Now that you're fluent in Greenwich time, you know that 06Z is approximately midnight (give or take an hour or two, depending on the time zone). That's intentional. The period from 06Z to 06Z corresponds roughly to a single calendar day. Typically, the minimum temperature during that period will take place around or a little before 12Z (just before sunrise), and the maximum temperature will take place around 21Z (in midafternoon). So you're basically making a simple one-day forecast.
You must forecast the maximum temperature, in degrees Fahrenheit; the minimum temperature, in degrees Fahrenheit; and the precipitation amount, by category. For the NCWFC, there are six categories, numbered 0 to 5:
| 0 | no measurable precipitation |
| 1 | precipitation amount 0.01 to 0.05 |
| 2 | precipitation amount 0.06 to 0.24 |
| 3 | precipitation amount 0.25 to 0.49 |
| 4 | precipitation amount 0.50 to 0.99 |
| 5 | precipitation amount 1 inch and above |
Each category, except for 0, has about the same probability of coming to pass. There's also a special precipitation amount which can be repeated, called a "trace". This means precipitation reached the ground at the observing station, but that less than 0.01 inches accumulated. If a trace happens, then a forecast of either category 0 or category 1 (or, in forecaster lingo, cat 0 or cat 1) are deemed to be correct.
Forecasts are scored by "error points". Each degree off on the maximum or minimum temperatures (or, in forecaster lingo, the max or min) is worth one error point. Each category off on the precipitation is worth four error points. Over a typical two-week period, you'll probably have as many error points on the max or min as you will on the precip category.
I'll score an example forecast: suppose you forecast a high of 59, a low of 42, and a precip category of 2. What actually happens is, the high is 60, the low is 38, and they receive 0.27 inches of rain. (These numbers are known as the "verification".) You would get 1 error point for the max, 4 error points for the min, and 4 error points for the rain (category 3 instead of category 2), for a total of 9 error points. Generally, a total error of 6 or less is very good, but it depends a lot on the weather.
Questions or Comments
Technical: E-mail John Fulton < jdfult@nimbus.met.tamu.edu >
Scientific: E-mail Dr. John Nielsen-Gammon. < nielsen@ariel.met.tamu.edu >
Copyright © 1996-2003 Texas A&M University, Texas A&M Atmospheric Sciences Department and Dr. John Nielsen-Gammon. All rights reserved.