ATS641 Forecast Contest Rules

The class, instructor, TA, and models will forecast against one another during two contests this semester. Performance will not affect grades, but participation will.

1. Locations

In order to familiarize ourselves with local weather and climate, we will forecast for Fort Collins every class day during each contest. Since predicting our local weather isn't always challenging, we will also forecast for a second, "away" city determined each day by the TA.

2. Schedule

We will forecast for 24 hour periods:
(1) from 1900 MST (2000 MDT) that day to 1900 MST (2000 MDT) the next day for Fort Collins, and
(2) from 0600 UTC that day to 0600 UTC the next day for the away city.
Forecasts are due promptly by 1700 LOCAL TIME (5pm).

3. Parameters

Forecasts will consist of three parameters: maximum temperature, minimum temperature, and 24-hour liquid (equivalent) precipitation accumulation. All forecast temperatures will be in Fahrenheit (F), and forecast precipitation in inches.

4. Submission Procedure

Submit your forecasts at: http://schumacher.atmos.colostate.edu/teaching/ats641/forecast_contest

username = Lastname (first letter capitalized!)
password = 1st two letters of first name + 1st two letters of last name + provided 4-digit numerical code

If a web browser is not available to you for a particular forecast, emailed submissions to Allie.Mazurek@colostate.edu are acceptable (though they require more work to process).

5. Verification

We will verify Fort Collins' weather at the campus station (FCL), and the away city's via METAR SAOs.

6. Scoring

One error point will be assessed for each °F the maximum and minimum temperature forecasts vary from observed values. Error points for precipitation will be assigned in the same was as the national WxChallenge contest:

.4 points for each .01 of error in the verification range from 0.00 - .10 inclusive
.3 points for each .01 of error in the verification range from 0.11 to 0.25 inclusive
.2 points for each .01 of error in the verification range from 0.26 to 0.50 inclusive
.1 points for each .01 of error in the verification range over 0.50

This is NOT an absolute error range. The score depends where you are on the scale. (Note: A verification of trace precipitation will be scored as 0.00" of precipitation.)

If you will be out of town for a conference or other such reason and will be unable to forecast, you may e-mail the TA in advance and request a model forecast for that period. Otherwise, late and absent forecasts will be given the consensus forecasts plus ten error points per city.

"Other Players"

To evaluate our performance with respect to some commonly available benchmarks, the TA will also enter forecasts for the following:

Class Consensus
Operational model output
National Weather Service

Consensus

The consensus player will represent the average forecast of all players (i.e. students, TA, instructor, models, and NWS). Typically, consensus is one of the best players in the game, as it eliminates a bit of each person's bias and incorporates (to some extent) everyone's individual insight. In order to beat consensus in the long run, you must have a good deal more skill than the collective skill of everyone.

Operational model output

The forecast models will get to play via the model output statistics (MOS) from NAM and GFS. The models will be given the MOS high and low temperature forecasts, and the raw gridded accumulation from that model. Away cities will be chosen to coincide with MOS output. For Fort Collins, the forecast temperatures and precipitation from KFNL will be used. Try to beat the models, or you will be phased out as a forecaster!

Page loaded at 07:14:23 MST from 3.15.31.27 (ec2-3-15-31-27.us-east-2.compute.amazonaws.com).
Created and maintained by Brian McNoldy, 2002-2011.