For the U.S. region, 4 Cressman passes are used, the finest of which is 250 km. For regional domains 3 passes range from 600-800 km down to about 200 km.
| Quality control: | Simple 1-pass large-influence objective analysis is interpolated back to the stations, where a difference field is produced. A set criteria for each field is used to decide what obs to toss. This is rather resolution dependent and, especially for the U.S. region, too many obs are being flagged right now. |
| Super-ob'ing | Isn't done. This means that major metro areas with many obs (e.g. Chicago) have too much influence on the analysis. The Chicago Blues (musical) influence is welcome, though. |
| First guess: | No FG is used. Particularly for hourly sfc analyses, using the previous hour would probably work well. But the quality control (QC) is simple, and I don't want the influence of bad observations to linger. There are hooks in the code to make use of a FG, but they are not in use now. |
| Data: | Standard METAR reports from the U.S. and Canada. I don't have the station tables set up to use Mexican reports or buoy data, yet. |
Rather more sophisticated analyses can be found from the Mesoscale Analysis and Prediction System and Local Analysis and Prediction System at the Forecast Systems Lab in Boulder, and at NCEP, CAPS/ADAS, and NCAR/RAP. See also the products available at Purdue, OSU, and COLA. Local detailed analyses are also run at various National Weather Service offices, such as this from NWSFO LBB.