The special terms were calculated as follows:

X terms, from geom ISR, multiplied by X ratio below | ||||

0.0 | * | 0.0 | 2.0 | 2.1 |

2.2 | 0.0 | 0.0 | 0.0 | 0.0 |

Y terms, from geom ISR, multiplied by Y ratio below | ||||

0.0 | 1.0 | * | 2.0 | 2.1 |

2.2 | 0.0 | 0.0 | 0.0 | 0.0 |

The "*" terms are the ratio of the X or Y scale at a particular epoch to the mean scale, quoted in the header of the geom file. This is the geometric mean scale of the detector, and has not changed since launch.

For example, in the NIC2 geom file these two terms would be

(X+Y)/2. 1.00451 = 0.0760062 (X mean from day 309-632) / 0.0756649 0.995488 = 0.0753235 (Y mean from day 309-632) / 0.0756649 ^ |---- These are the time-dependent scale numbersYou would only need to change the ratios to the geometric mean if the X and Y scales changed *relative* to each other. The global change of the overall nicmos focus does not change the coeffs in any way (because the ratio of X to Y stays the same). These coeffs allow drizzle to make the pixels square, and since drizzle works in pixel space it doesn't care what the geometric mean scale is doing with time.

If you have data spanning a large time period (long enough for the geometric mean scale to change) then you would account for this in calculating your xoff and yoff inputs to drizzle. (i.e. multiplying each arcsecond offset by the appropriate plate scale to get pixels at each epoch). There is NO NEED to change this file die to a mean plate scale change.

Download the geometric distortion coefficient files: