High-resolution Ensemble Forecasts Research Articles

Investigation of model forecast biases and skilful prediction for Assam heavy rainfall 2022

Extreme rainfall events (ERE) during the summer monsoon season have been occurring over most parts of India resulting in flooding and immense socio-economic loss. These extremes are becoming a frequent norm in the hilly and mountainous regions of the country such as Assam. Assam received one of the most historical EREs from 14–June 17, 2022. The present study analyses the performance of a suite of high-resolution ensemble model forecasts for this extreme event in terms of its intensity, and distribution with a lead time of up to 96 h. Furthermore, the 36 numerical experiments are carried out using two different land use and land cover (LULC) data sets (i.e. ISRO and USGS) and three different sets of parameterization schemes (i.e. planetary boundary layer, cumulus, and microphysics).Rainfall distributions in the case of USGS LULC are relatively less coherent and underestimated (60–260 mm/day) against IMD (80–300 mm/day) including the rainfall categories heavy (HR), very heavy (VHR), and extremely heavy (EHR) rainfall throughout the day-1 to day-4. Among all the ensembles (E1-E10), USGS (E6 - E10) has underestimated rainfall (140–260 mm/day) compared to ISRO (150–280 mm/day), specifically in MR and HR categories over the upper Assam (UAD) and lower Assam (LAD) divisions. Further, the Bias Correction Ensemble (BCE) technique is applied to minimize the forecast errors. A rigorous statistical analysis in terms of frequency distribution, Taylor diagram, and benchmark skill scores is carried out to elucidate the model biases. The set of the model ensembles using ISRO (E1- E5) and USGS (E6- E10) reasonably captured the HR, VHR, and EHR. In addition, throughout the forecast hour, BCE E5 (E10) is noted with the distinct realistic (underestimated) representation of model bias (5–20 %) (10–30 %) over all the subdivisions of Assam. Our results suggest that the combined efforts of ensembles of physical parameterization schemes, along with proper LULC, and the BCE approach are required to overcome challenges to improve the skills of rainfall events, particularly over complex terrains such as Assam.

Improving High-Resolution Ensemble Forecast (HREF) System Mesoscale Snowband Forecasts with Random Forests

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.

Assessing Variations in the Predictive Skill of Ensemble Snowband Forecasts with Object-Oriented Verification and Self-Organizing Maps

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.

An Iterative Approach toward Development of Ensemble Visualization Techniques for High-Impact Winter Weather Hazards: Part II: Product Evaluation

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.

An Iterative Approach toward Development of Ensemble Visualization Techniques for High-Impact Winter Weather Hazards: Part I: Product Development

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.

Integration of meteorology and geomorphology for enhanced understanding of post-fire debris-flow hazards

Through precipitation, the fields of meteorology and geomorphology are fundamentally linked, thus interdisciplinary efforts are needed to advance understanding and warning of rainfall-driven geohazards. With a focus on recent efforts specific to post-fire debris flows in California, our presentation provides an overview of the benefits and challenges of working in an interdisciplinary team of meteorologists and geomorphologists, as well as results of a recent project demonstrating advancement through the integration of these fields. In this project, we combine high-resolution ensemble precipitation forecasts with post-fire debris-flow models to explore the feasibility and potential value of providing probabilistic post-fire debris-flow hazard information over a burn scar. In sharing these examples, we emphasize the multi-benefit nature of these efforts, and encourage future interdisciplinary efforts that improve warning and mitigation of rainfall-driven geohazards.

The added value of high-resolution downscaling of the ECMWF-EPS for extreme precipitation forecasting

We have considered the most severe flood events that affected Liguria region (northwestern Italy) in the last decade, between October 2010 and October 2019. High-resolution numerical simulations of six events have been performed with the state-of-the-art non–hydrostatic Weather Research and Forecasting (WRF) mesoscale model. For each selected case study, 50 WRF simulations have been performed corresponding to the downscaling of the 50 EPS-ECMWF members. The main aim of the present work has been to answer the question on whether high-resolution ensemble forecasts bring an added value with respect to the original EPS forecasts. To answer this question, suitable statistical indices have been considered in order to assess model performance both in terms of mean properties of the ensemble and in terms of its spread. As a result of our study, the added value of WRF-based downscaling clearly emerged in relation both to the forecast of the intensity of the extreme precipitation considered in the present study and to their spatial localization. For example, a significant added value of the downscaling process emerges being the Continuous Ranked Probability Score (CRPS) skill score for the precipitation maxima well above 20 % for all considered events, with the exception of the event belonging to the year 2019 for which the improvement, although presents, is considerably lower. Nevertheless, the downscaled ensemble still remains underdisperive.

Enhancing Capacity for Short-Term Climate Change Adaptations in Agriculture in Serbia: Development of Integrated Agrometeorological Prediction System

The Integrated Agrometeorological Prediction System (IAPS) was a two-year project for the development of the long term forecast (LRF) for agricultural producers. Using LRF in decision-making, to reduce the risks and seize the opportunities, represents short-term adaptation to climate change. High-resolution ensemble forecasts (51 forecasts) were made for a period of 7 months and were initiated on the first day of each month. For the initial testing of the capacity of LRF to provide useful information for producers, 2017 was chosen as the test year as it had a very hot summer and severe drought, which caused significant impacts on agricultural production. LRF was very useful in predicting the variables which bear the memory of the longer period, such are growing degree days for the prediction of dates of the phenophases’ occurrences and the soil moisture of deeper soil layers as an indicator for the drought. Other project activities included field observations, communication with producers, web portal development, etc. Our results showed that the selected priority forecasting products were also identified by the producers as being the highest weather-related risks, the operational forecast implementation with the products designed for the use in agricultural production is proven to be urgent and necessary for decision-making, and required investments are affordable. The total cost of the full upgrade of agrometeorological climate services to meet current needs (including monitoring, seamless forecasting system development and the development of tools for information dissemination) was found to be about three orders of magnitude lower than the assessed losses in agricultural production in the two extreme years over the past decade.

Utilizing the High-Resolution Ensemble Forecast System to Produce Calibrated Probabilistic Thunderstorm Guidance

Abstract The High-Resolution Ensemble Forecast system (HREF) calibrated thunder guidance is a suite of probabilistic forecast products designed to predict the likelihood of at least one cloud-to-ground (CG) lightning flash within 20 km (12 miles) of a point during a given 1-, 4-, and 24-h time interval. This guidance takes advantage of a combination of storm attribute and environmental fields produced by the convection-allowing HREF to objectively improve upon lightning forecasts generated by the non-convection-allowing Short-Range Ensemble Forecast system (SREF). This study provides an overview of how the HREF calibrated thunder guidance was developed and calibrated to be statistically reliable against observed CG lightning flashes recorded by the National Lightning Detection Network (NLDN). Performance metrics for the 1-, 4-, and 24-h guidance are provided and compared to the respective SREF calibrated probabilistic lightning forecasts. The HREF calibrated thunder guidance has been implemented operationally within the National Weather Service and is now available to the public. Significance Statement The NOAA Storm Prediction Center has created a suite of new calibrated probabilistic thunderstorm guidance products from a convection-allowing model ensemble, the HREF. The new guidance is a notable improvement over the long-running SREF calibrated thunder guidance and is now operational across the National Weather Service.

Multivariate ensemble sensitivity analysis applied for an extreme rainfall over Indian subcontinent

Ensemble sensitivity analysis (ESA) uses sample statistics of ensemble forecasts to estimate relationships between forecast metrics and initial conditions. The ensemble sensitivity analysis is often considered as a simple univariate regression as it includes an approximation of analysis covariance matrix with corresponding diagonal elements. In this work, univariate ensemble sensitivity is extended to multivariate ensemble sensitivity that incorporates the contribution from the full covariance matrix in the sensitivity calculations. The performance of multivariate ensemble sensitivity over univariate is examined for meso- and convective scale ensemble forecasts of a heavy rainfall event that happened over the Chennai city in India in December 2015. The ensemble forecasts and analyses are generated using the Advanced Research - Weather Research and Forecasting (WRF) model Data Assimilation Research Testbed (DART) based Ensemble Kalman Filter (EnKF). Multivariate ensemble sensitivity shows organized sensitivity patterns, while the sensitivity values are found to be broadly distributed in univariate ensemble sensitivity. Both the methods are validated using a perturbed initial condition approach, and the results indicate that the multivariate ensemble sensitivity method is effective in predicting the forecast response closest to the actual model response compared to the univariate ensemble sensitivity. The impact of model error on sensitivity calculations is examined by generating a new set of ensembles that uses the Stochastic Kinetic Energy Backscatter Scheme (SKEBS). In the presence of added model error, the forecast response estimated by multivariate using SKEBS ensembles compares better with the actual response. It is found that the performance of the multivariate approach depends on the optimal choice of localization radius, and if insufficient localization is applied, the spurious long-distance correlation contaminates the performance of the multivariate ensemble sensitivity method. The impact of various forecast lead times on the univariate and multivariate ensemble sensitivity analysis indicates that responses using multivariate ensemble sensitivity are more accurate than univariate, especially at longer lead times when nonlinearity becomes significant. The performance of univariate and multivariate methods in convection-permitting scale is examined by using the high-resolution ensemble forecasts, and it is found that the multivariate sensitivity with localization substantially improves the estimates in finer scales.

A Preliminary Examination of Warm Season Precipitation Displacement Errors in the Upper Midwest in the HRRRE and HREF Ensembles

Abstract Increased operational use of convection-allowing models and ensembles offers substantial improvements for some aspects of convective weather forecasting; however, errors in quantitative precipitation forecasts (QPFs) from these models, especially those related to incorrect placement of heavy rainfall systems, limit their usefulness as an input into hydrological models. To improve understanding of QPF location errors, this study quantifies the displacement errors for the centroids of both 0–18-h accumulated rainfall and rainfall in the first hour after initiation of precipitation systems in both the High-Resolution Rapid Refresh Ensemble (HRRRE) and the High-Resolution Ensemble Forecast (HREF) for 30 events in the 2018 warm season. Ensemble member QPFs are compared to quantitative precipitation estimates (QPEs) obtained from the North Central River Forecast Center (NCRFC). HRRRE is found to have less spread in centroid locations than HREF, and both HRRRE and HREF 0–18-h QPF accumulations have less spread than the 1-h QPF accumulation when the precipitation event initiates. Furthermore, QPF centroids are most often displaced to the west in HRRRE for both 0–18-h QPF accumulation and the 1-h QPF accumulation when the precipitation event initiates. The 0–18-h QPF accumulation displacement errors can be reduced when adjustments are made to the forecasted position based upon displacement errors present in the first hour of precipitation, but only when the adjustments are a function of the intercardinal quadrant in which the initial hour QPF centroid was displaced.

Comparing and Interpreting Differently Designed Random Forests for Next-Day Severe Weather Hazard Prediction

Abstract Recent research has shown that random forests (RFs) can create skillful probabilistic severe weather hazard forecasts from numerical weather prediction (NWP) ensemble data. However, it remains unclear how RFs use NWP data and how predictors should be generated from NWP ensembles. This paper compares two methods for creating RFs for next-day severe weather prediction using simulated forecast data from the convection-allowing High-Resolution Ensemble Forecast System, version 2.1 (HREFv2.1). The first method uses predictors from individual ensemble members (IM) at the point of prediction, while the second uses ensemble mean (EM) predictors at multiple spatial points. IM and EM RFs are trained with all predictors as well as predictor subsets, and the Python module tree interpreter (TI) is used to assess RF variable importance and the relationships learned by the RFs. Results show that EM RFs have better objective skill compared to similarly configured IM RFs for all hazards, presumably because EM predictors contain less noise. In both IM and EM RFs, storm variables are found to be most important, followed by index and environment variables. Interestingly, RFs created from storm and index variables tend to produce forecasts with greater or equal skill than those from the all-predictor RFs. TI analysis shows that the RFs emphasize different predictors for different hazards in a way that makes physical sense. Further, TI shows that RFs create calibrated hazard probabilities based on complex, multivariate relationships that go well beyond thresholding 2–5-km updraft helicity.

