Random Forest Model to Assess Predictor Importance and Nowcast Severe Storms using High-Resolution Radar–GOES Satellite–Lightning Observations

2021 ◽  
Author(s):  
John R. Mecikalski ◽  
Thea N. Sandmæl ◽  
Elisa M. Murillo ◽  
Cameron R. Homeyer ◽  
Kristopher M. Bedka ◽  
...  
Keyword(s):  
Great Plains ◽  
Random Forests ◽  
Motion Vector ◽  
Radar Data ◽  
Lightning Flash ◽  
Severe Storms ◽  
Tornado Warnings ◽  
Predictor Importance ◽  
Scan Data ◽  
Total Lightning

AbstractFew studies have assessed combined satellite, lightning, and radar databases to diagnose severe storm potential. The research goal here is to evaluate next-generation, 60-second update frequency geostationary satellite and lightning information with ground-based radar to isolate which variables, when used in concert, provide skillful discriminatory information for identifying severe (hail ≥2.5 cm in diameter, winds ≥25 m s–1, tornadoes) versus non-severe storms. The focus of this study is predicting severe thunderstorm and tornado warnings. A total of 2,004 storms in 2014–2015 were objectively tracked with 49 potential predictor fields related to May, daytime Great Plains convective storms. All storms occurred when 1-min Geostationary Operational Environmental Satellite (GOES)–14 “super rapid scan” data were available. The study used three importance methods to assess predictor importance related to severe warnings, and random forests to provide a model and skill evaluation measuring the ability to predict severe storms. Three predictor importance methods show that GOES mesoscale atmospheric motion vector derived cloud-top divergence and above anvil cirrus plume presence provide the most satellite-based discriminatory power for diagnosing severe warnings. Other important fields include Earth Networks Total Lightning flash density, GOES estimated cloud-top vorticity, and overshooting-top presence. Severe warning predictions are significantly improved at the 95% confidence level when a few important satellite and lightning fields are combined with radar fields, versus when only radar data are used in the random forests model. This study provides a basis for including satellite and lightning fields within machine-learning models to help forecast severe weather.

scholarly journals Relationships between Deep Convection Updraft Characteristics and Satellite-Based Super Rapid Scan Mesoscale Atmospheric Motion Vector–Derived Flow

2018 ◽  
Vol 146 (10) ◽  
pp. 3461-3480 ◽  
Author(s):  
Jason M. Apke ◽  
John R. Mecikalski ◽  
Kristopher Bedka ◽  
Eugene W. McCaul ◽  
Cameron R. Homeyer ◽  
...  
Keyword(s):  
Doppler Radar ◽  
Deep Convection ◽  
Motion Vector ◽  
Three Dimensional ◽  
Radar Data ◽  
Severe Weather ◽  
Lightning Flash ◽  
Rapid Scan ◽  
Total Lightning ◽  
The Relationship

Abstract Rapid acceleration of cloud-top outflow near vigorous storm updrafts can be readily observed in Geostationary Operational Environmental Satellite-14 (GOES-14) super rapid scan (SRS; 60 s) mode data. Conventional wisdom implies that this outflow is related to the intensity of updrafts and the formation of severe weather. However, from an SRS satellite perspective, the pairing of observed expansion and updraft intensity has not been objectively derived and documented. The goal of this study is to relate GOES-14 SRS-derived cloud-top horizontal divergence (CTD) over deep convection to internal updraft characteristics, and document evolution for severe and nonsevere thunderstorms. A new SRS flow derivation system is presented here to estimate storm-scale (<20 km) CTD. This CTD field is coupled with other proxies for storm updraft location and intensity such as overshooting tops (OTs), total lightning flash rates, and three-dimensional flow fields derived from dual-Doppler radar data. Objectively identified OTs with (without) matching CTD maxima were more (less) likely to be associated with radar-observed deep convection and severe weather reports at the ground, suggesting that some OTs were incorrectly identified. The correlation between CTD magnitude, maximum updraft speed, and total lightning was strongly positive for a nonsupercell pulse storm, and weakly positive for a supercell with multiple updraft pulses present. The relationship for the supercell was nonlinear, though larger flash rates are found during periods of larger CTD. Analysis here suggests that combining CTD with OTs and total lightning could have severe weather nowcasting value.

