Search for: All records

Abstract Hailstorms cause billions of dollars in damage across the United States each year. Part of this cost could be reduced by increasing warning lead times. To contribute to this effort, we developed a nowcasting machine learning model that uses a 3D U-Net to produce gridded severe hail nowcasts for up to 40 min in advance. The three U-Net dimensions uniquely incorporate one temporal and two spatial dimensions. Our predictors consist of a combination of output from the National Severe Storms Laboratory Warn-on-Forecast System (WoFS) numerical weather prediction ensemble and remote sensing observations from Vaisala’s National Lightning Detection Network (NLDN). Ground truth for prediction was derived from the maximum expected size of hail calculated from the gridded NEXRAD WSR-88D radar (GridRad) dataset. Our U-Net was evaluated by comparing its test set performance against rigorous hail nowcasting baselines. These baselines included WoFS ensemble Hail and Cloud Growth Model (HAILCAST) and a logistic regression model trained on WoFS 2–5-km updraft helicity. The 3D U-Net outperformed both these baselines for all forecast period time steps. Its predictions yielded a neighborhood maximum critical success index (max CSI) of ∼0.48 and ∼0.30 at forecast minutes 20 and 40, respectively. These max CSIs exceeded the ensemble HAILCAST max CSIs by as much as ∼0.35. The NLDN observations were found to increase the U-Net performance by more than a factor of 4 at some time steps. This system has shown success when nowcasting hail during complex severe weather events, and if used in an operational environment, may prove valuable. 

Abstract Forecasting tornadogenesis remains a difficult problem in meteorology, especially for short-lived, predominantly nonsupercellular tornadic storms embedded within mesoscale convective systems (MCSs). This study compares populations of tornadic nonsupercellular MCS storm cells with their nontornadic counterparts, focusing on nontornadic storms that have similar radar characteristics to tornadic storms. Comparisons of single-polarization radar variables during storm lifetimes show that median values of low-level, midlevel, and column-maximum azimuthal shear, as well as low-level radial divergence, enable the highest degree of separation between tornadic and nontornadic storms. Focusing on low-level azimuthal shear values, null storms were randomly selected such that the distribution of null low-level azimuthal shear values matched the distribution of tornadic values. After isolating the null cases from the nontornadic population, signatures emerge in single-polarization data that enable discrimination between nontornadic and tornadic storms. In comparison, dual-polarization variables show little deviation between storm types. Tornadic storms both at tornadogenesis and at a 20-min lead time show collocation of the primary storm updraft with enhanced near-surface rotation and convergence, facilitating the nonmesocyclonic tornadogenesis processes. 

Increased wildfire events constitute a significant threat to life and property in the United States. Wildfire impact on severe storms and weather hazards is another pathway that threatens society, and our understanding of which is very limited. Here, we use unique modeling developments to explore the effects of wildfires in the western US (mainly California and Oregon) on precipitation and hail in the central US. We find that the western US wildfires notably increase the occurrences of heavy precipitation rates by 38% and significant severe hail (≥2 in.) by 34% in the central United States. Both heat and aerosols from wildfires play an important role. By enhancing surface high pressure and increasing westerly and southwesterly winds, wildfires in the western United States produce ( 1 ) stronger moisture and aerosol transport to the central United States and ( 2 ) larger wind shear and storm-relative helicity in the central United States. Both the meteorological environment more conducive to severe convective storms and increased aerosols contribute to the enhancements of heavy precipitation rates and large hail. Moreover, the local wildfires in the central US also enhance the severity of storms, but their impact is notably smaller than the impact of remote wildfires in California and Oregon because of the lessened severity of the local wildfires. As wildfires are projected to be more frequent and severe in a warmer climate, the influence of wildfires on severe weather in downwind regions may become increasingly important. 

Abstract Tropical Storm Bill produced over 400 mmof rainfall to portions of southern Oklahoma from 16-20 June 2015, adding to the catastrophic urban and river flooding that occurred throughout the region in the month prior to landfall. The unprecedented excessive precipitation event that occurred across Oklahoma and Texas during May and June 2015 resulted in anomalously high soil moisture and latent heat fluxes over the region, acting to increase the available boundary layer moisture. Tropical Storm Bill progressed inland over the region of anomalous soil moisture and latent heat fluxes which helped maintain polarimetric radar signatures associated with tropical, warm rain events. Vertical profiles of polarimetric radar variables such as Z H , Z DR , K DP , and ρ hv were analyzed in time and space over Texas and Oklahoma. The profiles suggest that Tropical Storm Bill maintained warm rain signatures and collision-coalescence processes as it tracked hundreds of kilometers inland away from the landfall point consistent with tropical cyclone precipitation characteristics. Dual-frequency precipitation radar observations from the NASA GPM DPR were also analyzed post-landfall and showed similar signatures of collision-coalescence while Bill moved over north Texas, southern Oklahoma, eastern Missouri, and western Kentucky. 

This paper synthesizes multiple methods for machine learning (ML) model interpretation and visualization (MIV) focusing on meteorological applications. ML has recently exploded in popularity in many fields, including meteorology. Although ML has been successful in meteorology, it has not been as widely accepted, primarily due to the perception that ML models are “black boxes,” meaning the ML methods are thought to take inputs and provide outputs but not to yield physically interpretable information to the user. This paper introduces and demonstrates multiple MIV techniques for both traditional ML and deep learning, to enable meteorologists to understand what ML models have learned. We discuss permutation-based predictor importance, forward and backward selection, saliency maps, class-activation maps, backward optimization, and novelty detection. We apply these methods at multiple spatiotemporal scales to tornado, hail, winter precipitation type, and convective-storm mode. By analyzing such a wide variety of applications, we intend for this work to demystify the black box of ML, offer insight in applying MIV techniques, and serve as a MIV toolbox for meteorologists and other physical scientists.