Convective Outlooks Research Articles

Short and Medium-Range Predictability of Warm-Season Derechos. Part II: Convection-Allowing Ensemble Forecasts

Abstract This study explores convection-allowing ensemble (CAE) forecasts of a subset of 24 warm-season (May–August) progressive derechos from 2012 to 2022. The derecho cases were stratified into low predictability and moderate predictability based on the performance of Storm Prediction Center’s Convective Outlooks. The Model for Prediction Across Scales (MPAS) is used with a global mesh of varying horizontal resolution, with 60 km spacing over most of the globe decreasing to 3 km spacing where the derechos occurred. A 10-member ensemble was initialized at 0000 UTC from 5 days before the derecho (D5) to the day of the event (D1). The CAE is evaluated using objective metrics as well as subjective assessments of convective mode, coverage, initiation timing, and spatial displacement of simulated convection. The objective evaluation indicates that the MPAS CAE has skill in predicting severe wind gusts, even in the medium range (3–5 days before). The skill diminishes faster with lead time in low-predictability cases compared to moderate-predictability cases. Overall, more than half of the ensemble members generated a sustained, progressive bowing mesoscale convective system (MCS) on D1 in both low- and moderate-predictability cases, but the percentages decrease substantially for forecasts with progressively greater lead time. Furthermore, the environmental differences among ensemble members that produce bowing MCSs, multicell clusters, and supercells are minimal. These results indicate that correctly predicting sustained bowing MCSs prior to observed derechos is challenging for the MPAS CAE for lead times longer than 1–2 days, reflecting convective-scale complexity and predictability limits.

Analysis of a Case of a Classical Supercell Storm in Bihar, India: Observation and Tracking

In this study, we use the S-band Doppler Weather Radar to analyze the weather of a prolonged classical supercell storm that occurred on April 7, 2018, across the Indian state of Bihar. In the early morning hours of April 7th, 2018, a supercell storm with its origins in a colliding cloud mass produced in the Himalayan foothills invaded the Himalayan foothills from the North-West of Bihar via East Uttar Pradesh. The echo top increased to over 14 kilometers as it moved through the Bihar districts of Siwan and Gopalganj. As it followed, the storm shifted direction, heading northwest. Maximum radar reflectivity reached up to 61 dBz at 11: 32 IST observation (seen from Doppler Weather Radar, Patna), which may be the highest reflectivity ever recorded at DWR (Doppler Weather Radar) Station, Patna. It stopped being a supercell at 13: 30 IST and convert into a multicellular storm. A strange hook echo could be seen off the Storm's back in the Vertical Integrated Liquid profile. The supercell had two linked outflows, one to the northeast and one to the southeast. The Supercell's inverted "V"-shaped front flank was another characteristic feature. Within a small vertical band along the echo wall, reflectance reached over 60 dB at its highest. There was significant divergence near the peak of the supercell storm, with the speed differential between the updraft and downdraft reaching around 60 m/s. The system lasted for about 7 to 8 hours, damaging hailstorms occurred often along the path. Hail stones larger than 6-7 mm in diameter were spotted (as per the observed report). This particular cell was determined to be a supercell based on its internal structure, reflectivity, duration, and ground-level weather pattern. The incidence, timing, and development of the storms were all accurately predicted by the convective outlook products 2 to 3 hours in advance.

Exploring the Usefulness of Machine Learning Severe Weather Guidance in the Warn-on-Forecast System: Results from the 2022 NOAA Hazardous Weather Testbed Spring Forecasting Experiment

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.

The effect of modifying the Storm Prediction Center Convective Outlook scale on perceived risk level

AbstractThe National Weather Service Storm Prediction Center Convective Outlook (CO) is a flagship probability‐derived forecast product in the United States (US). Although much attention has been devoted to how the terminology and risk category labels and colours in the CO are ordered, no study has yet to evaluate how end‐user risk perception is affected by switching the colour scales from discrete to continuous. Results from an online survey of 1409 US adults show that perceived level of risk increases when the colour scale is modified from discrete to continuous, even if the underlying forecast data do not change.

