7.7 Accuracy in Aperture Photometry

As a further verification, aperture photometry of the five HII regions shown in Figure 7.1 was carried out on the HST data as well as on GAIA simulated flux map data obtained by stacking of 50 simulated observations with $6\times4$ pixels/sample. The center of the HII regions was determined by visual inspection, the signal counts inside a radius of 0.5 arcsec were summed and a median background calculated inside an annulus of radii 0.5 arcsec and 2 arcsec was subtracted. In order to take into account the different exposure time of the two images, the electron counts obtained from GAIA flux maps were multiplied by the ratio between the exposure time of the HST original image (900 s) and the effective total exposure time of a flux map reconstructed from 50 observations (43.09 s).

Table 7.3: Aperture photometry of five HII regions in HST original image and GAIA simulated flux map. HST electron counts, GAIA electron counts scaled to HST exposure time, error in magnitudes of GAIA aperture photometry with respect to HST aperture photometry and $V$ magnitude of the HII regions in HST original image. A flux map obtained from 50 observations with $6\times4$ pixels/sample was used. The position of the HII regions on HST original image is given in the left image of Figure 7.1.

HII region	$\mathrm{E}_{HST}$	$\mathrm{E}_{GAIA}$	$\frac{\mathrm{E}_{GAIA}-\mathrm{E}_{HST}}{\mathrm{E}_{HST}}$	$V_{HST}$
	e$^-$	e$^-$	mag	mag
1	192769	159549	$-0.172$	18.8492
2	147170	115019	$-0.218$	19.1423
3	178634	137214	$-0.232$	18.9319
4	222181	198978	$-0.104$	18.6950
5	361727	304130	$-0.159$	18.1659

The results are given in Table 7.3. The second column gives HST electron counts corrected for the background, the third one gives GAIA electron counts corrected for the background and scaled to HST exposure time, the fourth one gives the error in magnitudes of GAIA aperture photometry with respect to HST aperture photometry and the fifth one gives the total $V$ magnitude of the HII region according to HST data and to WFPC2 photometric calibration obtained by [Holtzman et al. 1995b], adopting the color index of $V-I=1.2$ from [Prugniel and Héraudeau 1998].

A bias of about 0.2 mag towards faint magnitudes is clearly seen, together with a standard error of the result from 50 observations of about 0.05 mag, estimated from the agreement between the five values in the fourth column. A more careful inspection of the data shows that this bias is due to an underestimation of the signal counts of about 0.1 mag as well as to an overestimation of the background counts of about 0.3 mag. Both biases are easily understood as due to the relatively wide wings of GAIA PSF, which cause some energy to fall out of the 0.5 arcsec radius and thus in the outer annulus. A smaller contribution to the systematic error affecting the background determination seems to originate from the smearing of almost point-like features like faint stars and cosmic ray hits present in the HST image. Their smearing cause their electron counts, which as far as possible should not be considered in the background calculation, to spread over a fairly large area and thus to ``escape'' the median filtering used to reject them in the calculation of the background. These results could undoubtedly be improved, especially by application of PSF photometry instead of aperture photometry, as described e.g. by [Høg, Fabricius, Knude and Makarov 1999], but this was beyond the scope of the present study.