The 15th International Conference on Interactive Information and Processing Systems(IIPS) for Meteorology, Oceanography, and Hydrology

P1.5
AWIPS INTERACTIVE FORECAST VERIFICATION

James M. Frederick, NOAA/NWS, Tulsa, OK; and S. A. Amburn

Verification of forecasts is, or should be a critical function at any forecast office. Verification provides an assessment forecast quality, a baseline for judging changes in either forecast techniques or models, and also a mechanism to provide forecasters with information to learn from either their mistakes or successes. Assessing forecast quality and developing forecast baselines have been done for many years. However, to provide a forecaster with the information to help him/her learn and improve forecast skills requires near-real-time, interactive verification. This kind of verification (called SOOVER) has been developed at WFO Tulsa and has been operational in the AWIPS environment since October 1996.

Both statistical and graphical verification are available from a pull-down menu on AWIPS. The forecasters' temperature and precipitation probability forecast numbers (CCF) for any number of sites and MOS (FWC/FAN) numbers are verified for either four or five periods. Separate statistics are available for each period, or combined statistics are available for an entire forecast cycle. Absolute errors and biases for temperatures are produced for 00 UTC and 12 UTC forecast cycles. Temperature error frequency distribution, mean wet-PoP, and mean dry-PoP tables are also shown. The statistics may also be selected to show verification for the previous day, days, weeks, or even months. This is frequently helpful in determining MOS biases and trends from the previous few days. These types of data are also available in graphical mode to provide the information in a more visual manner.

The verification statistics are updated every 12 hours so a forecaster may review his/her first period verification at the beginning of shift the next day. This is desirable so the forecaster can see forecast outcomes while still remembering charts and other guidance on which the forecast was made. Forecast verification which is available days or weeks later is virtually useless to the forecaster as a learning tool.

Very little manual effort is required for SOOVER to run. Temperature and precipitation data are manually input to provide maximum quality control of the observations. Otherwise, all MOS and forecaster data is obtained automatically. The program runs from a cron and scripts located in AWIPS, with the SOOVER GUI written in C and Motif

The 15th International Conference on Interactive Information and Processing Systems(IIPS) for Meteorology, Oceanography, and Hydrology