NOAA National Severe Storms Laboratory

Abstract Winter weather creates dangerous conditions across the United States, requiring strong collaboration between the NWS and transportation partners. While prior research has examined how the public interprets winter weather products, less is known about how operational decision-makers use weather guidance to inform critical roadway decisions. This study addresses that gap through three semistructured focus groups with 15 transportation professionals that represent Departments of Transportation, public works agencies, and emergency management offices, and private-sector companies who also regularly experience winter weather. Using a phronetic iterative analysis, we first identified challenges and opportunities in communication and collaboration between transportation partners and the NWS. Findings show that unclear role definitions, differing timelines for road and weather impacts, and varied preferences for probabilistic information hinder coordination. At the same time, participants emphasized the value of audio–visual communication and collaborative forums such as the Pathfinder Project for aligning messages and decisions. We then apply systems theory to visualize how transportation partner participants synthesize weather forecasts from the public and private sectors, translate information into risk messaging, and refine decisions through collaborative discussions. These insights provide pathways for strengthening NWS–partner collaboration through the Pathfinder Project and for tailoring communication to the operational needs of transportation agencies. Significance Statement Winter storms often make roads hazardous, yet forecasts and road impacts do not always align. We asked transportation officials how they synthesize forecasts to keep drivers safe. They told us that unclear roles and complex forecasts can create challenges, but regular calls and programs like the Pathfinder Project help build trust and consistent messaging. We then applied systems theory to illustrate how partners synthesize information to make decisions, offering a path to improve future coordination.

Abstract Coastal breeze circulations [bay breeze (BB) and sea breeze (SB)] modulate weather conditions by advecting a maritime air mass onshore, altering air quality, and often initiating deep convection. Coastal breeze interactions with the planetary boundary layer (PBL) are not well simulated due to small-scale interactions being parameterized by numerical weather models. As part of the Tracking Aerosol Convection Interactions Experiment (TRACER) in Houston, Texas, PBL observations from remote sensors and uncrewed aerial systems provide a unique benchmark for evaluating model performance and investigating underlying processes. The Warn-on-Forecast System (WoFS) is a convection-allowing, 18-member ensemble designed to predict high-impact weather by combining rapidly updating data assimilation cycles with varying PBL and radiation parameterizations. The TRACER datasets are used to evaluate WoFS’s simulation of the PBL, coastal breezes, and their interactions with high vertical and temporal resolution. There is broad variability across members in the depth, arrival time, and intensity of the simulated SB case. Moreover, only a subset of members simulate the preceding BB, but none do so accurately. The radiation scheme impacts the ability to simulate a BB and the onset time of the SB. The mixing scheme for PBL parameterization lends differences in depth, intensity, and evolution. Nonlocal mixing schemes allow baroclinic circulations to develop more readily, thus simulating coastal flows better, but overestimate PBL depth and temperature. The influence parameterization schemes have on meteorological biases offers potential solutions to improve simulations of coastal PBL processes.

Artificial intelligence (AI) and machine learning (ML) tools are rapidly growing in capability and application across the weather enterprise. Fully AI-based numerical weather prediction (NWP) emulators are beginning to outperform traditional NWP, and many weather agencies have started to adopt ML-derived guidance products into the forecast process. For example, the United States National Weather Service’s Storm Prediction Center (SPC) has implemented a number of ML models to aid in the prediction and detection of tornadoes, severe wind, hail, and wildfires. However, the development of these AI/ML products and their subsequent transition into SPC operations revealed several challenges which potentially slowed their overall adoption into the forecasters’ workflow. This presentation will discuss several factors that impacted the adoption of AI/ML into forecast operations and highlight some best practices used by SPC to help streamline the research-to-operations transition. Case studies of AI/ML projects that were successfully transitioned into SPC operations will help illustrate the application of these best practices and showcase some of the common pitfalls faced by AI/ML development for operational applications.

Abstract Tornado outbreaks, involving multiple destructive tornadoes in close succession, result from interactions between large-scale atmospheric patterns and mesoscale dynamics. With a warming climate altering atmospheric behavior, the frequency and distribution of outbreak-favorable patterns may change, influencing future risks. Building on our recently introduced methodology for identifying tornado outbreak-favorable atmospheric patterns in a global climate model, we apply the method, for the first time, to future projections (2065–2100) across four Shared Socioeconomic Pathways using the MPI-ESM1-2-HR model. Focusing on May tornado outbreaks in the United States, we find that outbreak-supportive patterns are projected to increase in frequency and expand northward and eastward, beyond their historical geographic focus in the Great Plains and southeastern U.S. This application demonstrates how a dual-variable, threshold-based method can be applied to global climate model projections to investigate event-specific extremes, offering a pathway for assessing future risks of severe convective hazards under a warming climate.

