@article{radford_lackmann_goodwin_correia jr_harnos_2023, title={An Iterative Approach toward Development of Ensemble Visualization Techniques for High-Impact Winter Weather Hazards}, volume={104}, ISSN={["1520-0477"]}, DOI={10.1175/BAMS-D-22-0193.1}, abstractNote={Abstract We developed five prototype convection-allowing model ensemble visualization products with the goal of improving depictions of the timing of winter weather hazards. These products are interactive, web-based plots visualizing probabilistic onset times and durations of intense snowfall rates, probabilities of heavy snow at rush hour, periods of heightened impacts, and mesoscale snowband probabilities. Prototypes were evaluated in three experimental groups coordinated by the Weather Prediction Center (WPC) Hydrometeorological Testbed (HMT), with a total of 53 National Weather Service (NWS) forecasters. Forecasters were asked to complete a simple forecast exercise for a snowfall event, with a control group using the Storm Prediction Center’s (SPC) High-Resolution Ensemble Forecast (HREF) system viewer, and an experimental group using both the HREF viewer and the five experimental graphics. Forecast accuracy was similar between the groups, but the experimental group exhibited smaller mean absolute error for snowfall duration forecasts. A series of Likert-scale questions saw participants respond favorably to all of the products and indicated that they would use them in operational forecasts and in communicating information to core partners. Forecasters also felt that the new products improved their comprehension of ensemble spread and reduced the time required to complete the forecasting exercise. Follow-up plenary discussions reiterated that there is a high demand for ensemble products of this type, though a number of potential improvements, such as greater customizability, were suggested. Ultimately, we demonstrated that social science methods can be effectively employed in the atmospheric sciences to yield improved visualization products.}, number={9}, journal={BULLETIN OF THE AMERICAN METEOROLOGICAL SOCIETY}, author={Radford, Jacob T. and Lackmann, Gary M. and Goodwin, Jean and Correia Jr, James and Harnos, Kirstin}, year={2023}, month={Sep}, pages={E1649–E1669} } @article{radford_lackmann_goodwin_correia jr_harnos_2023, title={An Iterative Approach toward Development of Ensemble Visualization Techniques for High-Impact Winter Weather Hazards Part I: Product Development}, volume={104}, ISSN={["1520-0477"]}, DOI={10.1175/BAMS-D-22-0192.1}, abstractNote={Abstract We applied social science research principles to develop a suite of probabilistic winter weather forecasting visualizations for High-Resolution Ensemble Forecast (HREF) system output. This was achieved through an iterative, dialogic process with U.S. National Weather Service (NWS) forecasters to design nine new web-based, interactive products aimed toward improving visualizations of winter weather event magnitudes, characteristics, and timing. These products were influenced by feedback from a preliminary focus group, which emphasized the importance of product credibility, contextualization, and scalability. In a follow-up discussion, winter weather forecasting experts found the event timing products to have the greatest utility due to their association with impact-decision support services (IDSS). Furthermore, forecasters assessed snowfall rates as the most impactful variable rather than snowfall totals and radar reflectivity. The timing products include plots of probabilistic snowfall onset time and duration, rush hour intersection probabilities, and a combination meteogram. The onset and duration plots visualize the ensemble-average onset time and duration of a specified snowfall rate, as demonstrated in previous works, but with the addition of uncertainty information by visualizing the earliest, most likely, and latest potential onset times as well as the shortest, most likely, and longest potential durations. The rush hour product visualizes the probability of exceeding a specified snowfall rate during local commutes, and the combination meteogram allows rapid identification of high-impact periods by encoding probabilities of precipitation, precipitation-type probabilities, and average rates into one graphical tool. Examples of these interactive products are maintained on our companion website: www.visweather.com/bams2023.}, number={9}, journal={BULLETIN OF THE AMERICAN METEOROLOGICAL SOCIETY}, author={Radford, Jacob T. and Lackmann, Gary M. and Goodwin, Jean and Correia Jr, James and Harnos, Kirstin}, year={2023}, month={Sep}, pages={E1630–E1648} } @article{radford_lackmann_2023, title={Assessing Variations in the Predictive Skill of Ensemble Snowband Forecasts with Object-Oriented Verification and Self-Organizing Maps}, volume={38}, ISSN={["1520-0434"]}, DOI={10.1175/WAF-D-23-0004.1}, abstractNote={Abstract We used object-oriented verification and self-organizing maps (SOMs) to identify patterns in environmental parameters correlating with mesoscale snowband predictive skill by the High-Resolution Ensemble Forecast (HREF) system between 2017 and 2022. First, HREF snowband forecasts for 305 banding events were verified based on similarities between forecast and observed feature properties. HREF members performed comparably, demonstrating large positional errors, but the non-time-lagged High-Resolution Rapid Refresh member demonstrated the greatest overall skill. Observed banding events were clustered by 500-hPa geopotential height anomalies, mean sea level pressure, vertical velocity, frontogenesis, and saturation equivalent potential vorticity from the European Centre for Medium-Range Weather Forecasts Reanalysis version 5 using SOMs. Clusters reaffirmed the presence of midlevel frontogenesis, ascent, and reduced stability in most banding cases, and the predominant synoptic environments conducive to band development. Clusters were compared to determine whether patterns in the variables were correlated with predictive skill. Strength of upward motion was correlated with skill, with the strongest upward motion cases verifying 10% better than the weakest upward motion cases due to smaller positional error. Additionally, events with a single region of strong upward motion verified better than events with disorganized, but comparably intense, upward motion. The magnitude of