Uwarunkowania meteorologiczne silnych burz - analiza przypadku z 10 lipca 2020 roku

Considered the most dangerous type of thunderstorm, supercell thunderstorms appear in strictly defined meteorological conditions in the troposphere. Predicting their location is particularly problematic in an environment of strong vertical wind shear and low instability. An example of such a situation is the supercell that contributed to the development of the tornado on July 10, 2020 at 18:10 CEST in Ustronie Morskie. On that day, two dangerous supercells were also created, which, at the peak of activity, passed over Lubusz and Masovian Provinces. The tornado in Ustronie Morskie was formed on the wavy front, while the supercell developing in Lubusz and Mazovia Land was observed on the cold front. Large hail and damaging wind gusts associated with the downburst in the west of the country took place in higher atmospheric instability than the other cases. The MLCAPE was over 200 J kg–1 higher than that for peak activity of the other storms discussed in the paper. This could have been crucial for the formation of the largest hail recorded on July 10, 2020 in Poland. The highest values of wind shear and storm relative helicity occurred on the coast which could have influenced the formation of a tornado despite the low values of atmospheric instability. The hook echo radar signature in Mazovia Land developed with a similar MLCAPE as the tornado on the coast (nearly 500 J kg–1). On the other hand, the MUCAPE and SBCAPE was much higher there (approx. 500 J kg–1 more), and so was the LCL level. This proves a higher humidity deficit in the lower troposphere in the place where the hook echo passes. With regard to the ESTOFEX and Skywarn Polska convection outlook, the tornado was the most difficult phenomenon to predict. Hail and damaging wind gusts not related to rotation coincided to a greater extent with the convective outlook.

Assessing Public Interpretation of Original and Linguist-Suggested SPC Risk Categories in Spanish

Abstract Recent work has shown that the words used in the Storm Prediction Center’s convective outlook are not easily understood by members of the public. Furthermore, Spanish translations of the outlook information have also been shown to have interpretation challenges. This study uses survey data collected from the Severe Weather and Society Spanish Survey, a survey of Spanish speakers across the United States, to evaluate how U.S. residents receive, understand, and respond to weather forecasts and warnings. For this experiment, respondents were tasked with ranking the words and colors used in the SPC’s convective outlook. They were randomly assigned either 1) the words originally used by the SPC for Spanish translations or 2) a set of words suggested by linguistic experts familiar with Spanish dialects in the United States. We find Spanish speakers have similar challenges to English speakers when ordering the words the SPC uses. When using the translations proposed by the linguistic experts, we find the majority of Spanish speakers ranked the words in the intended order of associated risk. Spanish speakers also displayed similar ranking distributions for the colors in the outlook as English speakers, where both groups ranked red as the highest level of risk. These findings suggest the original translations used by the SPC convective outlook create barriers for Spanish speakers and that the expert translations more effectively communicate severe weather hazards to Spanish-speaking members of the public. Significance Statement The SPC’s convective outlook provides important information about the risk posed by severe storms to members of the public. While the SPC had official Spanish translations for the categorical labels used in the outlook, it was believed anecdotally that there was a disconnect between the words the SPC was using and the way the translated outlook was being interpreted by Spanish-speaking members of the public. This work verifies previous beliefs about the original translation set and confirms the reliability of a new set of translations developed by linguistic experts among Spanish-speaking members of the public.

Exploring the Differences in SPC Convective Outlook Interpretation Using Categorical and Numeric Information

Abstract While previous work has shown that the Storm Prediction Center (SPC) convective outlooks accurately capture meteorological outcomes, evidence suggests stakeholders and the public may misinterpret the categorical words currently used in the product. This work attempts to address this problem by investigating public reactions to alternative information formats that include the following numeric information: 1) numeric risk levels (i.e., “Level 2 of 5”) and 2) numeric probabilities (i.e., “a 5% chance”). In addition, it explores how different combinations of the categorical labels with numeric information may impact public reactions to the product. Survey data comes from the 2020 Severe Weather and Society Survey, a nationally representative survey of U.S. adults. Participants were shown varying combinations of the information formats of interest, and then rated their concern about the weather and the likelihood of changing plans in response to the given information. Results indicate that providing numeric information (in the form of levels or probabilities) increases the likelihood of participants correctly interpreting the convective outlook information relative to categorical labels alone. Including the categorical labels increases misinterpretation, regardless of whether numeric information was included alongside the labels. Finally, findings indicate participants’ numeracy (or their ability to understand and work with numbers) had an impact on correct interpretation of the order of the outlook labels. Although there are many challenges to correctly interpreting the SPC convective outlook, using only numeric labels instead of the current categorical labels may be a relatively straightforward change that could improve public interpretation of the product. Significance Statement The SPC convective outlook contains vital information that can help people prepare for a severe weather event. The categorical labels in this product are often ordered incorrectly by members of the public. This work shows using numeric levels or probabilities reduces the tendency for people to order the levels incorrectly.

