eXe

BWER Algorithm

The bounded weak echo region is an inverted cavity of radar echoes in a thunderstorm that is evidence of a strong updraft and rotation. It is one of the markers of a supercell thunderstorm, and is well correlated to severe weather occurrence.

A BWER can be viewed by radar as a 3-dimensional region of weak echoes bounded by a 3-dimensional region of strong echoes above and surrounding it horizontally. It is similar to searching for an inverted cup pattern in the radar echoes.

The BWER detection algorithm was originally written by InfoMagnetic Technologies of Winnipeg and implemented in the Radar Decision Support System (RDSS) in Winnipeg in the 1990s. It was modified slightly when it was ported to URP in 2002.

The algorithm is implemented in a two pass approach.

In the first pass, for each bin a search is carried out in the horizontal and vertical directions in the vicinity of the bin, looking for a sufficiently strong reflectivity gradient that can be thought of as the BWER wall. For each search direction where a BWER wall is encountered, a count number is incremented. Thus if there are 13 horizontal and vertical search directions, the maximum possible count would be 13. A higher count is a stronger likelihood that the radar bin is inside a BWER.

It is possible that while one bin may have a high count and appear quite bounded, its neighbour may be able to see a hole in the echoes and thus have a lower count. This region is not a true BWER since not all bins inside the echo cavity are completed bounded by the BWER wall. Thus a second pass is carried out for each radar bin to assess the "degree of binding" of the radar bin's BWER count to its neighbours' BWER counts. For each bin a search is carried out in the horizontal and vertical directions in the vicinity of the bin, looking for BWER counts that are less than the central radar bin's BWER count. The search ends when a BWER wall is reached or when the maximum search distance is reached, whichever is shorter. If a lower BWER count is encountered along the path, then the central radar bin's BWER count is decremented by one. Thus a final BWER count is generated for each radar bin. If it is greater than or equal to a predefined threshold, then the radar bin is said to be inside a BWER.

The current URP version of the algorithm searches in 8 horizontal directions, 4 upward-diagonal directions, and the vertical direction, for a total of 13 directions. All directions are weighted equally. The reflectivity gradient threshold required to be called a BWER wall is a configurable 8 dBZ per radar bin. The minimum reflectivity required in the BWER wall is a configurable 40.5 dBZ. The search distance is a configurable number of radar bins, currently set to 10 bins. The BWER count threshold is set to a configurable value of 0, where the count must greater than 0 for the radar bin to be called a BWER bin.

Current URP BWER algorithm looks in 8 horizontal directions, 4 diagonal directions and one in the vertical

Algorithm Deficiencies

There are not enough search directions in the upward-diagonal direction. The vertical direction is weighted the same as any other direction, so if one is in clear air surrounded by storm cells, there may be a BWER count score of 12 out of 13, and yet there would be no BWER, or updraft or even thunderstorm above that location. The reflectivity gradient threshold is an all or nothing threshold, where 8 dBZ per radar bin counts as a BWER wall, whereas 7.5 dBZ per bin is counted as nothing. The search distance is set to the number of adjacent radar bins to check for a BWER wall. While 10 radar bins are 10 km in the radial direction, 10 radar bins in the azimuthal direction are about 20 km wide at a 120 km range from the radar, and 40 km wide at a 240 km range from the radar. It would be better to set the search distance in kilometres and adjust the radar bin limits accordingly. There is also a coding bug where BWERs cannot be detected as they pass through the zero degree azimuth straight north from the radar.

The URP BWER algorithm detects the existence of BWERs and calculates the maximum height of the cavity, but it flags too many false BWERs, especially near the radar. Because of a coding bug, it can actually misses BWERs directly north of the radar. There is a requirement to reduce the number of false alarms so that forecasters can have confidence that a flagged BWER is a real BWER. In addition, it is desirable to calculate the BWER volume, since this information will supplement the BWER height and provide another measure of storm cell intensity. The current URP BWER module is perhaps the slowest running of all the URP modules because it examines in detail all 92160 radar bin columns around the radar, probing for cavities. A faster running module would reduce the overall time it takes to produce storm cell assessments for the forecaster, allowing him to make more timely decisions.