(Severe) Thunderstorms and Climate HAROLD BROOKS NOAA/NSSL HAROLD.BROOKS@NOAA.GOV
Big questions How and why are weather hazards distributed? Are things changing in time and will they? Begin with thunderstorm rain, then severe thunderstorms (tornado, winds>=50 kts, hail>=1 inch
What can we say about causes? Easy-things associated with increased CO 2, increased average temperature Hard-getting away from those Formal attribution studies give idea of probability Need physical relationship
National Heavy Precipitation Changes (www.ncdc.noaa.gov/extremes/cei)
Precip(1 in/hr) Hitchens et al. 2014
August precip for MSP Daily data back to 1871 (National Climatic Data Center) Heavy rain is almost certainly from thunderstorms Look at running 20 year averages
Reports-A logical place to start US reporting database Target of opportunity Changes in de jure and de facto standards Hail in other countries China-yes/no reports available at >500 sites with some size data Italy, France, and Spain-hailpad networks
Views on Tornado Data Quality Want consistency in absence of complete accuracy Tier 1-Date/time, location, deaths, (E)F1+ Tier 2-Length, (E)F scale for different periods Tier 3-Width, path, injuries Best thing-(e)f1+ for 1954-2013
Annual US Tornado Reports by F-scale 1400 1200 F0 F1+ 1000 Reports 800 600 400 200 0 1950 1960 1970 1980 1990 2000 2010 Year
US Reports Per Year (Divided by 2) 10000 Wind Hail Reports 1000 100 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 Year
Comparing Environment Estimates to Reports ~0.8%/year ~6.0%/year Environments account for ~7% of increase in reports
Report summary Lots of reporting changes make it hard to know Big interannual variability Hail is observed more consistently some places 17
China-Hail Frequency Xie et al. 2008 (GRL)
France/Italy Hailpad Data Occurrence Kinetic Energy Berthet et al. (ECSS 2009) Eccel et al. (2011)
Hail Obs Summary Little change to slight decrease in occurrence Small decrease in mean size, but increase in kinetic energy of hailfalls Start with slightly larger hail at beginning of fall Melt more because of higher freezing level height, particularly impacting small Leaves distribution shifted to larger stones Does it extend to larger sizes? 20
Severe thunderstorms Where and when in the US The observational record
Timing of Maximum Tornado Probability Brooks et al. 2003
Annual cycles of tornadoes Take smoothed fields in space and time Different time periods Focus on (E)F1+
What might have changed with tornadoes? Impacts of seasonal temperature swings? Timing of season "As spring moves up a week or two, tornado season will start in February instead of waiting for April When does tornado season start?
What has changed about tornado distributions? Appearance of increased variability Starting date Since 2002, set or tied records for monthly F1+ extremes Max-4 (Feb 08, Apr 11, May 03, Sept 04) Min-6 (Jan 03, Feb 10, May 05, Jun 02, Jul 12, Sept 09) Days per year (F1) decreased More tornadoes on biggest days
Timing of tornadoes in Plains (TX/OK/KS/NE) Long and Stoy (2014)
Temperature Impacts on Tornadoes Use warm and cold historical periods as proxy for change Implicit assumption-future patterns look like recent warm Start by looking at annual temps (NCDC US 48 states) (E)F1+ counts
ANNUAL TORS Cyan-Old, Magenta-New
Monthly temperatures Shorter time scale Look at warmest, coldest for each calendar month
16% Reduction/SD(Temp) Cold=515/ year Warm=442 Neutral=538
Spatial patterns All (E)F1+ tornado starting points for the 20X12 months Gaussian smoother (120 km)
ANNUAL DAYS
Summary of tornado observations Increased variability in recent years Possible temperature impacts (more in warm winters, fewer in warm summers) Change in location??
Ingredients for severe thunderstorms-the supercell Thunderstorms Low-level warm, moist air Mid-level (~2-10 km) relatively cold, dry air Something to lift the warm, moist air Combine first two to get energy available for storm (CAPE or Wmax) Organization Winds that increase and change direction with height over lowest few km From equator at surface, west aloft
Reanalysis Proximity Soundings (1997-9) 100 Sfc-6 km Wind Difference (m/s) Shear 10 1 Little severe Significant severe Significant tornado 'Best' discriminator 0.1 0.1 1 10 100 1000 10000 CAPE (J/kg) Energy
(Dan Cecil, Univ. of Alabama-Huntsville) Updated from Brooks et al (2003)
What will happen in the future Mean expected changes CAPE goes up (related to moisture increase) Shear goes down (decrease in equator-to-pole gradient) We care about combinations Climate model simulations Dynamical downscaling
Trapp et al. (2009) Regional Analyses
Updraft Shear Combination
Diffenbaugh et al. (2013)
Black dots: Ensemble S/N > 1 White dots: Ensemble S/N > 2 Diffenbaugh et al. (2013)
U.S. Population Density (persons km -2 )
Tornado Fatalities (Ashley 2007) Tornado Fatalities Killer Tornadoes Mobile Home Fatalities Permanent Home Fatalities
Regional Differences %Nocturnal Fatalities %Mobile Homes Forest Cover % in Poverty
Model summary Energy term increases Shear term decreases Overall, more environments favorable for severe storms Tendency to increase non-tornadic wind events Weaker evidence for long-term increase in tornadoes How do we look at long time series? New tool-20 th Century Reanalysis-surface pressure, monthly SST
20 th Century Reanalysis Large-scale cyclones underestimated?
Closing thoughts We need improved modelling Need to improve environment-event relationships Higher resolution, better reanalyses Increased use of high-res models Increased variability Tornadoes on fewer days, more on outbreak days More heavy rain Some suggestion from models of future variability nationally