Notes from Bob Hartman Last Revision February 04, 2003

There are really two parts to this, statistical accuracy  and systematic uncertainties. It seems highly unlikely that the beam calibrations can achieve the desired statistical uncertainties over a significant fraction of the (energy, theta, phi) space. It appears more reasonable to expect that simulations will provide the required statistics, and that the beam calibrations should be used to establish fiducial points for comparison, plus more detailed measurements at points where the simulations seem vulnerable. As a starting point, I would like to see the statistical (simulation) accuracy at least a factor of two below the ultimate requirements. Here are my suggestions:

Knowledge of Parameters

Parameter LAT requirements Goal EGRET
Aeff (>100 MeV) 20 % 10 % 10-15% > 70 MeV
Aeff (<100 MeV) 50 % 10 % 50-100% < 70 MeV (contamination)
E_resolution (>100 MeV) ?? 10% of resolution ~10% of resolution
E_resolution (<100 MeV) ?? 10% of resolution ~10% of resolution

PSF (68% & 95% containment):

Although these are convenient numbers for describing the resolution, the science analysis will use actual point spread functions, which for EGRET are sums of Gaussians, fitted to the calibration results (See figs 17, 18 of Thompson et al. 93). The shape of the PSF at each energy needs to be determined rather well in order to avoid artifacts in data analysis. In the end, the EGRET PSF's were better-determined from observations of bright pulsars than from ground calibration, because EGRET lasted longer than anticipated at calibration time.

 

Angle of incidence:

(Assuming this means the knowledge of the orientation of the LAT to the calibration beam) For EGRET, this was known to about 0.1 degree. Although not obvious initially, this turned out to be important because of what we call the the "fish-eye effect", i.e., the appearance of known sources closer to the axis than they should have been, when seen at wide angles. The source of this effect was never conclusively demonstrated, but is believed to be related to multiple scattering, supposedly causing increased detection efficiency toward the axis (as opposed to away from.) Only important at low energies. Since LAT has a much wider FOV than EGRET, this could be important. (It was first discovered in-flight in observations of known pulsars well off-axis, later confirmed and better quantified from calibration data.) For LAT, only the orientation of the tracker to the beam is likely to be significant.

Beam Flux:

A necessary component of the effective area determination. Its determination may dominate the error budget for effective area.

Deadtime:

In EGRET, this was done on-board in at least two different ways: First, each event telemetered contained a measurement of how long the trigger was disabled by that event, including any backup in the buffer. For LAT, a similar measurement would be more complex, because it would have to include also the time lost to triggers that didn't get telemetered. Second, there was an independent deadtime scaler that measured the deadtime (or maybe livetime, can't remember) in each 2.048 second packet. Fortunately, the two measurements agreed. However, during gamma-ray bursts, when the instrument was triggering at max rate, they weren'tgood enough to give a good livetime fraction. (We didn't anticipateseeing GeV gammas from GRB's.)