An improved algorithm for detecting blocking events
One shortcoming of these algorithms is that they cannot easily distinguish gradient reversals caused by the presence of a genuine blocking anticyclone from gradient reversals reflecting a mere saddle point between the primary polar cyclone and a cutoff low. This shortcoming can lead to false positives when establishing a climatology of (predicted or observed) blocking events.
We show that this problem can be remedied by incorporating diagnostics of streamline curvature into the blocking index algorithm. We call a flow pattern a block only if a spatially isolated, locally dominant extremum in anticyclonic curvature is found poleward of the location where a gradient reversal of -say- 500mb height is detected.
The requirement to incorporate velocity data into the detection algorithm is a complicating factor. However, the unfamiliar task of diagnosing streamline curvature is no more complex than that of diagnosing vorticity.