis with criteria established by the U.S. National Weather Service (USNWS). Five event types were defined to represent key meteorological hazards relevant to the study region (Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eExtreme weather threshold and associated flags.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eExtreme Event\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFlag(s)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eThreshold\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTemperature\u0026ndash;moisture hazard\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eE1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eTemp\u0026thinsp;\u0026gt;\u0026thinsp;24\u0026deg;C and AbsHum\u0026thinsp;\u0026gt;\u0026thinsp;20 g/m\u0026sup3;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWind chill event\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eE2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eTemp\u0026thinsp;\u0026le;\u0026thinsp;4.8\u0026deg;C and PkWind\u0026thinsp;\u0026ge;\u0026thinsp;3 mph\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003eLow temperature event\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eE3_1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eTemp\u0026thinsp;\u0026lt;\u0026thinsp;0\u0026deg;C\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eE3_2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eTemp \u0026lt; -5\u0026deg;C\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eE3_3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eTemp \u0026lt; -10\u0026deg;C\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHigh wind event\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eE4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePkWind\u0026thinsp;\u0026gt;\u0026thinsp;25 mph\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003eLow wind speed event\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eE5_1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePkWind\u0026thinsp;\u0026lt;\u0026thinsp;2.0 mph\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eE5_2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePkWind\u0026thinsp;\u0026lt;\u0026thinsp;1.0 mph\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eE5_3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePkWind\u0026thinsp;\u0026lt;\u0026thinsp;0.5 mph\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThe temperature\u0026ndash;moisture hazard (E1) denotes concurrent high temperature (\u0026gt;\u0026thinsp;24\u0026deg;C) and high absolute humidity (\u0026gt;\u0026thinsp;20 g m⁻\u0026sup3;), conditions known to elevate the heat index and increase the risk of heat-related illness. Wind chill events (E2) were identified when air temperature was \u0026le;\u0026thinsp;4.8\u0026deg;C and the wind speed was \u0026ge;\u0026thinsp;3 mph, reflecting conditions under which cold stress and frostbite risk intensify. Low-temperature events (E3) included three severity levels (\u0026lt;\u0026thinsp;0\u0026deg;C, \u0026lt; -5\u0026deg;C, and \u0026lt; -10\u0026deg;C) to represent progressively stronger cold exposure. High-wind events (E4) were defined by peak wind speeds exceeding 25 mph, corresponding to the operational threshold for suspending outdoor activities at ORNL. Finally, low-wind events (E5) were characterized by wind speeds below 2.0 mph, with additional levels at 1.0 mph and 0.5 mph, conditions of particular concern for emergency release scenarios under pollutant stagnation conditions.\u003c/p\u003e \u003cp\u003eThis standardized framework ensures consistent identification and classification of meteorological conditions of concern within the tower network at ORR, with multiple thresholds for low temperature and low wind speed events capturing varying levels of severity.\u003c/p\u003e \u003cp\u003eAfter determining thresholds for each type of extreme event, their occurrence frequency and duration were analyzed. Because the towers differed in measurement height and sensor configuration, tower-specific variable mappings were applied for event detection (Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e5\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSensor variables and measurement heights used for event detection across towers.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTower\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTemperature\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAbsolute Humidity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePeak Wind Speed\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e30m\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e15m\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e30m\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eB\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e30m\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ndash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e30m\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eD\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e35m\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e15m\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e35m\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eF\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e10m\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e10m\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e10m\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e25m\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ndash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e25m\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eY\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e33m\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026ndash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e33m\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eFor