Exploring Impacts of Rapid-Scan Radar Data on NWS Warning Decisions

2012 ◽  
Vol 27 (4) ◽  
pp. 1031-1044 ◽  
Author(s):  
Pamela L. Heinselman ◽  
Daphne S. LaDue ◽  
Heather Lazrus
Keyword(s):  
Decision Making ◽  
Process Analysis ◽  
Radar Data ◽  
Lead Times ◽  
Weather Radars ◽  
Rapid Scan ◽  
Tornado Warnings ◽  
Scan Time ◽  
Preliminary Study ◽  
Scan Data

Abstract Rapid-scan weather radars, such as the S-band phased array radar at the National Weather Radar Testbed in Norman, Oklahoma, improve precision in the depiction of severe storm processes. To explore potential impacts of such data on forecaster warning decision making, 12 National Weather Service forecasters participated in a preliminary study with two control conditions: 1) when radar scan time was similar to volume coverage pattern 12 (4.5 min) and 2) when radar scan time was faster (43 s). Under these control conditions, forecasters were paired and worked a tropical tornadic supercell case. Their decision processes were observed and audio was recorded, interactions with data displays were video recorded, and the products were archived. A debriefing was conducted with each of the six teams independently and jointly, to ascertain the forecaster decision-making process. Analysis of these data revealed that teams examining the same data sometimes came to different conclusions about whether and when to warn. Six factors contributing toward these differences were identified: 1) experience, 2) conceptual models, 3) confidence, 4) tolerance of possibly missing a tornado occurrence, 5) perceived threats, and 6) software issues. The three 43-s teams issued six warnings: three verified, two did not verify, and one event was missed. Warning lead times were the following: tornado, 18.6 and 11.5 min, and severe, 6 min. The three tornado warnings issued by the three 4.5-min teams verified, though warning lead times were shorter: 4.6 and 0 min (two teams). In this case, use of rapid-scan data showed the potential to extend warning lead time and improve forecasters’ confidence, compared to standard operations.

scholarly journals Evaluation of Different Storm Parameters as the Proxies for Gridded Total Lightning Flash Rates: A Convection-Allowing Model Study

Atmosphere ◽  
2021 ◽  
Vol 12 (1) ◽  
pp. 95
Author(s):  
Xinyao Qian ◽  
Haoliang Wang
Keyword(s):  
Data Assimilation ◽  
Mesoscale Convective System ◽  
Radar Data ◽  
Skill Score ◽  
Convective System ◽  
Test Case ◽  
Lightning Flash ◽  
Simulation Accuracy ◽  
Mesoscale Convective ◽  
Total Lightning

Lightning simulation is important for a variety of applications, including lightning forecast, atmospheric chemical simulation, and lightning data assimilation. In this study, the potential of five storm parameters (graupel volume, precipitation ice mass, radar echo volume, maximum updraft, and updraft volume) to be used as the proxy for the diagnosis of gridded total lightning flash rates has been investigated in a convection-allowing model. A mesoscale convective system occurred in the Guangdong province of China was selected as the test case. Radar data assimilation was used to improve the simulation accuracy of the convective clouds, hence providing strong instantaneous correlations between observed and simulated storm signatures. The areal coverage and magnitude of the simulated lightning flash rates were evaluated by comparing to those of the total lightning observations. Subjective and the Fractions Skill Score (FSS) evaluations suggest that all the five proxies tested in this study are useful to indicate general tendencies for the occurrence, region, and time of lightning at convection-allowing scale (FSS statistics for the threshold of 1 flash per 9 km2 per hour were around 0.7 for each scheme). The FSS values were decreasing as the lightning flash rate thresholds used for FSS computation increased for all the lightning diagnostic schemes with different proxies. For thresholds from 1 to 3 and 16 to 20 flashes per 9 km2 per hour, the graupel contents related schemes achieved higher FSS values compared to the other three schemes. For thresholds from 5 to 15 flashes per 9 km2 per hour, the updraft volume related scheme yielded the largest FSS. When the thresholds of lightning flash rates were greater than 13 flashes per 9 km2 per hour, the FSS values were below 0.5 for all the lightning diagnostic schemes with different proxies.