frontogenesis was uncorrelated with skill, but events with more upright frontogenesis collocated with the band centroid were better predicted than events with shallower slopes and low-level frontogenesis displaced toward warmer air. The skill variance associated with different vertical motion magnitudes could assist forecasters in modulating forecast confidence, while the most common types of errors documented here may be beneficial to model developers in refining HREF member snowfall forecasts. Significance Statement High-resolution numerical weather prediction (NWP) models generally have limited predictive skill for mesoscale snowband forecasts. Even so, some snowbands are forecast by NWP models with much greater skill than others. In this work, we apply artificial intelligence to group snowband events based on atmospheric conditions and then determine whether different groups are easier or harder for models to predict. Identification of these groups could help forecasters know when to trust or be skeptical of NWP output and help developers improve snowband formation processes in NWP models.}, number={9}, journal={WEATHER AND FORECASTING}, author={Radford, Jacob T. and Lackmann, Gary M.}, year={2023}, month={Sep}, pages={1673–1693} } @article{radford_lackmann_2023, title={Improving High-Resolution Ensemble Forecast (HREF) System Mesoscale Snowband Forecasts with Random Forests}, volume={38}, ISSN={["1520-0434"]}, DOI={10.1175/WAF-D-23-0005.1}, abstractNote={Abstract Mesoscale snowbands are impactful winter weather phenomena but are challenging to predict due to small-scale forcings and ingredients. Previous work has established that even deterministic convection-allowing models (CAMs) often struggle to represent these features with much precision and recommended the application of ingredients-based or probabilistic forecast strategies. Based on these recommendations, we develop and evaluate four different models for forecasting snowbands. The first model, referred to as the “HREF threshold probability” model, detects band development in High-Resolution Ensemble Forecast (HREF) system members’ 1000-m simulated reflectivities, then uses these detections to calculate a snowband probability. The second model is a random forest incorporating features explicitly linked to snowbands, such as the detection of bands in each HREF member and statistical summaries of simulated reflectivity and the categorical snow field. The third model is a random forest model incorporating snowband ingredients, such as midtropospheric frontogenesis, moist symmetric stability, and vertical velocity. Last, the fourth model combines the features of the explicit and implicit random forests. Binary band predictions based upon the HREF threshold probability model resulted in a critical success index 27% higher than the average HREF member. The explicit feature random forest model further improved performance by an additional 11%, with statistics of the reflectivity field holding the most predictive value. The implicit and combined random forests slightly underperformed the explicit random forest, perhaps due to a large number of noisy, correlated features. Ultimately, we demonstrate that simple probabilistic snowband forecasting strategies can yield substantial improvements over deterministic CAMs. Significance Statement Mesoscale snowbands have the potential for major societal impacts but are difficult to predict due to their small spatial scales. Previous work has shown that individual high-resolution numerical weather prediction (NWP) models struggle to predict whether or not a snowband will occur. In this work, we evaluate whether a probabilistic forecast strategy using high-resolution ensemble NWP output leads to improved snowband forecasts, and whether we can gain additional predictive skill by combining this output with artificial intelligence (AI) methods. AI can also help us understand the environmental factors associated with snowbands and compare environmental importance in forecasting to just using the model output snowfall forecasts explicitly.}, number={9}, journal={WEATHER AND FORECASTING}, author={Radford, Jacob T. and Lackmann, Gary M.}, year={2023}, month={Sep}, pages={1695–1706} } @article{radford_lackmann_baxter_2019, title={An Evaluation of Snowband Predictability in the High-Resolution Rapid Refresh}, volume={34}, ISSN={["1520-0434"]}, DOI={10.1175/WAF-D-19-0089.1}, abstractNote={Abstract Narrow regions of intense, banded snowfall present hazardous travel conditions due to rapid onset, high precipitation rates, and lowered visibility. Despite their importance, there are few verification studies of snowbands in operational forecast models. The objective of this study is to evaluate the ability of the High-Resolution Rapid Refresh (HRRR) model to predict snowbands in the United States east of the Rocky Mountains. An automated band-detection algorithm was applied to a 3-yr period of simulated and observed radar reflectivity to compare snowband climatologies. This algorithm uses the distributions of reflectivities in contiguous precipitation regions to determine a band intensity threshold. The predictability of snowbands on a case-by-case basis was also evaluated using an object-oriented approach. The distribution of HRRR forecast banding resembles that of the observations, but with a significant positive frequency bias. This may partially be due to underrepresentation of observed bands in our verification dataset due to limited radar coverage in portions of the central United States. On a case-by-case basis, traditional skill metrics indicate limited predictability, but allowing for small timing discrepancies dramatically improves scores. Object-oriented verification yields mixed results, with 30% of forecasts receiving a score indicative of a well-predicted event. However, 69% of cases have at least one forecast lead demonstrating skill, suggesting the HRRR is successful in depicting environments conducive to band formation. These results suggest adopting a probabilistic, ensemble approach, and indicate that the deterministic HRRR is best suited for the identification of regions of elevated snowband risk and not precise timing or location information.}, number={5}, journal={WEATHER AND FORECASTING}, author={Radford, Jacob T. and Lackmann, Gary M. and Baxter, Martin A.}, year={2019}, month={Oct}, pages={1477–1494} }