Evaluation of an experimental Warn‐on‐Forecast 3DVAR analysis and forecast system on quasi‐real‐time short‐term forecasts of high‐impact weather events

AbstractThe purpose of this study is to demonstrate the capability of an experimental, weather‐adaptive, high‐resolution, deterministic Warn‐on‐Forecast (WoF) analysis and forecast system (WoF3DVAR‐AFS) for predicting high‐impact severe weather events that occurred during the Hazardous Weather Testbed 2019 Spring Forecast Experiments. WoF3DVAR‐AFS uses a three‐dimensional variational (3DVAR) method as its core data assimilation system and the Advanced Research Version of the Weather Research and Forecasting (WRF‐ARW) model as its forward model. Surface measurements provided in meteorological aviation reports and the Oklahoma Mesonet, Doppler radar data, and spaceborne total lightning observations provided by the Geostationary Lightning Mapper are assimilated at 15‐min frequency over a target domain determined by the “Day 1” Convective Outlook product from the Storm Prediction Center. The chief goal of this system is to complement probabilistic forecasts generated by ensemble analysis and forecast systems, such as the experimental Warn‐on‐Forecast System (WoFS) with a higher‐resolution deterministic member to aid forecasters' decision‐making. We performed both qualitative and quantitative evaluations on 0–6 hr forecasts launched hourly from 1900 to 0300 UTC the next day for each of the 12 cases. Aggregated subjective forecast evaluation metrics from each individual case, as well as detailed comparison against available verification datasets, suggest that the forecasts are generally skillful in terms of composite reflectivity fields, quantitative precipitation forecasts, and the strength and location of rotation tracks and damaging winds. This study presents initial efforts to assess the performance of WoF3DVAR‐AFS and provides possible directions for further improvements, including the development of a weather‐adaptive, dual‐resolution analysis and forecast system hybrid with an ensemble system, such as the experimental Warn‐on‐Forecast system.

Colorful Language: Investigating Public Interpretation of the Storm Prediction Center Convective Outlook

AbstractAlthough severe weather forecast products, such as the Storm Prediction Center (SPC) convective outlook, are much more accurate than climatology at day-to-week time scales, tornadoes and severe thunderstorms claim dozens of lives and cause billions of dollars in damage every year. While the accuracy of this outlook has been well documented, less work has been done to explore the comprehension of the product by non-expert users like the general public. This study seeks to fill this key knowledge gap by collecting data from a representative survey of U.S. adults in the lower 48 states about their use and interpretation of the SPC convective outlook. Participants in this study were asked to rank the words and colors used in the outlook from least to greatest risk, and their answers were compared through visualizations and statistical tests across multiple demographics. Results show that the US public ranks the outlook colors similarly to their ordering in the outlook but switch the positions of several of the outlook words as compared to the operational product. Logistic regression models also reveal that more numerate individuals more correctly rank the SPC outlook words and colors. These findings suggest that the words used in the convective outlook may confuse non-expert users, and that future work should continue to use input from public surveys to test potential improvements in the choice of outlook words. Using more easily understood words may help to increase the outlook’s decision support value and potentially reduce the harm caused by severe weather events.

The Impact of the Storm Prediction Center's Convective Outlooks and Watches on Emergency Management Operational Planning

For many emergency managers (EMs) and National Weather Service (NWS) forecasters, Convective Outlooks issued by the Storm Prediction Center (SPC) influence the preparation for near-term severe weather events. However, research into how and when EMs utilize that information, and how it influences their emergency operations plan, is limited. Therefore, to better understand how SPC Convective Outlooks are used for severe weather planning, a survey was conducted of NWS core partners in the emergency management sector. The results show EMs prefer to wait until an Enhanced Risk for severe thunderstorms is issued to prepare for severe weather. In addition, the Day 2 Convective Outlook serves as the threshold for higher, value-based decision making. The survey was also used to analyze how the issuance of different risk levels in SPC Convective Outlooks impact emergency management preparedness compared to preparations conducted when a Convective Watch is issued.