Exploring Hourly Updating Probabilistic Guidance in the 2021 Spring Forecasting Experiment with Objective and Subjective Verification

Abstract This study explored how forecasters can best use the two main forms of operational convection-allowing model guidance: the High-Resolution Ensemble Forecast (HREF) system and the hourly High-Resolution Rapid Refresh (HRRR). The former represents a wider range of possible outcomes, but the latter updates much more frequently and incorporates newer observations. HREF and time-lagged High-Resolution Rapid Refresh (HRRR-TL) probabilistic forecasts of reflectivity and updraft helicity, as well as two methods of combining HREF and HRRR into hourly updating blended guidance, were evaluated for the 2021 Spring Forecasting Experiment (SFE) period. In both objective skill and the subjective ratings of SFE participants, the 1200 UTC HREF proved difficult to outperform over this sample of events, even when incorporating HRRR initializations as late as 1800 UTC. It was usually better to use either of the experimental blending techniques than to simply discard the older HREF in favor of newer HRRR solutions. The greater model diversity and dispersion of solutions within the HREF is likely primarily responsible for this result. A possible bias in diurnal convection initiation timing and coverage in the newly upgraded HRRRv4 was also investigated, including on subdomains targeted to weakly forced diurnal initiation, and was found to have little or no systematic effect on HRRRv4’s operational utility.

A Progress Report on the Development of the High-Resolution Rapid Refresh Ensemble

AbstractThe High-Resolution Rapid Refresh Ensemble (HRRRE) is a 36-member ensemble analysis system with 9 forecast members that utilizes the Advanced Research version of the Weather Research and Forecasting (ARW-WRF) dynamic core and the physics suite from the operational Rapid Refresh/High-Resolution Rapid Refresh deterministic modeling system. A goal of HRRRE development is a system with sufficient spread among members, comparable in magnitude to the random error in the ensemble mean, to represent the range of possible future atmospheric states. HRRRE member diversity has traditionally been obtained by perturbing the initial and lateral boundary conditions of each member, but recent development has focused on implementing stochastic approaches in HRRRE to generate additional spread. These techniques were tested in retrospective experiments and in the May 2019 Hazardous Weather Testbed Spring Experiment (HWT-SE). Results show a 6%–25% increase in the ensemble spread in 2-m temperature, 2-m mixing ratio, and 10-m wind speed when stochastic parameter perturbations are used in HRRRE (HRRRE-SPP). Case studies from HWT-SE demonstrate that HRRRE-SPP performed similar to or better than the operational High-Resolution Ensemble Forecast system, version 2 (HREFv2), and the nonstochastic HRRRE. However, subjective evaluations provided by HWT-SE forecasters indicated that overall, HRRRE-SPP predicted lower probabilities of severe weather (using updraft helicity as a proxy) compared to HREFv2. A statistical analysis of the performance of HRRRE-SPP and HREFv2 from the 2019 summer convective season supports this claim, but also demonstrates that the two systems have similar reliability for prediction of severe weather using updraft helicity.

