11.1.5 Conditional Probability of Precipitation Type
Feedback received after the 2003 field demonstration indicated that the end users had a
strong desire for information related to the evolution of precipitation phase during winter
weather events. Given the complexity of the physics and the difficulties that arise in
providing a deterministic output, the MDSS team chose to develop a probabilistic
precipitation type product and present it in an intuitive graphical form. Because it was
very well received by the end users, the graphical product is described here so that future
implementers can use this example to guide their own product design process.
The conditional probability of precipitation type describes the probability that the
precipitation will be rain, snow or ice if any precipitation occurs. The MDSS graphical
product combines the probability of precipitation and the conditional probability of
precipitation type. The conditional probability of precipitation type is normalized to the
total probability of precipitation so the user can get a quick sense of the likelihood of any
precipitation and the probability that it will be a specific precipitation type or mix of
types. An example of the product is shown in Fig. 11.6.
Fig. 11.6. Sample graphic showing the 48 hour forecast of conditional probability of
precipitation type for I-35 North of Des Moines, Iowa on 19 February 2004.
In this case, the total probability of precipitation rose to a peak of 75% between 6 and 7
AM on 19 February 2004. The event was predicted to start as rain and then slowly
transition to snow at around 9 AM. The MDSS is configured to declare a specific
precipitation type based on criteria that includes a combination of total probability of
precipitation, conditional probability of precipitation type, and precipitation rate. To
declare a precipitation event, the MDSS was configured as follows (for the winter 2003-
2004 demonstration):
a) Probability of precipitation was >= 25%
b) Precipitation rate was >= 0.01 mm/hour
105
Previous Page
