1 / 17

The Ingredients Based Tornado Parameter

The Ingredients Based Tornado Parameter. Matt Onderlinde. Motivation. At least 550 fatalities due to tornadoes in the US in 2011 http://www.noaanews.noaa.gov/2011_tornado_information.html SPC maintains an extensive tornado database which can be compared to NCEP model analyses

grant
Download Presentation

The Ingredients Based Tornado Parameter

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. The Ingredients Based Tornado Parameter Matt Onderlinde

  2. Motivation • At least 550 fatalities due to tornadoes in the US in 2011http://www.noaanews.noaa.gov/2011_tornado_information.html • SPC maintains an extensive tornado database which can be compared to NCEP model analyses • I attempt to derive statistical relationships between tornadoes and model variables

  3. Hypothesis • Existing metrics for tornado parameters exist. Would the consideration of additional environmental factors improve these parameters? • Can we use the SPC tornado reports database and model analyses to derive a parameter directly?

  4. Study Domain All tornado reports and RUC analyses data are regridded to the grid shown here. The IBTP is calculated on this grid along with the SPC’s significant tornado parameter (STP) for comparison.

  5. Variables to consider • CAPE • CIN • 0-6 km vector wind shear • 0-1 km vector wind shear • 2 m temperature • 2m dew point • LCL height • 0-1 km SRH • 0-3 km SRH • 1000-850 mb mean RH • 850-600 mb lapse rate • 850-700 mb lapse rate • Surface moisture convergence Variables considered by STP • CAPE • CIN • Effective Bulk Shear • Effective SRH • LCL Height STP = (mlCAPE/1500 J kg-1) * ((2000-mlLCL)/1000 m) * (ESRH/150 m2 s-2) * (EBWD/20 m s-1) * ((200+mlCIN)/150 J kg-1) Plus if-statements….

  6. Methodology • Calculate each model variable from the previous slide on our grid for the time of all tornadoes in the Plains (2005 – 2009)- Note, 2010 will be used for independent verification • RUC analyses are used • Correlate all variables with each other and check for inter-correlations • Correlate all variables with tornado occurrence

  7. Methodology • Normalize each variable according to how frequently specific ranges of that variable occur (example in a minute) • Make histograms of our normalized variables and then fit functions to the height of the bins • Use these functions to scale each variable between 0 and 1 so we can use each variable as a predictor

  8. Methodology • Multiply each scaled variable (now ranging between 0 and 1) together to get a “tornado potential value” between 0 and 1 • This allows us to effectively deal with “necessary but insufficient” atmospheric variables

  9. Scaled variable histogram 2 m dew point (F) Normalization process draws out the true relationship between dew point and tornadoes

  10. Scaled variable histograms (examples) Generally accept the red curve, however there are exceptions, like CAPE So when the IBTP is calculated, each variable is scaled based on the red curves

  11. Results and Verification Correlation with Tornadoes There are high inter-correlations between some variables. For example the 2 lapse rates are highly correlated with each other. The final variables chosen for the IBTP were : 1) CONVECTIVE AVAILABLE POTENTIAL ENERGY (CAPE) 2) LOW LEVEL RELATIVE HUMIDITY 3) 0-1 KM STORM RELATIVE HELICITY 4) 850-700 MB LAPSE RATE 5) 2 M DEW POINT 6) 2 M TEMPERATURE 7) LCL HEIGHT 8) MOISTURE CONVERGENCE 9) 0-6 KM VECTOR WIND SHEAR

  12. Results and Verification • IBTP correlates slightly better than STP for Plains tornadoes in 2010 • IBTP and STP values on the Plains grid are compared directly with the number of tornadoes that occur in the grid box during the forecast hour

  13. Specifics of Verification Correlations • Clearly there are thousands of grid points where no tornadoes occur… (532 tornadoes in 2010, versus 198,188 null points) • So if the parameter just forecasted 0 for all points at all times, it would correlate well with the gridded tornado reports • To get a sense of actual skill, I randomly select 532 null grid points and include them in the correlation calculation... So correct negative forecasts count equally as much as correct positive forecasts • I repeat this process 2000 times in a loop and average the correlation coefficients

  14. Example Case – May 10, 2010 Both IBTP and STP nicely capture the spatial extent of the tornadoes. Note that the 2 parameters are snapshots at 21 UTC whereas the SPC reports are for the entire day. Case studies are good but comprehensive statistical verification is desired to determine which parameter best predicts tornado likelihood.

  15. Example 2 – May 19, 2010

  16. Future Work • Double, Triple check that I’ve calculated these correlations accurately • Add 2011 into the independent verification • Publish?

  17. Example 3 – June 8, 2010

More Related