Browse

You are looking at 51 - 60 of 3,259 items for :

  • Weather and Forecasting x
  • Refine by Access: All Content x
Clear All
Ryan A. Sobash
and
David A. Ahijevych

Abstract

The High Resolution Rapid Refresh (HRRR) model provides hourly-updating forecasts of convective-scale phenomena, which can be used to infer the potential for convective hazards (e.g., tornadoes, hail, and wind gusts), across the United States. We used deterministic 2019–2020 HRRR version 4 (HRRRv4) forecasts to train neural networks (NNs) to generate 4-hourly probabilistic convective hazard forecasts (NNPFs) for HRRRv4 initializations in 2021, using storm reports as ground truth. The NNPFs were compared to the skill of a smoothed updraft helicity (UH) baseline to quantify the benefit of the NNs. NNPF skill varied by initialization time and time of day, but were all superior to the UH forecast. NNPFs valid at hours between 18 UTC – 00 UTC were most skillful in aggregate, significantly exceeding the baseline forecast skill. Overnight NNPFs (i.e., valid 06–12 UTC) were least skillful, indicating a diurnal cycle in hazard predictability that was present across all HRRRv4 initializations. We explored the sensitivity of HRRRv4 NNPF skill to NN training choices. Including an additional year of 2021 HRRRv4 forecasts for training slightly improved skill for 2022 HRRRv4 NNPFs, while reducing the training dataset size by 40% using only forecasts with storm reports was not detrimental to forecast skill. Finally, NNs trained with 2018–2020 HRRRv3 forecasts led to a reduction in NNPF skill when applied to 2021 HRRRv4 forecasts. In addition to documenting practical predictability challenges with convective hazard prediction, these findings reinforce the need for a consistent model configuration for optimal results when training NNs and provide best practices when constructing a training dataset with operational convection-allowing model forecasts.

Restricted access
P. W. Miller
,
C. Li
,
K. Xu
,
S. Caparotta
, and
R.V. Rohli

Abstract

On 13 April 2021, a mesoscale convective system (MCS) swept across the southeastern Louisiana coast, capsizing the 39-m Seacor Power roughly 7 km from shore and leaving 13 mariners drowned or missing. In addition to the severe straight-line winds that sank the vessel, sustained surface winds >20 m s−1 behind the leading convection persisted well after the main convective band, inhibiting search and rescue efforts. Though complete historical fatality statistics are unavailable, the 13 deaths associated with this event likely represent one of the deadliest severe convective weather events in modern U.S. maritime history. This analysis integrates in-situ, remotely sensed, and reanalysis datasets to reconstruct the 2021 Seacor Power accident as well as ascertain its depiction in day-of operational convection-allowing model (CAM) guidance. Results suggest that the MCS formed along an unanalyzed coastal boundary and developed a strong meso-high to the east of the wreck as it moved offshore. The resulting zonally oriented pressure gradient directed stiff easterly winds over the wreck for several hours, even as the squall line had propagated well away from the coast. This multi-hour period of severe weather along the Louisiana coast was relatively well resolved by morning-of CAM guidance, providing optimism that future such events may be anticipated with the lead times required by vulnerable sea craft to reach safe harbor. Future severe convective weather watches containing marine zones might include a “marine” section detailing the potential sea conditions, analogous to the “aviation” section in current severe weather watches.

Restricted access
Audrey Casper
,
Emma S. Nuss
,
Christine M. Baker
,
Melissa Moulton
, and
Gregory Dusek

Abstract

Rip currents, fast offshore-directed flows, are the leading cause of death and rescues on surf beaches worldwide. The National Oceanic and Atmospheric Administration (NOAA) seeks to minimize this threat by providing rip-current hazard likelihood forecasts based on environmental conditions from the Nearshore Wave Prediction System. Rip currents come in several types, including bathymetric rip currents that form when waves break on sandbars interspersed with channels and transient rip currents that form when there are breaking waves coming from multiple directions. The NOAA model was developed and tested in an area where bathymetric rip currents may be the most prevalent type of rip current. Therefore, model performance in regions where other types of rip currents (e.g., transient rip currents) may be more ubiquitous remains unknown. To investigate the efficacy of the NOAA model guidance in the context of different rip-current types, we compared modeled rip-current probabilities with physical-based parameterizations of bathymetric and transient rip-current speeds. We also compared these probabilities to lifeguard observations of bathymetric and transient rip currents from Salt Creek Beach, California, in summer and fall 2021. We found that the NOAA model skillfully predicts a wide range of hazardous parameterized bathymetric speeds but generally underpredicts hazardous transient rip-current speeds and the hazardous rip currents observed at Salt Creek Beach. Our results demonstrate how wave parameters, including directional spread, may serve as environmental indicators of rip-current hazard. By evaluating factors that influence the skill of modeled rip-current predictions, we strive toward improved rip-current hazard forecasting.