Abstract Idealized simulations described in Part I investigated the importance of near-surface [low level (LL); 0–500 m] line-parallel shear, varied in both direction and magnitude, for modulation of quasi-linear convective system (QLCS) updraft and vortex attributes. Here, in Part II, we explore dynamical processes contributing to these attributes. Results indicate that LL shear strongly impacts the strength and streamwise nature of LL inflow vorticity, while free lower-tropospheric line-normal shear modulates storm propagation and LL storm-relative flow (largely through modifications to ground-relative flow). These combine to impact the storm-relative helicity (SRH) of QLCS inflow, with higher SRH for near-storm environments with increased LL hodograph curvature and length (i.e., enhanced LL line-parallel shear). Increased influx of streamwise vorticity fosters the stronger and wider rotating updrafts identified previously, which are driven by intense midlevel dynamic accelerations. When coincident with near-surface circulation, these strongly forced updrafts can support intense, long-lived mesovortices. These results demonstrate that changes to the orientation and magnitude of the ambient LL shear vector, as well as the vertical distribution of shear, can have considerable influence on the underlying dynamics governing the propagation, updrafts, and rotation within squall lines—even for a fixed line-normal environmental shear value. Altogether, this study (along with Part I) provides crucial context for understanding near-storm environmental impacts on QLCS morphology and rotation. Significance Statement Research related to linearly organized storms has largely focused on vertical changes in winds (i.e., shear) oriented normal to squall lines, but squall lines that produce considerable rotation often have a component of low-level shear oriented parallel to storms. This study, Part II of a larger work, assesses how these shear components influence the updrafts within squall lines. Line-normal shear and ambient flow impact how storms propagate, while line-parallel shear impacts their near-surface rotation. These factors combine to influence the overall strength and rotation within updrafts of quasi-linear convective systems (QLCSs) and their ability to support near-surface vortices that often precede squall-line tornadoes. These physical relationships can help improve our understanding and prediction of QLCS evolution and associated hazards.

Abstract Assimilating polarimetric radar data within a double-moment microphysics scheme requires that both hydrometeor mixing ratios and number concentrations be updated simultaneously to effectively utilize the radar information. This study directly assimilates polarimetric radar data in a fraternal twin observing system simulation experiment (OSSE) using both mixing ratios and number concentrations as analysis variables within a variational approach. A newly developed set of parameterized forward operators for polarimetric radar data, incorporating a new continuous melting model, is employed. To address challenges in minimizing the cost function, a power transformation function is applied to the analysis variables of mixing ratios and number concentrations. This approach alleviates issues arising from the very large dynamic range of number concentrations and the highly nonlinear relationship between the model’s hydrometeors and radar variables. Results from several groups of sensitivity experiments show that updating number concentrations using an appropriate power transformation function together with mixing ratios of hydrometeors reduces the analysis errors of radar variables and improves the analysis of polarimetric radar signatures. Updating number concentrations proves to be quite sensitive when assimilating differential reflectivity, while the additional assimilation of specific differential phase yields smaller analysis errors for reflectivity and mixing ratios of water vapor and rainwater compared to differential reflectivity assimilation alone. Experiments with smaller observation errors provide better analyses of the radar variables but also increase model variable analysis errors. Among the threshold values tested for reflectivity and polarimetric variables, assimilating polarimetric variables at grids where reflectivity exceeds 15 dB Z provides the best qualitative analysis.

Abstract Forecasting the impacts of wildfires has taken on more importance in recent years due to their increased frequency and significant human impacts. Wildfires often generate large smoke plumes which can loft aerosols deep into the atmosphere, which are then transported far downstream by the prevailing environment. Convection-allowing models such as the HRRR-Smoke currently forecast the evolution of smoke plumes, but many uncertainties in these forecasts remain. A smoke enabled version of the Warn-on-Forecast System (WoFS) has been developed to combine the existing smoke injection algorithms into a short-term (0–6 h) probabilistic forecast system. WoFS cycles at 15-min intervals, allowing for continuous updates of rapidly changing wildfire characteristics derived from geostationary satellite data. However, neither system assimilates any observations associated with the actual smoke plume. Few real-time, high-density observations of smoke aerosol concentrations are currently available. Precipitation radars, such as the WSR-88D, can detect biomass burning debris (BBD) lofted by wildfires. Smoke aerosols themselves are too small to be readily detectable from these radars, but the lofted debris is larger. Returned radar reflectivity and radial velocity provide an estimate on the spatial distribution of smoke and its movement within the environment. Radial velocity observations can be assimilated with minimal modifications to the existing system. This work explores the impact of assimilating radial velocity observations of BBD for three large wildfires in 2024. Assimilating these data has a large impact on the forecast smoke plume characteristics. Comparisons of forecast smoke with visible satellite imagery and radar reflectivity show that the forecasted spatial extent of wildfire smoke is improved. Significance Statement This research utilizes measurements of lofted smoke and burned debris from wildfires by Doppler radars to improve smoke forecasts in numerical weather prediction models. These data act to adjust the model wind field with actual measurements where no observations are currently assimilated into the system. Applying these data to three case studies showed that smoke forecasts are improved using these data, and this work represents the first of its kind utilizing Doppler radial velocity measurements for smoke forecasting applications.

