The first frame effect correction seems to depend most strongly on the amount of time that has passed since the last read of the array. However, it is not a simple or even montonic function of time. Therefore, we will correct it using an empirical lookup table approach. The algorithm will work as follows:

1. When correcting a single DCE, when the module is run it queries the ending time of the previous DCE made with this array. The module can either make the query itself, or receive it as a command line entry from a third entity such as the cal server. The ending time can be derived from the difference between AINTBEG of the current DCE and ATIMEEND of the previous DCE. This approach currently may fail due to a firmware bug, but is anticipated to be fixed in a later revision. Alternately, if both DCEs have not yet actually been downlinked then this value could be approximated by differencing AINTBEG of the current DCE and the predicted end of the previous DCE based on it's commanded start time and commanded frame time, which presumably exist in the SODB.

2. Once the delta time has been determined, the module will compute the correction to be made from a lookup table. This lookup table will be provided by the instrument team and will consist of a column of times along with corresponding offsets. For times not explicitly included in the table the value should be interpolated from the table. There may be multiple tables per array, corresponding to different combinations of fowler number and wait time.

3. To apply the correction (which may be negative), the offset should be added to every pixel in the image. In adddition to "value offset" mode, the module should also allow an optional "image offset" mode, in which the correction factor is used to scale an input image (also provided by the IT), which is then added to the DCE being processed. In principle the module could always do the latter, using a flat input image, but using a constant might save in file-reading overhead.