Significance Statement

The purpose of this study is to evaluate how well the NOAA rip-current hazard model predicts different rip-current types. Accurate forecasting of rip currents is important because rip currents are the leading cause of death and rescues at surf beaches worldwide. By comparing the performance of the NOAA model to parameterized rip-current speed and lifeguard observations of rip-current strength, we highlighted the model’s decreased ability to predict hazardous transient rip currents compared to hazardous bathymetric rip currents. Because bathymetric and transient rip currents are driven by different environmental conditions, an improved hazard model must be sensitive to these different conditions to predict a greater range of hazardous rip currents.

Restricted access
John D. Horel
and
James T. Powell

Abstract

While many studies have examined intense rainfall and flash flooding during the North American monsoon (NAM) in Arizona, Nevada, and New Mexico, less attention has focused on the NAM’s extension into southwestern Utah. This study relates flash flood reports and Multi-Radar Multi-Sensor (MRMS) precipitation across southwestern Utah to atmospheric moisture content and instability analyses and forecasts from the High-Resolution Rapid Refresh (HRRR) model during the 2021–23 monsoon seasons. MRMS quantitative precipitation estimates over southwestern Utah during the summer depend largely on the areal coverage from the KICX WSR-88D radar near Cedar City, Utah. Those estimates are generally consistent with the limited number of precipitation gauge reports in the region except at extended distances from the radar. A strong relationship is evident between days with widespread precipitation and afternoons with above-average precipitable water (PWAT) and convective available potential energy (CAPE) estimated from HRRR analyses across the region. Time-lagged ensembles of HRRR forecasts (initialization times from 0300 to 0600 UTC) that are 13–18 h prior to the afternoon period when convection is initiating (1800–2100 UTC) are useful for situational awareness of widespread precipitation events after adjusting for underprediction of afternoon CAPE. Improved skill is possible using random forest classification relying only on PWAT and CAPE to predict days experiencing excessive (upper quartile) precipitation. Such HRRR predictions may be useful for forecasters at the Salt Lake City National Weather Service Forecast Office to assist in issuing flash flood potential statements for visitors to national parks and other recreational areas in the region.

Significance Statement

Summer flash floods in southwestern Utah are a risk to area residents and millions of visitors annually to the region’s national parks, monuments, and recreational areas. The likelihood of flash floods within the region’s catchments depends on the intense afternoon and early evening convection initiated by lift and instability primarily due to terrain–flow interactions over elevated plateaus and mountains. Forecasts at lead times of 13–18 h of moisture and instability from the operational High-Resolution Rapid Refresh model have the potential to predict summer afternoons that are likely to have increased risks for higher rainfall amounts across southwestern Utah, although they are not expected to predict the likelihood of flash floods in any specific locale.

Open access
Timothy J. Wagner
,
Ralph A. Petersen
,
Richard D. Mamrosh
,
Jordan Gerth
,
Curtis H. Marshall
, and
James M. O’Sullivan

