Radar Algorithms

Radar is a visual tool. The meteorologist looks at radar images in the hope that certain signatures will be present based on a conceptual model of an event that has been created using prior analysis and diagnosis. For example, if the meteorologist believes that supercells are likely, they will be looking for certain behaviors from cells on radar. A meteorologist expecting supercell behavior is not going to be looking for bright bands associated with freezing rain.

Many of the signatures that appear on a radar image are subtle and are not readily apparent to the meteorologist's eye. Significant signatures can be just a few pixels across and can easily get lost in the riot of colours that sometimes appear on a radar image. The problem is exacerbated by requirements that the meteorologist keep track of not just one radar but perhaps three, five, or even ten radars at the same time.

This is where radar algorithms come in. They mine the digital information contained within the radar's volume scan to highlight relevant signatures that otherwise might be missed by the meteorologist. They also can remove irrelevant signatures, like AP.

No algorithm is perfect; they all come with caveats and asterisks for their use. A meteorologist who does not understand how a particular algorithm works will quickly be reduced to blindly following the algorithm's output, often with detrimental results. Experience with the Radar Decision Support System (RDSS), a first-generation expert system developed in the mid-1990s by the forecast center is Winnipeg, showed that meteorologists who weren't familiar with the algorithms being used in RDSS often issued warning based on colours; if RDSS flagged a cell as being severe (it went red), warnings were issued, even when the cell was flagged because of problems with the algorithm.

Steve Weiss, from the Storm Prediction Center (SPC) in Norman, Oklahoma during a keynote address at the last Forecaster's Forum held by the MSC in 2006, said that if a meteorologist does not know how an index is calculated or how an algorithm works, that meteorologist has no business using it.

The goal of this module is to highlight current algorithms in use by the MSC's Unified Radar Processor (URP), improved radar algorithm, and an validation that has been done to test them. Much of the on-going algorithm development is targeted toward inclusion in NinJo.