A multiple source approach to operational lightning prediction

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Wednesday, 26 January 2011: 4:00 PM
A multiple source approach to operational lightning prediction
602/603 (Washington State Convention Center)
Gail Hartfield, NOAA/NWS, Raleigh, NC
Manuscript (562.6 kB)

Lightning poses a significant threat to life and property. Thirty-four people were killed by lightning in the United States in 2009 (http://www.weather.gov/os/hazstats.shtml) and many more were injured. Yearly damages and losses due to lightning strike damage in the United States are estimated at 5 to 6 billion dollars (National Lightning Safety Institute, 2008). While the National Weather Service has greatly improved its forecasts and warnings for severe storms, tornadoes, and floods, the ability to accurately predict lightning activity on multiple time scales – from a few minutes to one day – remains elusive. Understanding of the lightning production process is limited and largely theoretical, as it occurs on the poorly-sampled storm scale with observational data primarily limited to cloud-to-ground (CG) strikes over much of the country.

Since 2007 the National Weather Service in Raleigh has investigated excessive lightning events in the Carolinas, including the creation of a lightning climatology for central North Carolina and evaluations of pre-storm tropospheric conditions favoring high lightning activity. A review of past lightning studies and publications, as well as local studies of significant lightning events, revealed key environmental parameters which suggest an increased likelihood of convection with excessive lightning, such as high mixed-layer convective available potential energy (CAPE), high CAPE in the prime electrification zone (from -10 degrees C to -30 degrees C), and increasing precipitable water. Based on these and other parameters from various models and sources, including experimental lightning probabilities from the NWS Storm Prediction Center (SPC), NWS Raleigh forecasters produced experimental daily “lightning outlooks” from May through September in 2008 and 2009.

A mention of the elevated risk of excessive lightning was included in the morning Hazardous Weather Outlook on 24 of the 153 forecast days in 2009. Of these, eight outlooks specifically noted the extreme nature of the expected lightning, and these included three of the top four excessive lightning days (as defined by the total number of CG strikes in 24 hours) in 2009 in central North Carolina. Media response has been positive, with television meteorologists in multiple North Carolina broadcast markets mentioning the NWS-indicated lightning risk in their weathercasts. Initial subjective forecast verification suggests that the human-determined lightning forecasts improve upon any single-model automated method, however further objective verification is needed. Challenges also persist regarding the optimal methods for conveying lightning forecast information to users such as utility companies and the general public.

Beginning in the summer of 2010, NWS Raleigh forecasters began subjective evaluations of an experimental lightning threat parameter produced by the convection-allowing 4-km resolution Weather Research and Forecast (WRF) model run by the National Severe Storms Laboratory (NSSL). This parameter, devised by researchers at NSSL and the Short Term Prediction Research and Transition Center (SPoRT), combines the upward graupel flux at -15 degrees C (which factors in the updraft strength and presence of graupel in the critical lightning production region) and the column integrated graupel. Incorporation of this output field from an explicit-convection model, into the existing process of evaluating pre-storm environmental conditions from traditional parameterized models, results in a pseudo-“ensemble” approach to lightning prediction. With forecasters assessing output from a variety of forecast systems, including models with both parameterized and explicit convection, to determine the overall threat of excessive lightning, there is less reliance on a single model or parameter which may or may not be successful on a particular day. Evaluations of this forecast process for convective events in 2009 and 2010 will be presented.