Abstract

Aircraft-based observations (ABOs) are an important component of the global observation system. Observations of pressure, temperature, and wind are obtained from thousands of routine commercial flights daily via the Aircraft Meteorological Data Relay (AMDAR) program, while a subset of approximately 145 aircraft globally (and 135 within the conterminous United States) also produces observations of water vapor from the Water Vapor Sensing System–II (WVSS–II). Aircraft equipped with WVSS–II provide the basic parameters as radiosonde observations throughout most of the troposphere, often at higher temporal and spatial frequency. Since these aircraft are operated according to the demands of passenger and cargo, the availability of aircraft profiles varies significantly in space and time, with more profiles during daytime and early evening than overnight, more profiles on weekdays than weekends, and more during the summer months. The number of available profiles was significantly impacted by reductions in travel during the COVID-19 pandemic but has recovered substantially. The potential for aircraft profiles to support the operational radiosonde network is explored, including the effect of various spatial and temporal matching criteria. Radiosonde launches at 0000 UTC that are well aligned with aircraft profiles are found across the conterminous United States, but well-covered 1200 UTC launches are strongly biased to the east. ABO coverage of asynoptic launch times is also explored. The busiest sites usually have multiple compatible aircraft profiles at both synoptic and asynoptic times. This redundancy lends robustness to the observation network and enables forecasters to monitor atmospheric evolution more continuously throughout the day.

Significance Statement

Some commercial aircraft make the same observations as weather balloons, but there is not a good record of how frequently these observations are made at specific locations. This paper does a census of where the airplane profile observations are most likely to be found and shows where they are duplicating weather balloon observations and where they are filling in the gaps in the weather balloon network.

Restricted access
Montgomery L. Flora
,
Burkely Gallo
,
Corey K. Potvin
,
Adam J. Clark
, and
Katie Wilson

Abstract

Artificial intelligence (AI) is gaining popularity for severe weather forecasting. Recently, the authors developed an AI system using machine learning (ML) to produce probabilistic guidance for severe weather hazards, including tornadoes, large hail, and severe winds, using the National Severe Storms Laboratory’s (NSSL) Warn-on-Forecast System (WoFS) as input. Known as WoFS-ML-Severe, it performed well in retrospective cases, but its operational usefulness had yet to be determined. To examine the potential usefulness of the ML guidance, we conducted a control and treatment (experimental) group experiment during the 2022 NOAA Hazardous Weather Testbed Spring Forecasting Experiment (HWT-SFE). The control group had full access to WoFS, while the experimental group had access to WoFS and ML products. Explainability graphics were also integrated into the WoFS web viewer. Both groups issued 1-h convective outlooks for each hazard. After issuing their forecasts, we surveyed participants on their confidence, the number of products viewed, and the usefulness of the ML guidance. We found the ML-based outlooks outperformed non-ML-based outlooks for multiple verification metrics for all three hazards and were rated subjectively higher by the participants. However, the difference in confidence between the two groups was not significant, and the experimental group self-reported viewing more products than the control group. Participants had mixed sentiments toward explainability products as it improved their understanding of the input/output relationships, but viewing them added to their workload. Although the experiment demonstrated the usefulness of ML guidance for severe weather forecasting, there are avenues to improve upon the ML guidance, and more training and exposure are needed to exploit its benefits fully.

Significance Statement

We developed an artificial intelligence (AI) system to predict tornadoes, large hail, and damaging straight-line winds. The AI system was leveraged in real time during the 2022 NOAA Hazardous Weather Testbed Spring Forecasting Experiment. This study reveals that forecasters using AI guidance produced more reliable and spatially accurate outlooks than those without. While AI and complementary explainability products did not reduce forecaster workload, both demonstrated great potential for improving severe weather forecasting. This research also highlights the importance of user feedback in refining AI tools for severe weather forecasting.

Restricted access
Karl Schneider
,
Kelly Lombardo
,
Matthew R. Kumjian
, and
Kevin Bowley

Abstract

Convective snow (CS) presents a significant hazard to motorists and is one of the leading causes of weather-related fatalities on Pennsylvania roadways. Thus, understanding environmental factors promoting CS formation and organization is critical for providing relevant and accurate information to those impacted. Prior research has been limited, mainly focusing on frontal CS bands often called “snow squalls”; thus, these studies do not account for the diversity of CS organizational modes that is frequently observed, highlighting a need for a robust climatology of broader CS events. To identify such events, a novel, radar-based CS detection algorithm was developed and applied to WSR-88D radar data from 10 cold seasons in central Pennsylvania, during which 159 cases were identified. Distinct convective organization modes were identified: linear (frontal) snow squalls, single cells, multicells, and streamer bands. Each algorithm-flagged radar scan containing CS was manually classified as one of these modes. Interestingly, the most-studied frontal mode only occurred <5% of the time, whereas multicellular modes dominated CS occurrence. Using the times associated with each CS mode, synoptic and local environmental information from model analyses was investigated. Key characteristics of CS environments compared to null cases include a 500-hPa trough in the vicinity, lower-tropospheric conditional instability, and sufficient moisture. Environments favorable for the different CS modes featured statistically significant differences in the 500-hPa trough axis position, surface-based CAPE, and the unstable layer depth, among others. These results provide insights into forecasting CS mode, explicitly presented in a forecasting decision tree.