Generating Probabilistic Next-Day Severe Weather Forecasts from Convection-Allowing Ensembles Using Random Forests

AbstractExtracting explicit severe weather forecast guidance from convection-allowing ensembles (CAEs) is challenging since CAEs cannot directly simulate individual severe weather hazards. Currently, CAE-based severe weather probabilities must be inferred from one or more storm-related variables, which may require extensive calibration and/or contain limited information. Machine learning (ML) offers a way to obtain severe weather forecast probabilities from CAEs by relating CAE forecast variables to observed severe weather reports. This paper develops and verifies a random forest (RF)-based ML method for creating day 1 (1200–1200 UTC) severe weather hazard probabilities and categorical outlooks based on 0000 UTC Storm-Scale Ensemble of Opportunity (SSEO) forecast data and observed Storm Prediction Center (SPC) storm reports. RF forecast probabilities are compared against severe weather forecasts from calibrated SSEO 2–5-km updraft helicity (UH) forecasts and SPC convective outlooks issued at 0600 UTC. Continuous RF probabilities routinely have the highest Brier skill scores (BSSs), regardless of whether the forecasts are evaluated over the full domain or regional/seasonal subsets. Even when RF probabilities are truncated at the probability levels issued by the SPC, the RF forecasts often have BSSs better than or comparable to corresponding UH and SPC forecasts. Relative to the UH and SPC forecasts, the RF approach performs best for severe wind and hail prediction during the spring and summer (i.e., March–August). Overall, it is concluded that the RF method presented here provides skillful, reliable CAE-derived severe weather probabilities that may be useful to severe weather forecasters and decision-makers.

Forecasting Severe Weather with Random Forests

AbstractUsing nine years of historical forecasts spanning April 2003–April 2012 from NOAA’s Second Generation Global Ensemble Forecast System Reforecast (GEFS/R) ensemble, random forest (RF) models are trained to make probabilistic predictions of severe weather across the contiguous United States (CONUS) at Days 1–3, with separate models for tornado, hail, and severe wind prediction at Day 1 in an analogous fashion to the Storm Prediction Center’s (SPC’s) convective outlooks. Separate models are also trained for the western, central, and eastern CONUS. Input predictors include fields associated with severe weather prediction, including CAPE, CIN, wind shear, and numerous other variables. Predictor inputs incorporate the simulated spatiotemporal evolution of these atmospheric fields throughout the forecast period in the vicinity of the forecast point. These trained RF models are applied to unseen inputs from April 2012 to December 2016, and their forecasts are evaluated alongside the equivalent SPC outlooks. The RFs objectively make statistical deductions about the relationships between various simulated atmospheric fields and observations of different severe weather phenomena that accord with the community’s physical understandings about severe weather forecasting. Using these quantified flow-dependent relationships, the RF outlooks are found to produce calibrated probabilistic forecasts that slightly underperform SPC outlooks at Day 1, but significantly outperform their outlooks at Days 2 and 3. In all cases, a blend of the SPC and RF outlooks significantly outperforms the SPC outlooks alone, suggesting that use of RFs can improve operational severe weather forecasting throughout the Day 1–3 period.

Hindcasting the First Tornado Forecast in Europe: 25 June 1967

Abstract The tornado outbreak of 24–25 June 1967 was the most damaging in the history of western Europe, producing 7 F2–F5 tornadoes, 232 injuries, and 15 fatalities across France, Belgium, and the Netherlands. Following tornadoes in France on 24 June, the Royal Netherlands Meteorological Institute (KNMI) issued a tornado forecast for 25 June, which became the first ever—and first verified—tornado forecast in Europe. Fifty-two years later, tornadoes are still not usually forecast by most European national meteorological services, and a pan-European counterpart to the NOAA/NWS/Storm Prediction Center (SPC) does not exist to provide convective outlook guidance; yet, tornadoes remain an extant threat. This article asks, “What would a modern-day forecast of the 24–25 June 1967 outbreak look like?” To answer this question, a model simulation of the event is used in three ways: 20-km grid-spacing output to produce a SPC-style convective outlook provided by the European Storm Forecast Experiment (ESTOFEX), 800-m grid-spacing output to analyze simulated reflectivity and surface winds in a nowcasting analog, and 800-m grid-spacing output to produce storm-total footprints of updraft helicity maxima to compare to observed tornado tracks. The model simulates a large supercell on 24 June and weaker embedded mesocyclones on 25 June forming along a stationary front, allowing the ESTOFEX outlooks to correctly identify the threat. Updraft helicity footprints indicate multiple mesocyclones on both days within 40–50 km and 3–4 h of observed tornado tracks, demonstrating the ability to hindcast a large European tornado outbreak.

