[Cowystats] COWY-ASA Reminder Tom Hamill Feb 28th; pharmaSUG scholarhip.

[Matt Pocernich]

## Feb 28  Tom Hamill (NOAA) Improving Probabilistic Weather Forecasts
## Jun 3-6  PharmaSUG, June 3-6 - new student scholarship -
scholarship deadline March 30th.


Next Wednesday, Tom Hamill will be speaking at Metro.  While
the topic is weather, the issues discussed in the talk apply to many
areas of statistical modelling.  I have seen Tom speak on several
occasions and can guarantee and interesting talk.

Secondly, registration is open for PharmaSUG (SUG = SAS Users Group). 
More information is listed below.  This year they have started a student 
scholarship program.  

No one has asked but if there are students applying for scholarships
such as this one or lets say the Gertrude Cox scholarship who would
like comments on their application I suspect the chapter has members
with willing to critique these applications.  Just at thought.


Thanks,

Matt

##################################

Wednesday February 28th - 4pm 

Improving Probabilistic Weather Forecasts

Thomas M. Hamill
NOAA Earth System Research Lab, Physical Sciences Division
Boulder, CO

Metro State College Denver 
Jointly sponsored by Metro Department of Meteorology, Metro Department of Statistics and CO/WY ASA. 
Refreshments provided.

In the Science Building Room 103  Metro State College Denver
See  http://www.mscd.edu/enroll/admissions/campus/map/ for a campus
and parking information.  Note that the campus is very well served by
RTD light rail lines.

Abstract ***

Weather forecasts, we all know, inevitably have errors, sometimes small,
sometimes large. There are two main sources of errors in numerical
weather predictions.  The first is "chaos," or the sensitive dependence
on the initial condition.  Any small error in the initial description of
the weather will tend to grow larger and larger as the forecast lead
increases, though the error growth rate may vary from one day to the
next.  The other source of error is imperfections in the numerical model
used to make the weather forecast.

While the current practice is to issue deterministic forecasts ("high of
60 today") a more intellectually honest way of communicating the
forecast is to do so probabilistically, indicating in some fashion the
uncertainty in the forecast.
In this talk we will demonstrate how a long time series of past
forecasts, which we call "reforecasts" and associated weather
observations can be used to generate skillful, reliable probabilistic
forecasts, forecasts that account for the uncertainty due to chaos,
forecasts that statistically adjust for many of the systematic errors
introduced by the forecast model.  We'll show that a statistically
adjusted weather forecast is much more skillful than a forecast without
this statistical adjustment.



##################

Registration is open for PharmaSUG, June 3-6, at the Hyatt Regency
Denver at Colorado Convention Center. The website is
www.pharmasug.org. Early registration ends April 20th. 

This year we've started a student scholarship program. Details can be
found  on the website.  Follow the links to registration.

http://www.pharmasug.org/content/view/21/43/


-- 
Matt Pocernich
National Center for Atmospheric Research
Research Applications Laboratory
(303) 497-8312