Significance Statement

Convective snow events such as snow squalls are a leading cause of weather-related deaths on Pennsylvania roads. Research into these events is limited, thus negatively impacting forecast skill. To better understand convective snow event frequency of occurrence, inter- and intra-annual variability, and their supporting environments, we performed a 10-yr radar-based climatology of these events. We report the results of this climatology and on the statistically significant differences in their supporting environments. The latter are used to propose a forecasting framework for convective snow, which may improve the predictability of convective snow in an operational setting.

Restricted access
Free access
Brett S. Borchardt
,
Keith D. Sherburn
, and
Russ S. Schumacher

Abstract

Identifying radar signatures indicative of damaging surface winds produced by convection remains a challenge for operational meteorologists, especially within environments characterized by strong low-level static stability and convection for which inflow is presumably entirely above the planetary boundary layer. Numerical model simulations suggest the most prevalent method through which elevated convection generates damaging surface winds is via “up–down” trajectories, where a near-surface stable layer is dynamically lifted and then dropped with little to no connection to momentum associated with the elevated convection itself. Recently, a number of unique convective episodes during which damaging surface winds were produced by apparently elevated convection coincident with mesoscale gravity waves were identified and cataloged for study. A novel radar signature indicative of damaging surface winds produced by elevated convection is introduced through six representative cases. One case is then explored further via a high-resolution model simulation and related to the conceptual model of up–down trajectories. Understanding the processes responsible for, and radar signature indicative of, damaging surface winds produced by gravity wave coincident convection will help operational forecasters identify and ultimately warn for a previously underappreciated phenomenon that poses a threat to lives and property.

Significance Statement

We identified unique radar and observational signatures of thunderstorms that produce damaging surface winds through a recently discovered mechanism. The radar and observational signatures can be used to issue warnings to protect lives and property in situations where damaging winds were previously unexpected. Key observational signatures include associated increases in surface pressure, sustained wind, and wind gust magnitudes, as well as little to no change or an increase in surface temperature. In addition, base radar data exhibit a divergence signature, including in regions of little or no detectable precipitation. Additional study is needed to answer why some atmospheric environments are supportive of the unique damaging-wind-producing mechanism while others are not.

Restricted access
Christopher Rodell
,
Rosie Howard
,
Piyush Jain
,
Nadya Moisseeva
,
Timothy Chui
, and
Roland Stull

Abstract

Wildfire agencies use fire danger rating systems (FDRSs) to deploy resources and issue public safety measures. The most widely used FDRS is the Canadian fire weather index (FWI) system, which uses weather inputs to estimate the potential for wildfires to start and spread. Current FWI forecasts provide a daily numerical value, representing potential fire severity at an assumed midafternoon time for peak fire activity. This assumption, based on typical diurnal weather patterns, is not always valid. To address this, we developed an hourly FWI (HFWI) system using numerical weather prediction. We validate HFWI against the traditional daily FWI (DFWI) by comparing HFWI forecasts with observation-derived DFWI values from 917 surface fire weather stations in western North America. Results indicate strong correlations between forecasted HFWI and the observation-derived DFWI. A positive mean bias in the daily maximum values of HFWI compared to the traditional DFWI suggests that HFWI can better capture severe fire weather variations regardless of when they occur. We confirm this by comparing HFWI with hourly fire radiative power (FRP) satellite observations for nine wildfire case studies in Canada and the United States. We demonstrate HFWI’s ability to forecast shifts in fire danger timing, especially during intensified fire activity in the late evening and early morning hours, while allowing for multiple periods of increased fire danger per day—a contrast to the conventional DFWI. This research highlights the HFWI system’s value in improving fire danger assessments and predictions, hopefully enhancing wildfire management, especially during atypical fire behavior.

Open access