each tower, events were first detected independently using tower-specific variable mappings and then merged across towers using a union approach, ensuring that overlapping or adjacent intervals were not double-counted. Event days were counted on a calendar-day basis (i.e., multiple events or multiple towers on the same day were counted once). Figures\u0026nbsp;\u003cspan refid=\"Fig13\" class=\"InternalRef\"\u003e11\u003c/span\u003e and \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e12\u003c/span\u003e summarize the monthly occurrence of each event and highlight the variation in event duration through its mean, minimum, and maximum values.\u003c/p\u003e \u003cp\u003eAs shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig13\" class=\"InternalRef\"\u003e11\u003c/span\u003e, distinct seasonal patterns emerge. Temperature\u0026ndash;moisture hazards (E1) are concentrated in the summer months (June \u0026ndash; August), with an average of 6 \u0026minus;\u0026thinsp;14 days per month; the maximum single-event duration in July exceeded 1,000 min, highlighting potential heat stress during sustained hot and humid episodes. Wind chill events (E2) dominate the winter season (November\u0026ndash;January), especially December, with more than 20 days per month and maximum durations surpassing 5,000 min, reflecting persistent cold and windy conditions. Low temperature events (E3) show a similar winter concentration, with extremes occasionally exceeding 10,000 minutes, posing risks to infrastructure and operations. In contrast, high wind events (E4) are distributed more broadly, peaking in spring and autumn (March\u0026ndash;May). Although the monthly frequencies are lower (~\u0026thinsp;15\u0026ndash;19 days), their repeated occurrence implies potential operational disruptions. Finally, low wind speed events (E5) are most frequent in summer and early autumn (July\u0026ndash;September), with up to 30 event days per month and durations exceeding 3,000 min, emphasizing the possibility of exacerbating poor air quality conditions.\u003c/p\u003e \u003cp\u003eOverall, the results demonstrate that each event type exhibits clear seasonal peaks and distinctive persistence characteristics. These findings underscore the importance of tailoring risk mitigation and preparedness strategies to the seasonal hazard profiles observed across the tower network.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eWind Patterns related to Extreme Events\u003c/h2\u003e \u003cp\u003eWind pattern analysis was also conducted to further interpret the synoptic mechanisms behind observed extreme meteorological events in ORNL. Results show that low-wind conditions, often associated with high pollutant concentration potential, were highly correlated with daytime terrain-driven thermal winds (40%) and cross-valley winds from the west (27%). These conditions reflect the complex interactions of air flow with the Great Valley\u0026rsquo;s topography \u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e, where weak thermally induced circulations and lateral deflection of airflow limit horizontal dispersion and promote pollutant buildup. However, unstable atmospheric conditions sometimes counteract these effects.\u003c/p\u003e \u003cp\u003eWind chill events were significantly associated with cross-valley winds from the northwest, north, and north-northeast sectors, accounting for 19\u0026ndash;24% of cases overall and increasing to 35\u0026ndash;50% during the cool season (October\u0026ndash;March). This pattern suggests that cold air masses from higher latitudes often cross local terrain during winter, amplifying cold stress. Similarly, low-temperature events (below 0\u0026deg;C) were closely linked to NNE\u0026ndash;NE and WNW\u0026ndash;NNW winds, 18% and 17% overall (rising to 34\u0026ndash;36% in the cool season), indicating an association with both cold-air drainage along the valley axis as well as synoptic-scale cold air advection episodes.\u003c/p\u003e \u003cp\u003eIn contrast, high temperature and moisture hazards were observed mainly during the warm months (April\u0026ndash;September), comprising 14% of warm-season observations, and were dominated by southwest-to-northeast up-valley winds that transported warm, humid air from the Gulf of America/Mexico. These up-valley flows, often deflected from the west into the Great Valley, may enhance the accumulation of moisture within the Great Valley, contributing to thermal discomfort and elevated cooling demand, especially under low wind conditions that are typical of summer.\u003c/p\u003e \u003cp\u003eOverall, wind pattern analysis highlights the critical role of valley-modulated airflow in shaping ORNL and ORR local climate, pollutant dispersion, and thermal risk. Integrating these terrain\u0026ndash;wind\u0026ndash;event relationships into predictive and optimization frameworks, such as HVAC control and environmental management, can improve resilience, comfort, and energy efficiency under site-specific meteorological conditions.