Transitions Newsletter Header

Community Connections: Object-based Verification at WPC

Summer 2013

The Weather Prediction Center (WPC) at NCEP has been using MODE to supplement its traditional verification techniques since April 2010. The Method for Object-based Diagnostic Evaluation (MODE), a utility that is part of the MET verification package, has been developed with substantial support from the DTC. Both are systematically expanded and maintained for specific DTC tasks and an array of outside users. MODE output is available to WPC forecasters in real time through an internal website that displays graphical verification for forecasts of 24 hr precipitation valid at 1200 UTC (see the figure). Forecasters can select the forecast lead time (Day 1 – 36 hr, Day 2 – 60 hr, or Day 3 – 84 hr) and precipitation threshold (0.50 in, 1.0 in, and 2.0 in), then view the corresponding verification for WPC forecasts and 9 numerical models.

“Two things that seem to resonate with our forecasters the most are the real time aspect of the website and the visual nature of the comparison.”

The graphical nature of the MODE verification allows for a quick comparison of forecasts in a way that goes beyond traditional threat scores and bias values to consider other measures of forecast quality (distance between forecast and observed objects, differences in angle of orientation and object size, etc.). The most recent update to the website attempts to better quantify these qualities by adding statistical comparisons of the interest value and the displacement distance between matched objects to complement the traditional graphical comparisons. Future plans include additional statistical information on the website, including longer term summaries (monthly, annually, etc.), and making the website available to the public.

Running MODE on a national scale at an operational center like WPC can present some unique challenges since MODE must be able to correctly identify precipitation objects from meteorological phenomena as varied as cool season synoptic scale storms to warm season convection. Determining the ideal configuration is still a work in progress, but it is an essential piece of the puzzle in order to build forecaster confidence in the utility of object-based verification.

Contributed by Faye Barthild, NCEP WPC