Ticker for April 21, 2003

                
MESONET TICKER ... MESONET TICKER ... MESONET TICKER ... MESONET TICKER ...
April 21, 2003     April 21, 2003     April 21, 2003     April 21, 2003


Point of Clarification

Yesterday's Ticker flippantly and clumsily mingled the concepts of
"records set" and "occurrence of extreme events", resulting in more
than a little feedback.  More than a little feedback calls for more
than a little explaining, so here it goes:

Despite our language, we didn't count the number of "broken records",
we counted the number of "extreme events", meaning we ran through the
entire data set before we declared a "winner".  In a static climate,
these should be randomly distributed within the dataset.

With that being said, there is a small systematic bias in the method
we employed Friday: in the event of a tie, we only gave credit to the
first event.  So, if the extreme for July 12th at Ardmore was 106F,
set in 1936 and 1984, only the 1930s got credit.

We also did it the opposite way, in which the latest occurrence got
credit.  This changed the numbers very slightly, but the overall
results were largely the same.

What we're brewing now is a method such that multiple occurrences
"share" a fraction of the record.  In other words, if the extreme
event occurred in 1934, 1936 and 1978, then the 30s will get credit
for .66667 of a record and the 70s will get credit for .33333.

Thanks for all the feedback and for keeping our language on its toes.