Lightning Data Assimilation Scheme in a 4DVAR System and its Impact on Very-Short-Term Convective Forecasting

2020 ◽  
Author(s):  
Xian Xiao ◽  
Juanzhen Sun ◽  
Xiushu Qie ◽  
Zhuming Ying ◽  
Lei Ji ◽  
...  
Keyword(s):  
Vertical Velocity ◽  
Doppler Radar ◽  
Analysis Data ◽  
Radar Data ◽  
Lightning Flash ◽  
Short Term ◽  
Distance Weighted ◽  
Dynamical Consistency ◽  
Analysis System ◽  
Total Lightning

AbstractA proof-of-concept method for the assimilation of total lightning observations in the 4DVAR framework is proposed and implemented into the Variational Doppler Radar Analysis System (VDRAS). Its performance is evaluated for the very-short-term precipitation forecasts of a localized convective event over northeastern China. The lightning DA scheme assimilated pseudo observations for vertical velocity fields derived from observed total lightning rates and statistically computed vertical velocity profile from VDRAS analysis data. To reduce representative errors of the derived vertical velocity, a distance-weighted horizontal interpolation is applied to the input data prior to the DA. The case study reveals that although 0–2 hour precipitation nowcasts are improved by assimilating lightning data alone compared to CTRL (no radar or lightning) and RAD (radar only), better results are obtained when the lightning data are assimilated with radar data simultaneously. The assimilation of both data sources results in improved dynamical consistency with enhanced updraft and latent heat as well as improved moisture distributions. Additional experiments are conducted to evaluate the sensitivity of the combined DA scheme to varied vertical velocity profiles, radii of horizontal interpolation, binning time intervals, and relationships used to estimate the maximum vertical velocity from lightning flash rates. It is shown that the scheme is robust to these variations with both radar and lightning assimilated data.

scholarly journals Colorado Plowable Hailstorms: Synoptic Weather, Radar, and Lightning Characteristics

2016 ◽  
Vol 31 (2) ◽  
pp. 663-693 ◽  
Author(s):  
Evan A. Kalina ◽  
Katja Friedrich ◽  
Brian C. Motta ◽  
Wiebke Deierling ◽  
Geoffrey T. Stano ◽  
...  
Keyword(s):  
Precipitable Water ◽  
Radar Data ◽  
Colorado Front Range ◽  
Lightning Flash ◽  
Front Range ◽  
Freezing Level ◽  
Synoptic Weather ◽  
Westerly Flow ◽  
The Times ◽  
Total Lightning

Abstract Synoptic weather, S-band dual-polarization radar, and total lightning observations are analyzed from four thunderstorms that produced “plowable” hail accumulations of 15–60 cm in localized areas of the Colorado Front Range. Results indicate that moist, relatively slow (5–15 m s−1) southwesterly-to-westerly flow at 500 hPa and postfrontal low-level upslope flow, with 2-m dewpoint temperatures of 11°–19°C at 1200 LST, were present on each plowable hail day. This pattern resulted in column-integrated precipitable water values that were 132%–184% of the monthly means and freezing-level heights that were 100–700 m higher than average. Radar data indicate that between one and three maxima in reflectivity Z (68–75 dBZ) and 50-dBZ echo-top height (11–15 km MSL) occurred over the lifetime of each hailstorm. These maxima, which imply an enhancement in updraft strength, resulted in increased graupel and hail production and accumulating hail at the surface within 30 min of the highest echo tops. The hail core had Z ~ 70 dBZ, differential reflectivity ZDR from 0 to −4 dB, and correlation coefficient ρHV of 0.80–0.95. Time–height plots reveal that these minima in ZDR and ρHV gradually descended to the surface after originating at heights of 6–10 km MSL ~15–60 min prior to accumulating hailfall. Hail accumulations estimated from the radar data pinpoint the times and locations of plowable hail, with depths greater than 5 cm collocated with the plowable hail reports. Three of the four hail events were accompanied by lightning flash rates near the maximum observed thus far within the thunderstorm.

