Update: 4:06 pm. Doppler radar continues to show scattered showers and a few T-storms north of Redwood Falls. The rain is more widespread and heavier west of the Twin Cities. Our hit-or-miss shower pattern will linger into the evening hours, and a few storms over far southern MN may become severe. A tornado watch has been posted for Iowa and the extreme southern counties of Minnesota until 10 pm, including Austin, Albert Lea, Fairmont and Worthington. Conditions are ripe for potentially violent weather near the Iowa border - I do expect some warnings by the dinner hour. The combination of low-level moisture, ample wind shear and increasing instability will create an environment where rotating, spinning "supercell" thunderstorms can survive for several hours. These are the storms that show up on Doppler radar, but unfortunately only 2 or 3 out of 10 of these spinning T-storms go on to spin up a tornado. For that reason a significant percentage of "Doppler-indicated" tornado warnings turn out to be false alarms. Strong rotation almost always warrants a warning, but not ever spinning T-storm goes on to spawn a twister - the limitations of the science. The NWS and local meteorologists always err on the side of safety. We'd rather over-warn than under-warn. The biggest fear: tornado surivors staring into a television camera saying "it came without warning. There was no warning!" As a result of this paradox meteorologists tend to play it safe and if a rotating storm is detected - the warning goes out, even if there's not confirmation of a tornado from professionally trained SKYWARN spotters on the ground.
The bottom line: meteorologists don't want to be in a position of "crying wolf". But it beats the alternative of playing it so safe that the only tornado warnings that get issued are the result of storm spotters actually SEEING a tornado. If a storm spins a warning goes out. The system is far from perfect, but it shows the limits of science and Doppler radar technology - it only goes so far.