February 3, 2000


All three predictands are valid on the hour every 3 hours from 6 to 72 hours after the initial time of the 0000 and 1200 UTC AVN model runs. The predictands were derived from the present weather reports in the hourly surface observations.


Developmental Sample: Forecast equations were developed using data from the following dates:

CONUS sites:
Sep. 16, 1997 -> May 15, 1998
Sep. 16, 1998 -> May 15, 1999
Alaska sites:
Sep. 1, 1997 -> May 31, 1998
Sep. 1, 1998 -> May 31, 1999

Test Sample: In order to evaluate the proposed MOS system, we divided the developmental sample into two parts: a dependent sample used to develop test forecast equations, and an independent sample on which to evaluate the forecasts made by the test equations. We chose an 88 day independent test period for verification consisting of:

Oct. 16 -> 31, 1998
Nov. 16 -> 30, 1998
Dec. 16 -> 31, 1998
Jan. 16 -> 31, 1999
Feb. 16 -> 28, 1999
Mar. 16 -> 31, 1999

Verification: The P-scores for the operational NGM MOS, primary AVN MOS, and secondary AVN MOS precipitation type probability forecasts will be shown at the February 3 CAFTI meeting, and are available on the Web at www.nws.noaa.gov/tdl/synop/results.htm Results were computed for both the 0000 and 1200 UTC model cycles for projections from 6 to 72 hours after the initial model runs. The P-scores for 283 sites in the CONUS and Alaska combined show that the new AVN MOS precipitation type guidance is comparable in accuracy to the operational NGM MOS guidance at all projections. For the 0000 UTC cycle, the AVN MOS is as accurate if not more accurate than the NGM MOS at virtually all projections beyond 30 hours.