\u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThe combination of high-risk temperature\u0026ndash;moisture conditions in summer and significant wind chill effects in winter presents an environmental challenge to ORNL\u0026rsquo;s operational resilience. Our analyses indicate that the laboratory experiences frequent weather extremes at both ends of the temperature spectrum. In summer, higher temperatures coupled with greater humidity elevate the risks of heat-related illness, equipment overheating, and reduced worker productivity \u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e. In winter, outbreaks of cold, windy weather increase risks of hypothermia, ice formation, and weather-related disruptions \u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e. Effective management helps to mitigate the effects of environmental extremes that can endanger staff well-being and interrupt mission-critical operations.\u003c/p\u003e \u003cp\u003eOur climatological records show trends consistent with broader patterns but also capture the recent intensification of extremes. For instance, while the mean annual temperature at Oak Ridge rose by roughly 1.3\u0026deg;C between the 1970s and 2000s \u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e, our results highlight that the most consequential changes are occurring in the frequency and magnitude of extreme events\u0026mdash;heat index surges and wind chill drops\u0026mdash;rather than in gradual shifts of mean temperature. Our analysis provides site-specific evidence of these regional trends, providing more insight of data trends that were otherwise undiscovered using EDA approaches. Such findings are important to improve predictability of extreme weather events that may impact site operations. Our results validate that EDA can be a useful approach for assessing data trend indicative of conditions that may impair operation and potentially contribute to the DOE standards. Specifically, using statistical and EDA techniques in treating at least five years of meteorological data as a novel approach, which is required by DOE-STD-3009-2014 \u003csup\u003e12\u003c/sup\u003e, will improve risk analysis efficiency significantly for safety designs, strategies, and operations and serve the DOE-STD-1189-2016 \u003csup\u003e13\u003c/sup\u003e objectives by providing site relevant applications following DOE-HDBK-1224-2024 \u003csup\u003e14\u003c/sup\u003e. The ability to analyze extreme weather patterns that have impact on site operation is important to conform to the DOE-STD-1020-2016\u003csup\u003e15\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eClustering analysis proved especially useful for identifying compound extreme events, such as joint high-temperature and high-humidity episodes or low-temperature and high-wind conditions, that are directly relevant to safety planning. Based on these findings, climate adaptation planning at ORNL should integrate both observed trends and projected changes into its risk management strategies. Practical actions may include updating heat-stress management protocols (for example, using heat index thresholds to guide work/rest cycles on hot days) and strengthening cooling infrastructure in critical facilities. Winter preparations should also be enhanced by insulating buildings and pipes against deep freezes, maintaining vegetation to limit wind damage, and ensuring contingency plans for essential operations during severe cold and wind events. These measures align with broader DOE resilience initiatives that address site-specific climate. For ORNL, the data clearly justify proactive measures: hot, humid summers and more anomalous winter events present tangible risks to a laboratory of this scale and importance.\u003c/p\u003e \u003cp\u003eOverall, exploratory data analysis provides a detailed picture of how local climate conditions can affect site daily operations. The subtropical humid climate of Oak Ridge and the surrounding region results in high heat stress potential, while winters can exhibit highly variable conditions sometimes associated with strong winds that heighten cold stress. Thus, local climate sometimes produces high-risk weather conditions, dangerous heat levels in summer and significant wind chills in winter. When interpreted alongside regional climate data, the results suggest that ORNL is experiencing seasonal climate extremes that are typical of regional atmospheric dynamics on the northern fringe of the subtropical climate zone (Koppen Climate \u0026ndash; Cfa), yielding differing extreme events in all seasons. The findings are consistent with other analyses in the Southeast and across DOE facilities, confirming that even within a short time span, statistically significant annual-scale climate shifts can be detected when high-quality data are used.