FogNet: A multiscale 3D CNN with double-branch dense block and attention mechanism for fog prediction

The reduction of visibility adversely affects land, marine, and air transportation. Thus, the ability to skillfully predict fog would provide utility. We predict fog visibility categories below 1600 m, 3200 m and 6400 m by post-processing numerical weather prediction model output and satellite-based sea surface temperature (SST) using a 3D-Convolutional Neural Network (3D-CNN). The target is an airport located on a barrier island adjacent to a major US port; measured visibility from this airport serves as a proxy for fog that develops over the port. The features chosen to calibrate and test the model originate from the North American Mesoscale Forecast System, with values of each feature organized on a 32 × 32 horizontal grid; the SSTs were obtained from the NASA Multiscale Ultra Resolution dataset. The input to the model is organized as a high dimensional cube containing 288 to 384 layers of 2D horizontal fields of meteorological variables (predictor maps). In this 3D-CNN (hereafter, FogNet), two parallel branches of feature extraction have been designed, one for spatially auto-correlated features (spatial-wise dense block and attention module), and the other for correlation between input variables (variable-wise dense block and attention mechanism.) To extract features representing processes occurring at different scales, a 3D multiscale dilated convolution is used. Data from 2009 to 2017 (2018 to 2020) are used to calibrate (test) the model. FogNet performance results for 6, 12− and 24−h lead times are compared to results from the High-Resolution Ensemble Forecast (HREF) system. FogNet outperformed HREF using 8 standard evaluation metrics.

Area-covering postprocessing of ensemble precipitation forecasts using topographical and seasonal conditions

Probabilistic weather forecasts from ensemble systems require statistical postprocessing to yield calibrated and sharp predictive distributions. This paper presents an area-covering postprocessing method for ensemble precipitation predictions. We rely on the ensemble model output statistics (EMOS) approach, which generates probabilistic forecasts with a parametric distribution whose parameters depend on (statistics of) the ensemble prediction. A case study with daily precipitation predictions across Switzerland highlights that postprocessing at observation locations indeed improves high-resolution ensemble forecasts, with 4.5% CRPS reduction on average in the case of a lead time of 1 day. Our main aim is to achieve such an improvement without binding the model to stations, by leveraging topographical covariates. Specifically, regression coefficients are estimated by weighting the training data in relation to the topographical similarity between their station of origin and the prediction location. In our case study, this approach is found to reproduce the performance of the local model without using local historical data for calibration. We further identify that one key difficulty is that postprocessing often degrades the performance of the ensemble forecast during summer and early autumn. To mitigate, we additionally estimate on the training set whether postprocessing at a specific location is expected to improve the prediction. If not, the direct model output is used. This extension reduces the CRPS of the topographical model by up to another 1.7 % on average at the price of a slight degradation in calibration. In this case, the highest improvement is achieved for a lead time of 4 days.

Effects of Horizontal Grid Spacing and Inflow Environment on Forecasts of Cyclic Mesocyclogenesis in NSSL’s Warn-on-Forecast System (WoFS)

AbstractCyclic mesocyclogenesis is the process by which a supercell produces multiple mesocyclones with similar life cycles. The frequency of cyclic mesocyclogenesis has been linked to tornado potential, with higher frequencies decreasing the potential for tornadogenesis. Thus, the ability to predict the presence and frequency of cycling in supercells may be beneficial to forecasters for assessing tornado potential. However, idealized simulations of cyclic mesocyclogenesis have found it to be highly sensitive to environmental and computational parameters. Thus, whether convective-allowing models can resolve and predict cycling has yet to be determined. This study tests the capability of a storm-scale, ensemble prediction system to resolve the cycling process and predict its frequency. Forecasts for three cyclic supercells occurring in May 2017 are generated by NSSL’s Warn-on-Forecast System (WoFS) using 3- and 1-km grid spacing. Rare cases of cyclic-like processes were identified at 3 km, but cycling occurred more frequently at 1 km. WoFS predicted variation in cycling frequencies for the storms that were similar to observed variations in frequency. Object-based identification of mesocyclones was used to extract environmental parameters from a storm-relative inflow sector from each mesocyclone. Lower magnitudes of 0–1-km storm-relative helicity and significant tornado parameter are present for the two more frequently cycling supercells, and higher values are present for the case with the fewest cycles. These results provide initial evidence that high-resolution ensemble forecasts can potentially provide useful guidance on the likelihood and cycling frequency of cyclic supercells.

What Does a Convection-Allowing Ensemble of Opportunity Buy Us in Forecasting Thunderstorms?

