How old is that picture

A volume scanOne full pass through every tilt in the pattern. A complete new picture of the storm arrives only when it finishes, which is where the four to six minute gap comes from, though extra low-level sweeps can refresh the bottom of it sooner. is a stack of cones

The radar does not take a picture. It sweeps a full circle at one elevation, tilts up, sweeps again, and repeats through a set of elevations. That set is a volume coverage pattern, and one complete pass through it is a volume scan. What your app shows as a single image is assembled from a sweep that took time to make.

Going back to the bottom, more often

Which is what the supplementary low-level scanning schemes exist to fix. Rather than waiting a full volume to look at the lowest tilt again, the radar inserts extra low-level sweeps inside the volume.

• One extra low-level cut roughly halves the time between looks at the bottom of the storm.
• Up to three extra cuts can bring the lowest-tilt update down to somewhere around 75 to 90 seconds.
• Terminating the high tilts early when there is nothing up there shortens the volume without losing anything, and buys the same effect a different way.

The reason any of this exists is in the record. The service assessment for the 2011 Joplin tornado found that "limited scans at lowest elevation slices during this time impacted the WFOWeather Forecast OfficeA local National Weather Service office. It issues the warnings for its own area, and it is the office your report reaches.'s ability to ascertain the magnitude of the tornado." Lesson 10 returns to that event.

Level IIThe full-resolution base radar data as recorded: reflectivity, velocity, spectrum width and the dual-polarisation fields at every elevation. Large, complete, and what research and the better desktop applications use. and Level IIIDerived and reduced radar products generated from Level II: single tilts, algorithm output, accumulations. Smaller and quicker to distribute, and what most public-facing displays are built from.

Level II

The full-resolution base data: reflectivityHow much of the radar pulse bounced back from whatever the beam was passing through on that tilt. It is the familiar green-to-red precipitation picture, and it describes what is up in the beam, not what is reaching the ground., velocityThe radar product showing how fast the rain, hail and debris in the beam are moving toward or away from the radar. Only motion along the beam is measured, and rotation shows up here as inbound and outbound speeds side by side., spectrum width and the dual-polarisationA radar upgrade that sends pulses both horizontally and vertically, so the returns say something about the shape of what they hit. It is what lets radar recognise lofted debris. fields, at every elevation, as the radar recorded them. Large, complete, and what research and the better desktop applications use.

Level III

Derived and reduced products generated from Level II: individual tilts, algorithm output, precipitation totals. Smaller, quicker to move, and most public-facing displays are built from it. It arrives after Level II because it is made from it.

The arithmetic that matters

Here is the calculation nobody does and everybody should. Your app shows an image. That image is old by the sum of several delays: the sweep itself took time, the volume had to finish and be assembled, products had to be generated, and the data had to reach your phone.

Two habits follow. Read the timestamp on every image, every time, and treat an app that does not show you one as unusable for this purpose. And watch the trend rather than the frame: three frames tell you where a storm is going, which partly compensates for the fact that no frame tells you where it is now.