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThis study demonstrates the value of combining trend analysis, clustering, and threshold-based event definitions. Together, these methods allow us to quantify subtle changes, distinguish normal from extreme conditions, and identify threshold exceedances that have real operational impacts. For a research campus like ORNL, such information is essential for evidence-based decisions on safety and infrastructure. Continued high-resolution meteorological monitoring will be important to track how these trends evolve, whether winter cooling persists or reverses, and whether heat extremes worsen steadily or accelerate. These insights can inform ORNL\u0026rsquo;s climate adaptation and mitigation strategies, ensuring that the laboratory remains safe, resilient, and well prepared for future meteorological challenges. By situating ORNL\u0026rsquo;s data within the broader scientific context of climate variations, our findings highlight both local manifestations of weather extremes and their regional significance. The agreement between our findings and established meteorological trends strengthens confidence in these interpretations and reinforces the need for continued observation and analysis. In summary, EDA is a viable approach to capture key aspects of broader climate dynamics and provides actionable knowledge for improving environmental safety using a six-year meteorological record that is fully qualified to support ORNL\u0026rsquo;s operation.\u003c/p\u003e"},{"header":"Declarations","content":" \u003cp\u003e \u003cstrong\u003eCompeting interests\u003c/strong\u003e \u003cp\u003eThe authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.\u003c/p\u003e \u003c/p\u003e\u003ch2\u003eFunding\u003c/h2\u003e \u003cp\u003eResearch efforts were supported by the Nuclear Safety Research and Development (NSR\u0026amp;D) program sponsored by National Nuclear Security Administration (NNSA) Office of Environment, Health, Safety and Security (EHSS) of the U.S. Department of Energy (DOE).\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eConceptualization and design: X.Y.Y and Z.L.; Data curation: M.S., K.B. and X.Y.Y.; Investigation: Z.L., M.S. and X.Y.Y.; Formal analysis: Z.L.; Validation: M.S. and X.Y.Y.; Writing\u0026mdash;original draft: Z.L., K.B., and X.Y.Y.; Writing\u0026mdash;review and editing: Z.L., K.B., and X.Y.Y.; Visualization: Z.L. and M.S.; Supervision and funding acquisition: X.Y.Y... All authors have read and agreed to the published version of the manuscript.\u003c/p\u003e\u003ch2\u003eAcknowledgements\u003c/h2\u003e \u003cp\u003eThis manuscript has been authored by UT-Battelle, LLC, under contract DE-AC05-00OR22725 with the US DOE. The US government retains and the publisher, by accepting the article for publication, acknowledges that the US government retains a nonexclusive, paid-up, irrevocable, worldwide license to publish or reproduce the published form of this manuscript, or allow others to do so, for US government purposes. DOE will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.energy.gov/doe-public-access-plan\u003c/span\u003e\u003cspan address=\"https://www.energy.gov/doe-public-access-plan\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e).\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eThe meteorological dataset used in this study is openly available on Zenodo at: [https://zenodo.org/records/15171289](https:/zenodo.org/records/15171289)\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAyadi, R., Forouheshfar, Y. \u0026amp; Moghadas, O. Enhancing system resilience to climate change through artificial intelligence: a systematic literature review. \u003cem\u003eFront Clim\u003c/em\u003e \u003cb\u003e7\u003c/b\u003e, (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSteckler, M. R., Birdwell, K. R., Xu, H., YU, X. Y. \u0026amp; Descriptor High Temporal Resolution Meteorological Data at Oak Ridge Reservation (ORR-HiResMet). \u003cem\u003eIEEE Data Descriptions\u003c/em\u003e. \u003cb\u003e2\u003c/b\u003e, 118\u0026ndash;124 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhou, H., Ren, H., Royer, P., Hou, H. \u0026amp; Yu, X. Y. Big Data Analytics for Long-Term Meteorological Observations at Hanford Site. \u003cem\u003eAtmosphere\u003c/em\u003e \u003cb\u003e13\u003c/b\u003e, 136 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSanhudo, L., Rodrigues, J. \u0026amp; Filho, \u0026Ecirc;. V. Multivariate time series clustering and forecasting for building energy analysis: Application to weather data quality control. \u003cem\u003eJ. Building Eng.\u003c/em\u003e \u003cb\u003e35\u003c/b\u003e, 101996 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eISO - ISO 8601 \u0026mdash; Date. and time format. \u003cem\u003eISO\u003c/em\u003e (2017). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.iso.org/iso-8601-date-and-time-format.html\u003c/span\u003e\u003cspan address=\"https://www.iso.org/iso-8601-date-and-time-format.html\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUS Department of Commerce, N. What is the heat index? \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.weather.gov/ama/heatindex\u003c/span\u003e\u003cspan address=\"https://www.weather.gov/ama/heatindex\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUS Department of Commerce, N. Understanding Wind Chill. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.weather.gov/safety/cold-wind-chill-chart\u003c/span\u003e\u003cspan address=\"https://www.weather.gov/safety/cold-wind-chill-chart\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEnergy, U.-B. L. for the U. D. of. Wind Regimes in Complex Terrain of the Great Valley of Eastern Tennessee\u0026hellip; ORNL. https://www.ornl.gov/publication/wind-regimes-complex-terrain-great-valley-eastern-tennessee(2011).