In article <[EMAIL PROTECTED]>, "Richard A. Beldin"
<[EMAIL PROTECTED]> wrote:
> This is a multi-part message in MIME format.
> --------------3C6331B8C260BF767681A8B3
> Content-Type: text/plain; charset=us-ascii
> Content-Transfer-Encoding: 7bit
>
> The "definition" via axioms provides a mathematical structure that we
> interpret either as "relative frequency" or as "degrees of belief".
> Indeed, I think that any phenomenon which satisfies the axioms can serve
> as an "interpretation". As they say, "If it walks like a duck, ...".
>
> As far as the probability of rain tomorrow, I always explained to my
> students that the language is so imprecise that the numerical value has
> only rhetorical utility. We need to know:
> 1) How much rain in cm. ?
> 2) In which locations?
> 3) During what time span?
>
> Does 70% probability mean that it rains in 70% of the locations or 70%
> of the time or what?
>
> Your instincts are correct. That example is severely flawed because we
> have not made the experiment clear.
>
> Continue to question the simple examples. You will learn from it.
Although in this case, in the US, the definition is quite precise for
the National Weather Service. From their Operations Manual, it is "the
likelihood of occurrence...of a precipitation event at any given point
in the forecast area. The time period to which the PoP applies must
be clearly stated (or unambiguously inferred from the forecast wording)
since, without this, a numerical PoP value is meaningless." Elsewhere
in the Ops Manual, a "precipitation event" is the occurrence of at least
0.01" of liquid equivalent precipitation (i.e., rain, melted snow). It
has been shown that, given this definition, the PoP is equal to the
expected areal coverage of the precipitation.
Whether other groups issuing PoPs (media, other countries' weather
services) follow the same definition, I don't know. At the locations
where verification data are available, NWS PoP forecasts out through
48 hours are remarkably reliable. That is, if the PoP is N%,
precipitation occurs very close to N% of the time.
Harold
--
[EMAIL PROTECTED] http://www.nssl.noaa.gov/~brooks/
Standard disclaimer
Head, Mesoscale Applications Group, National Severe Storms Laboratory
Norman, Oklahoma
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