AbstractThe High Resolution Ensemble Forecast v2.1 (HREFv2.1), an operational convection-allowing model (CAM) ensemble, is an “ensemble of opportunity” wherein forecasts from several independently designed deterministic CAMs are aggregated and postprocessed together. Multiple dimensions of diversity in the HREFv2.1 ensemble membership contribute to ensemble spread, including model core, physics parameterization schemes, initial conditions (ICs), and time lagging. In this study, HREFv2.1 forecasts are compared against the High Resolution Rapid Refresh Ensemble (HRRRE) and the Multiscale data Assimilation and Predictability (MAP) ensemble, two experimental CAM ensembles that ran during the 5-week Spring Forecasting Experiment (SFE) in spring 2018. The HRRRE and MAP are formally designed ensembles with spread achieved primarily through perturbed ICs. Verification in this study focuses on composite radar reflectivity and updraft helicity to assess ensemble performance in forecasting convective storms. The HREFv2.1 shows the highest overall skill for these forecasts, matching subjective real-time impressions from SFE participants. Analysis of the skill and variance of ensemble member forecasts suggests that the HREFv2.1 exhibits greater spread and more effectively samples model uncertainty than the HRRRE or MAP. These results imply that to optimize skill in forecasting convective storms at 1–2-day lead times, future CAM ensembles should employ either diverse membership designs or sophisticated perturbation schemes capable of representing model uncertainty with comparable efficacy.

Evaluation of an Application of Probabilistic Quantitative Precipitation Forecasts for Flood Forecasting

Probabilistic streamflow forecasts using precipitation derived from ensemble-based Probabilistic Quantitative Precipitation Forecasts (PQPFs) are examined. The PQPFs provide rainfall amounts associated with probabilities of exceedance for all grid points, which are averaged to the watershed scale for input to the operational Sacramento Soil Moisture Accounting hydrologic model to generate probabilistic streamflow predictions. The technique was tested using both the High-Resolution Rapid Refresh Ensemble (HRRRE) and the High-Resolution Ensemble Forecast version 2.0 (HREF) for 11 river basins across the upper Midwest for 109 cases. The resulting discharges associated with low probability of exceedance values were too large; no events were observed having discharges above the 10% exceedance value predicted from the technique applied to both ensembles, and no events were observed having discharges above the 25% exceedance value from the HREF-based forecast. The large differences are due to using the same precipitation exceedance value at all points; it is unlikely that all watershed points would experience the heavy rainfall associated with the 5% probability of exceedance. The technique likely can be improved through calibration of the basin-average precipitation forecasts based on typical distributions of precipitation within the convective systems that dominate warm-season precipitation events or calibration of the resulting probabilistic discharge forecasts.

Calibration of Machine Learning–Based Probabilistic Hail Predictions for Operational Forecasting

Abstract In this study, we use machine learning (ML) to improve hail prediction by postprocessing numerical weather prediction (NWP) data from the new High-Resolution Ensemble Forecast system, version 2 (HREFv2). Multiple operational models and ensembles currently predict hail, however ML models are more computationally efficient and do not require the physical assumptions associated with explicit predictions. Calibrating the ML-based predictions toward familiar forecaster output allows for a combination of higher skill associated with ML models and increased forecaster trust in the output. The observational dataset used to train and verify the random forest model is the Maximum Estimated Size of Hail (MESH), a Multi-Radar Multi-Sensor (MRMS) product. To build trust in the predictions, the ML-based hail predictions are calibrated using isotonic regression. The target datasets for isotonic regression include the local storm reports and Storm Prediction Center (SPC) practically perfect data. Verification of the ML predictions indicates that the probability magnitudes output from the calibrated models closely resemble the day-1 SPC outlook and practically perfect data. The ML model calibrated toward the local storm reports exhibited better or similar skill to the uncalibrated predictions, while decreasing model bias. Increases in reliability and skill after calibration may increase forecaster trust in the automated hail predictions.