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHabibi, P. et al. Climate change and heat stress resilient outdoor workers: findings from systematic literature review. \u003cem\u003eBMC Public. Health\u003c/em\u003e. \u003cb\u003e24\u003c/b\u003e, 1711 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCDC \u0026amp; Preventing Hypothermia \u003cem\u003eWinter Weather\u003c/em\u003e (2024). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.cdc.gov/winter-weather/prevention/index.html\u003c/span\u003e\u003cspan address=\"https://www.cdc.gov/winter-weather/prevention/index.html\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAppendix B_Climate. Overview of the Oak Ridge Area.pdf.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePreparation of Nonreactor Nuclear Facility Documented Safety Analysis. (Invoked). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.standards.doe.gov/standards-documents/3000/3009-astd-2014\u003c/span\u003e\u003cspan address=\"https://www.standards.doe.gov/standards-documents/3000/3009-astd-2014\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIntegration of Safety into the Design Process. (Invoked). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.standards.doe.gov/standards-documents/1100/1189-astd-2016\u003c/span\u003e\u003cspan address=\"https://www.standards.doe.gov/standards-documents/1100/1189-astd-2016\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHazard and Accident Analysis Handbook. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.standards.doe.gov/standards-documents/1200/1224-bhdbk-2024\u003c/span\u003e\u003cspan address=\"https://www.standards.doe.gov/standards-documents/1200/1224-bhdbk-2024\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNatural Phenomena Hazards Analysis and Design Criteria for DOE Facilities. (Invoked). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.standards.doe.gov/standards-documents/1000/1020-astd-2016\u003c/span\u003e\u003cspan address=\"https://www.standards.doe.gov/standards-documents/1000/1020-astd-2016\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Exploratory Data Analysis (EDA), Extreme Weather Events, Trend Analysis, Unsupervised Clustering, Multivariate Time Series","lastPublishedDoi":"10.21203/rs.3.rs-8855721/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8855721/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Big data challenges are commonly encountered when conducting radiological and chemical hazard analysis to support operations of nuclear facilities in the Department of Energy (DOE). Extreme weather significantly influences environmental conditions and site operations, underscoring the need to incorporate accurate weather characterization in hazard assessments and safety planning. We applied exploratory data analysis (EDA) to six years (2017–2022) of high-resolution meteorological observations from six towers at the Oak Ridge Reservation (ORR). In addition to statistical methods to capture data distributions, EDA were used to examine trends using the Mann–Kendall test and Sen’s slope estimation. The results reveal asymmetric seasonal trends, namely warming in summer and cooling in winter. Clustering analysis was employed to interpret underlying patterns, identifying frequent co-occurrence of high temperature and high humidity during summer. Extreme weather events were further defined using feature-specific thresholds (e.g., temperature–moisture hazards, wind chill events, high-wind conditions), informed by regulatory guidelines and clustering outcomes. Our results show that EDA approaches can effectively assess big, long-term meteorological datasets and extract actionable information for site operations, particularly in relation to potentially hazardous extreme events. This study demonstrates that EDA is important in extreme event classification before applying machine learning in modeling.","manuscriptTitle":"Exploratory Data Analysis of Long-Term Oak Ridge Reservation Meteorological Data for Extreme Weather Event Discovery","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-03-12 18:30:16","doi":"10.21203/rs.3.rs-8855721/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewerAgreed","content":"249137642465638482384070213287485188536","date":"2026-03-20T16:17:26+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"282143513896214067271871377726689815931","date":"2026-03-09T17:10:09+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-03-09T16:18:10+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-02-17T10:24:40+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-02-12T14:09:41+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-02-12T14:03:57+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2026-02-11T21:27:49+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"eced492c-4290-4399-af02-3646d17ed15b","owner":[],"postedDate":"March 12th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":64199184,"name":"Earth and environmental sciences/Climate sciences"},{"id":64199185,"name":"Physical sciences/Engineering"},{"id":64199186,"name":"Earth and environmental sciences/Environmental sciences"},{"id":64199187,"name":"Earth and environmental sciences/Natural hazards"}],"tags":[],"updatedAt":"2026-03-12T18:30:17+00:00","versionOfRecord":[],"versionCreatedAt":"2026-03-12 18:30:16","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8855721","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8855721","identity":"rs-8855721","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}