The dprog/dt feature in AWIPS allows a user to see model trends for a specified forecast time by presenting a sequence of charts taken from previous runs, each valid at the same time. Implied in the trends is the idea that the model makes only partial corrections from one run to the next because errors in the short term forecast are transmitted into the initialization of the next run.

The applications shown here attempt to remedy the problem by completing the trends to create an "extrapolated" run. This run would normally never be seen because the next operational run is available after only a few hours. Examples on D2D show charts from the extrapolated run overlaid with those of the newest run. Verification to the eventual initialization shows how well the extrapolated run compares with the newest run alone.

In order to preserve the internal meteorological relationships it is necessary to operate on entire model runs (i.e., all fields), treating each run as a complete entity. Since both the newest run and the preceding run(s) preserve the relationships separately, the linear combination (based on the mathematical serpentine function) which generates the extrapolated run, should preserve them also.