Perception and Comprehension of the Extended Forecast Graphic: A Survey of Broadcast Meteorologists and the Public

AbstractThere have been multiple efforts in recent years to simplify visual weather forecast products, with the goal of more efficient risk communication for the general public. Many meteorological forecast products, such as the cone of uncertainty, storm surge graphics, warning polygons, and Storm Prediction Center (SPC) convective outlooks, have created varying levels of public confusion resulting in revisions, modifications, and improvements. However, the perception and comprehension of private weather graphics produced by television stations has been largely overlooked in peer-reviewed research. The goal of this study is to explore how the extended forecast graphic, more commonly known as the 7, 10 day, etc., is utilized by broadcasters and understood by the public. Data were gathered from surveys with the general public and also from broadcast meteorologists. Results suggest this graphic is a source of confusion and highlights a disconnect between the meteorologists producing the graphic and the content prioritized by their audiences. Specifically, timing and intensity of any precipitation or adverse weather events are the two most important variables to consider from the viewpoint of the public. These variables are generally absent from the extended forecast graphic, thus forcing the public to draw their own conclusions, which may differ from what the meteorologist intends to convey. Other results suggest the placement of forecast high and low temperatures, use of probability of precipitation, icon inconsistency, and length of time the graphic is shown also contribute to public confusion and misunderstanding.

An Analysis of Subdaily Severe Thunderstorm Probabilities for the United States

Abstract One of the challenges of providing probabilistic information on a multitude of spatiotemporal scales is ensuring that information is both accurate and useful to decision-makers. Focusing on larger spatiotemporal scales (i.e., from convective outlook to weather watch scales), historical severe weather reports are analyzed to begin to understand the spatiotemporal scales that hazardous weather events are contained within. Reports from the Storm Prediction Center’s report archive are placed onto grids of differing spatial scales and then split into 24-h convective outlook days (1200–1200 UTC). These grids are then analyzed temporally to assess over what fraction of the day a single location would generally experience severe weather events. Different combinations of temporal and spatial scales are tested to determine how the reference class (or the choice of what scales to use) alters the probabilities of severe weather events. Results indicate that at any given point in the United States on any given day, more than 95% of the daily reports within 40 km of the point occur in a 4-h period. Therefore, the SPC 24-h convective outlook probabilities can be interpreted as 4-h convective outlook probabilities without a significant change in meaning. Additionally, probabilities and threat periods are analyzed at each location and different times of year. These results indicate little variability in the duration of severe weather events, which allows for a consistent definition of an “event” for all locations in the continental United States.

The Influence of Tornado Activity, Impact, Memory, and Sentiment on Tornado Perception Accuracy among College Students

A survey consisting of open-ended and closed responses was administered at three universities in the eastern USA. The home counties of survey participants represented climatological tornado risks spanning from rarely impacted to frequently impacted. The first objective of this research was to classify climatological tornado risk for each county so that analyses of tornado perception accuracy could be evaluated. Perception accuracy was defined as the difference between what each participant perceived minus what actually happened. A manual classification scheme was created that uses the Storm Prediction Center’s Convective Outlook framework as county climatological risk categories. Participants from high-risk counties statistically significantly overestimated the numbers of violent tornadoes compared to participants from every risk category but moderate. Furthermore, participants from high-risk counties had significantly greater tornado impacts, thus validating the classification of high-risk. Participants from high, moderate, and slight-risk counties significantly overestimated the number of strong tornadoes compared to participants from enhanced-risk counties. There appeared to be no relationships between tornado memory and tornado sentiment with tornado perception accuracy. Possible explanations for the overestimation of the numbers of violent tornadoes in high-risk counties are discussed.