scholarly journals Numerical Simulations of Lightning and Storm Charge of the 29–30 May 2004 Geary, Oklahoma, Supercell Thunderstorm Using EnKF Mobile Radar Data Assimilation

2014 ◽  
Vol 142 (11) ◽  
pp. 3977-3997 ◽  
Author(s):  
Kristin M. Calhoun ◽  
Edward R. Mansell ◽  
Donald R. MacGorman ◽  
David C. Dowell
Keyword(s):  
Data Assimilation ◽  
Radar Data ◽  
Main Body ◽  
Lightning Flash ◽  
Wind Fields ◽  
Propagation Length ◽  
Kinematic Evolution ◽  
Supercell Storm ◽  
Total Lightning ◽  
Radar Data Assimilation

Abstract Results from simulations are compared with dual-Doppler and total lightning observations of the 29–30 May 2004 high-precipitation supercell storm from the Thunderstorm Electrification and Lightning Experiment (TELEX). The simulations use two-moment microphysics with six hydrometeor categories and parameterizations for electrification and lightning while employing an ensemble Kalman filter for mobile radar data assimilation. Data assimilation was utilized specifically to produce a storm similar to the observed for ancillary analysis of the electrification and lightning associated with the supercell storm. The simulated reflectivity and wind fields well approximated that of the observed storm. Additionally, the simulated lightning flash rates were very large, as was observed. The simulation reveals details of the charge distribution and dependence of lightning on storm kinematics, characteristics that could not be observed directly. Storm electrification was predominately confined to the updraft core, but the persistence of both positive and negative charging of graupel in this region, combined with the kinematic evolution, limited the extent of charged areas of the same polarity. Thus, the propagation length of lightning flashes in this region was also limited. Away from the updraft core, regions of charge had greater areal extent, allowing flashes to travel farther without termination due to unfavorable charge potential. Finally, while the simulation produced the observed lightning holes and high-altitude lightning seen in the observations, it failed to produce the observed lightning initiations (or even lightning channels) in the distant downstream anvil as seen in the observed storm. Instead, the simulated lightning was confined to the main body of the storm.

scholarly journals Evaluation of Fengyun-4A Lightning Mapping Imager (LMI) Performance during Multiple Convective Episodes over Beijing

2021 ◽  
Vol 13 (9) ◽  
pp. 1746
Author(s):  
Zhixiong Chen ◽  
Xiushu Qie ◽  
Juanzhen Sun ◽  
Xian Xiao ◽  
Yuxin Zhang ◽  
...  
Keyword(s):  
Doppler Radar ◽  
Detection Efficiency ◽  
Ground Truth ◽  
Radar Data ◽  
Detection Capability ◽  
Cloud Properties ◽  
Order Of Magnitude ◽  
Total Lightning ◽  
Matching Criteria ◽  
Event Based

