For a given time, this model estimates the amount of defocus at a particular camera. The model is a function of telemetered temperatures and secular non-thermal terms. The temperature-dependent part of the model is taken from Di Nino et al. 2008 while the long-term secular component is based on more recent fits given in Niemi et al., 2010. In addition to the temperature terms and secular fits, the model includes zero point offsets characterizing the focus offsets between cameras and channels.
Note (May 2014): We have re-posted a revised focus model, containing an updated secular fit to better characterize the focus trend since March 29, 2011. We have added a linear term with slope of -4.232e-03 microns/day or -1.55 microns a year beginning on March 29th 2011. This correction has been applied to the online model and the yearly summary files. Any results measured from this site before May 23rd 2014 will not have this correction applied. If you obtained focus model values from this site prior to May 23rd, for HST observations made after March 29th 2011, then obtaining those values again now will improve agreement with measurements.
Note (February 2020): Studies during 2019 incorporating a more reliable focus measurement method have resulted in an improved model. Also the linear term referred to above has been modified so as not to apply beyond 29 August 2014. Any values obtained from the model beyond the end of 2014 should be reviewed and replaced by the values produced by a rerunning of the current model.
Because long-term fits must be occasionally performed and added to the temperature-dependent component of the model, the model will over time accumulate a zero point offset. While these are occasionally removed (see the above notes), it is always generally true that the model is more useful and descriptive when applied to understand relative focus changes over shorter periods (i.e., orbits/days).
Resources for Hubble’s Focus Model
Accordion
The interface to the model presents you with the following input choices:
- Display: You have the option of displaying:
- the focus position calculated by the model,
- an actual measured focus position if available,
- a plot comparing both.
- Camera: You may choose to model the focus in ACS (HRC,WFC1, or WFC2), WFC3 (UVIS1 or UVIS2), or WFPC2 (PC).
- Year: Defaults to current year
- Date: Supply in mm/dd format
- Time Period:
- For the measured or comparison display options, a drop-down menu offers the actual times when measurements are available.
- For the modeled focus display option, you supply the date and start/stop times in 24 hour hh:mm format (UT, same frame used in HST science image headers). Any time period falling within 1 calendar day can be supplied.
The model produces an output webpage (example) containing the following:
- the average focus value over the time period requested (measured and/or modeled values as applicable).
- a link to a space-delimited text file containing Date, Julian date, Time, and Model value
- an in-line plot, downloadable as a Portable Network Graphic (.png) file.
All focus values are expressed as the number of microns of longitudinal motion of the Secondary Mirror (sometimes referred to as "piston" or "despace"). 1 micron at the secondary induces 6 nanometers of rms wavefront error.
A summary of model values covering one year at a time may be obtained as a single text file. The values are given as a function of Modified Julian date along with a time in normal date and time format. The focus values given are for a central camera position. Camera specific corrections are up to about half a micron and are listed below in units of microns.
WFPC2 PC 0.23
ACS HRC -0.25
ACS WFC1 0.09
ACS WFC2 -0.09
WFC3 UVIS1 -0.34
WFC3 UVIS2 0.53
Results are available by selecting one of the years below. A right click allows you to download the file.
The results of comparing the model with the actual measurements normally show agreement within a few microns. A fuller discussion of the model's fidelity is given in Cox & Niemi 2011. The differences between the measured and modeled values are summarized in the diagram below extracted from the ISR. It shows the cumulative fraction of cases which fall within any deviation. 1 sigma accuracy is ~1.6 microns (depending on the specific SI), and 80% of the time the agreement is within 2 microns, though occasional errors of up to 4 microns are seen. Since normal fluctuations within an orbit can be 6 microns or more, the model provides useful improvements to the knowledge of the focus position.
The Tiny Tim point-spread function modeling software has been modified recently to accept output from this focus model (Krist & Hook 2011). The application of modeled and empirical PSFs to science analysis is discussed briefly elsewhere.