These values were chosen based on subjective analyses prior to and during the early part
of the field season. The precipitation type was declared when the following criteria were
met:
a) Ice declared, if the conditional probability of ice was >= 25%
b) If the conditional probability of ice was < 25%, the declared precipitation
type is the maximum conditional probability between rain and snow.
The users indicated and reiterated several times that ice is a large winter maintenance
problem and they wanted to have a conservative heads-up if there was any chance of
freezing rain; therefore, the 25% threshold was chosen. In the graphic example above, the
declared precipitation type is shown above the precipitation probability graph.
The performance of the precipitation type field was judged based on what conditional
probability values were forecast for each precipitation type when a given precipitation
type was forecast. This was done for all METAR sites in Iowa and the results were
combined. For example, snow was observed 1697 times (hourly observations) during the
project, and on 81% of those occasions, the conditional probability of snow forecast for
all of those times fell within the highest bin (90-100% chance that precipitation will fall
as snow; Fig. 11.7a). For those same times, the conditional probabilities of rain and ice
were 0-20% almost every time. Rain was observed 694 times during the project, and on
two-thirds of those occasions, the conditional probability of rain exceeded 80%, while the
conditional probability of snow rarely exceeded 80% (17 times; Fig. 11.7b). Results for
the freezing rain were inconclusive because it was almost never observed during the 2004
demonstration.
106
Previous Page