This study investigates the characteristics of space-borne Lightning Mapping Imager (LMI) lightning products and their relationships with cloud properties using ground-based total lightning observations from the Beijing Broadband Lightning Network (BLNET) and cloud information from S-band Doppler radar data. LMI showed generally consistent lightning spatial distributions with those of BLNET, and yielded a considerable lightning detection capability over regions with complex terrain. The ratios between the LMI events, groups and flashes were approximately 9:3:1, and the number of LMI-detected flashes was roughly one order of magnitude smaller than the number of BLNET-detected flashes. However, in different convective episodes, the LMI detection capability was likely to be affected by cloud properties, especially in strongly electrified convective episodes associated with frequent lightning discharging and thick cloud depth. As a result, LMI tended to detect lightning flashes located in weaker and shallower cloud portions associated with fewer cloud shielding effects. With reference to the BLNET total lightning data as the ground truth of observation (both intra-cloud lightning and cloud-to-ground lightning flashes), the LMI event-based detection efficiency (DE) was estimated to reach 28% under rational spatiotemporal matching criteria (1.5 s and 65 km) over Beijing. In terms of LMI flash-based DE, it was much reduced compared with event-based DE. The LMI flash-based ranged between 1.5% and 3.5% with 1.5 s and 35–65 km matching scales. For 330 ms and 35 km, the spatiotemporal matching criteria used to evaluate Geostationary Lightning Mapper (GLM), the LMI flash-based DE was smaller (<1%).

scholarly journals A Total Lightning Trending Algorithm to Identify Severe Thunderstorms

2010 ◽  
Vol 27 (1) ◽  
pp. 3-22 ◽  
Author(s):  
Patrick N. Gatlin ◽  
Steven J. Goodman
Keyword(s):  
Severe Weather ◽  
Lightning Flash ◽  
Severe Thunderstorm ◽  
Critical Success ◽  
Severe Thunderstorms ◽  
Weather Events ◽  
Total Lightning ◽  
Jump Algorithm ◽  
Early Indication ◽  
Critical Success Index

Abstract An algorithm that provides an early indication of impending severe weather from observed trends in thunderstorm total lightning flash rates has been developed. The algorithm framework has been tested on 20 thunderstorms, including 1 nonsevere storm, which occurred over the course of six separate days during the spring months of 2002 and 2003. The identified surges in lightning rate (or jumps) are compared against 110 documented severe weather events produced by these thunderstorms as they moved across portions of northern Alabama and southern Tennessee. Lightning jumps precede 90% of these severe weather events, with as much as a 27-min advance notification of impending severe weather on the ground. However, 37% of lightning jumps are not followed by severe weather reports. Various configurations of the algorithm are tested, and the highest critical success index attained is 0.49. Results suggest that this lightning jump algorithm may be a useful operational diagnostic tool for severe thunderstorm potential.

A Variational Method for Filling in Missing Data in Doppler Velocity Fields

2021 ◽  
Author(s):  
VINCENT T. WOOD ◽  
ROBERT P. DAVIES-JONES ◽  
ALAN SHAPIRO
Keyword(s):  
Missing Data ◽  
Doppler Radar ◽  
Signal To Noise Ratio ◽  
Original Data ◽  
Radar Data ◽  
Velocity Fields ◽  
Doppler Velocity ◽  
Doppler Radar Data ◽  
Ground Clutter ◽  
Tornado Warnings

AbstractSingle-Doppler radar data are often missing in important regions of a severe storm due to low return power, low signal-to-noise ratio, ground clutter associated with normal and anomalous propagation, and missing radials associated with partial or total beam blockage. Missing data impact the ability of WSR-88D algorithms to detect severe weather. To aid the algorithms, we develop a variational technique that fills in Doppler velocity data voids smoothly by minimizing Doppler velocity gradients while not modifying good data. This method provides estimates of the analysed variable in data voids without creating extrema.Actual single-Doppler radar data of four tornadoes are used to demonstrate the variational algorithm. In two cases, data are missing in the original data, and in the other two, data are voided artificially. The filled-in data match the voided data well in smoothly varying Doppler velocity fields. Near singularities such as tornadic vortex signatures, the match is poor as anticipated. The algorithm does not create any velocity peaks in the former data voids, thus preventing false triggering of tornado warnings. Doppler circulation is used herein as a far-field tornado detection and advance-warning parameter. In almost all cases, the measured circulation is quite insensitive to the data that have been voided and then filled. The tornado threat is still apparent.

Sign in / Sign up

Export Citation Format

Share Document