Meteorological Differences Characterizing Tornado Outbreak Forecasts of Varying Quality

Tornado outbreaks (TOs) are a major hazard to life and property for locations east of the Rocky Mountains. Improving tornado outbreak (TO) forecasts will help minimize risks associated with these major events. In this study, we present a methodology for quantifying TO forecasts of varying quality, based on Storm Prediction Center convective outlook forecasts, and provide synoptic and mesoscale composite analyses to identify important features characterizing these events. Synoptic-scale composites from the North American Regional Reanalysis (NARR) are presented for TO forecasts at three forecast quality levels, H-class (high quality), M-class (medium quality), and L-class (low quality), as well as false alarm TO forecasts. H-class and false alarm TO forecasts share many meteorological similarities, particularly in the synoptic-scale, though false alarm events show less well-defined low-level synoptic-scale features. M- and L-class TOs present environments dominated by mesoscale thermodynamic processes (particularly dryline structures), contrasting H-class TOs which are clearly synoptically driven. Simulations of these composites reveal higher instability in M- and L-class TOs that lack key kinematic structures that characterize H-class TOs. The results presented offer important forecast feedback that can help inform future TO predictions and ultimately produce improved TO forecast quality.

Evolution of WRF-HAILCAST during the 2014–16 NOAA/Hazardous Weather Testbed Spring Forecasting Experiments

Abstract Four different versions of the HAILCAST hail model have been tested as part of the 2014–16 NOAA Hazardous Weather Testbed (HWT) Spring Forecasting Experiments. HAILCAST was run as part of the National Severe Storms Laboratory (NSSL) WRF Ensemble during 2014–16 and the Community Leveraged Unified Ensemble (CLUE) in 2016. Objective verification using the Multi-Radar Multi-Sensor maximum expected size of hail (MRMS MESH) product was conducted using both object-based and neighborhood grid-based verification. Subjective verification and feedback was provided by HWT participants. Hourly maximum storm surrogate fields at a variety of thresholds and Storm Prediction Center (SPC) convective outlooks were also evaluated for comparison. HAILCAST was found to improve with each version due to feedback from the 2014–16 HWTs. The 2016 version of HAILCAST was equivalent to or exceeded the skill of the tested storm surrogates across a variety of thresholds. The post-2016 version of HAILCAST was found to improve 50-mm hail forecasts through object-based verification, but 25-mm hail forecasting ability declined as measured through neighborhood grid-based verification. The skill of the storm surrogate fields varied widely as the threshold values used to determine hail size were varied. HAILCAST was found not to require such tuning, as it produced consistent results even when used across different model configurations and horizontal grid spacings. Additionally, different storm surrogate fields performed at varying levels of skill when forecasting 25- versus 50-mm hail, hinting at the different convective modes typically associated with small versus large sizes of hail. HAILCAST was able to match results relatively consistently with the best-performing storm surrogate field across multiple hail size thresholds.

Convective Outlook: Storm Prediction

Convective Outlook: Storm Prediction

Probabilistic Verification of Storm Prediction Center Convective Outlooks

Abstract Eight years’ worth of day 1 and 4.5 years’ worth of day 2–3 probabilistic convective outlooks from the Storm Prediction Center (SPC) are converted to probability grids spanning the continental United States (CONUS). These results are then evaluated using standard probabilistic forecast metrics including the Brier skill score and reliability diagrams. Forecasts are gridded in two different ways: one with a high-resolution grid and interpolation between probability contours and another on an 80-km-spaced grid without interpolation. Overall, the highest skill is found for severe wind forecasts and the lowest skill is observed for tornadoes; for significant severe criteria, the opposite discrepancy is observed, with highest forecast skill for significant tornadoes and approximately no overall forecast skill for significant severe winds. Highest climatology-relative skill is generally observed over the central and northern Great Plains and Midwest, with the lowest—and often negative—skill seen in the West, southern Texas, and the Atlantic Southeast. No discernible year-to-year trend in skill was identified; seasonally, forecasts verified the best in the spring and late autumn and worst in the summer and early autumn. Forecasts are also evaluated in CAPE-versus-shear parameter space; forecasts struggle most in very low shear but also in high-shear, low-CAPE environments. In aggregate, forecasts for all variables verified more skillfully using interpolated probability grids, suggesting utility in interpreting forecasts as a continuous field. Forecast reliability results depend substantially on the interpretation of the forecast fields, but day 1 and day 2–3 tornado outlooks consistently exhibit an underforecast bias.