Abstract The Warn-on-Forecast System (WoFS) is a convection-allowing ensemble that rapidly assimilates high-resolution radar, satellite, and other observational data to enhance forecasting capabilities at 0–6-h lead times. However, data latency and model spinup can reduce WoFS’ utility, especially within the first hour after initialization. On the other hand, Probability of Severe (ProbSevere), version 2 (PS2), is a set of statistical models that considers radar, satellite, and numerical weather prediction environment data to provide skillful severe hail, wind, and tornado probabilities at 0–1-h lead times, updated every 2 min. To enable seamless, probabilistic severe weather hazard forecasts leveraging the strengths of both systems, a random forest (RF) algorithm is developed that considers predictors from both WoFS and PS2. Experiments are conducted using different combinations of predictors at various spatial radii and lead times, and predictor importance is assessed. RFs configured using predictors from both WoFS and PS2 are found to outperform RFs configured from only one system, with the greatest gains in forecast skill from the all-predictor RFs coming at lead times less than 90 min. PS2 (WoFS) predictors are identified as more important at earlier (later) lead times. Severe hail forecasts were the most skillful of the three hazards, followed by severe wind and tornadoes. This new algorithm shows how combining complementary datasets using machine learning can improve short-term severe hazard guidance. Upcoming work will report on the results from testing during real-time forecasting experiments. Significance Statement A machine learning algorithm is used to combine two complimentary data sources for short-term severe weather forecasting. One is a high-resolution, rapidly updating numerical weather prediction model, while the other is a set of statistical models based on observations. The results show that combining these data sources results in more skillful forecasts than using either source alone. This algorithm will enhance short-term severe weather forecasting capabilities and provide insight into severe weather predictability at short lead times over local regions.

Abstract The Fujita (F) and enhanced Fujita (EF) scales have formed the baseline for tornado intensity estimation in the U.S. tornado climatology. Ahead of the EF scale’s debut in 2007, one stated goal was to maintain consistency in the ratings between the two scales (i.e., a rating from the EF scale would have the same climatological “meaning” as a rating from the F scale). However, the highly publicized “drought” of EF5 tornadoes since 2013 brings the broader consistency of present-day tornado rating practices relative to the F-scale era into question. Among a population of significant (EF2–EF5) tornadoes from 2014 to 2023 versus a climatological population of F2–F5 tornadoes from the last 30 years of F-scale use (1977–2006), the fraction of significant tornadoes rated EF2 has increased, while fractions of EF3, EF4, and EF5 have all decreased. Subsequently, the rate of fatalities associated with a given EF3 or EF4 tornado in 2014–23 is substantially greater than for F3 or F4 tornadoes in the 30-yr climatology, with the average EF4 tornado from 2014 to 2023 being over 2.5 times deadlier than the average F4 tornado from 1977 to 2006. Evaluation of an extended violent tornado [(E)F4–(E)F5] climatology back to 1880 reveals that the current frequency of (E)F4–(E)F5 ratings is the lowest it has been across the entire extended 144-yr period, and depending on the analysis method, the recent fatality rate for a typical EF4 tornado is the highest fatality rate associated with (E)F4 tornadoes in the past 55–72 years. Significance Statement When the enhanced Fujita scale replaced the Fujita scale in 2007, one goal was to maintain consistency across the records of ratings applied to tornadoes across the two scales. However, for tornadoes rated (E)F2 or greater, a larger fraction is currently rated EF2 than in the last 30 years of the Fujita-scale era, with smaller fractions of EF3s and EF4s and no EF5s since 2013. The frequency of EF4 and EF5 ratings over the past 10 years is the lowest it has been across 144 years of record. This downshift in higher ratings may also be partially responsible for more fatalities being observed per EF3–EF4 tornado today than in the last 30 years of the Fujita-scale era.

Abstract Despite decades of study, much remains unknown about the nature and fall behavior of natural hailstones. This incomplete understanding of the spatiotemporal distribution of hailfall and hail characteristics is limiting improvements in radar-based hail detection and quantification algorithms. Similarly, uncertainties about how hailstones fall in nature (e.g., the “tumbling” nature of falling hail) increase uncertainties in forward operators and numerical models. Herein, we introduce a new high-resolution (12.6 megapixel), high-speed (∼330 frames per second), stereoscopic imager designed to quantify the size, shape, and fall behavior of naturally occurring hailstones. Images collected in 2022 show that fine-scale features associated with falling hail are resolved, including the water torus around melting hailstones and lobes when present. Other microphysical processes such as drop shedding, breakup of large drops, and hydrometeor collisions are also observed. The stereoscopic nature of the imager allows us to estimate the size and, when tracked through time, the three-dimensional velocity of hydrometeors. The high frame rate allows estimation of rotation/gyration rates, and some sampled hailstones rotate at rates greater than 50 Hz. With this imager, we expect to (1) improve radar-based hail algorithms by providing a more complete record/sample of hydrometeor characteristics (e.g., hail sizes, start and stop times of hail fall, etc.), (2) quantify the canting angle distribution of and meltwater distribution on falling hailstones to improve polarimetric radar forward operators, and (3) provide direct observations of real hailstones and other large hydrometeors to improve numerical models.