Enhancing tornado intensity estimates with Doppler radar

We are developing a means to supplement subjective methods of evaluating tornado intensity using operational Doppler radar networks. Accepted method s of estimating tornado intensity and recording event occurrences have introduced subjectivity to the existing tornado climatology, with population biases intrinsic to the dataset . Using the Weather Surveillance Radar – 1988 Doppler (WSR-88D) network, we relate WSR-88D velocity data to higher resolution, tornado-scale velocity measurements from mobile Doppler on Wheels (DOW) research radars rather than to damage. Using these methods, we are creating a statistical model that will allow us to objectively estimate ground-level intensity of tornadoes using only the WSR-88D network.

For each event, we determine a maximum differential velocity measured by the DOW (ΔVDOW); we then determine the associated maximum of the WSR-88D differential velocity (ΔV88D) within a 15-minute time window, to allow for variations in sampling. Our current regression model is based on 10 events with confirmed tornadoes and exhibits the following linear relationship: ΔV88D ≈ 0.4(ΔVDOW) + 19, with a linear correlation coefficient of 0.84. Ongoing work to include additional cases, including null (non-tornadic) events, is underway. We acknowledge that there are limitations to our simple model, which presumes the velocity data necessarily are associated with a confirmed tornado; we have begun to address (1) the range degradation issues described by Wood and Brown (1997) as well as (2) the uncertainty related to the sampling of multiple vortex tornadoes. This is being achieved currently with the aid of simple vortex models.