Estimate of the performances LBT-WFI


As discussed above, LIS is designed to reproduce actual images of the sky. We have extensively used this software to produce simulated images of the Hubble Deep Field (HDF) as if it was observed with LBT-WFI or - for comparison - with other instruments. Just as an example, in figure 14 the original HDF (30 hours in the F606W band) is displayed, to be compared with the expected V=22 frame from LBT-WFI (figure 15), observed with a much shorter integration time (0.5 hours). As will be detailed below, objects at V are easily detected in LBT-WFI even with this short integration time. In figure 16, the same HDF field is shown as expected from NTT-SUSI2, with a longer exposure time.



Figure 14:The Hubble Deep Field (chip #3) in the F606W band; total exposure time is 30 hours



Figure 15:A simulation of the same HDF field as observed with LBT-WFI in 1800 s of integration, with a seeing of 1''.



Figure 16:A simulation of the Hubble Deep Field as observed with NTT-SUSI2 in 7200 s of integration, with a seeing of 1''.

Images of this kind have been produced in several bands and for different exposure times, and analyzed using the standard software SExtractor (Bertin et al). The aim of this exercise is to estimate the performances of the LBT-WFI when the clustering, morphology and colors of real astronomical objects are taken into account. In addition, other instruments have been simulated (in particular NTT-SUSI2 and the VLT Test Camera) to allow a comparison with the LBT-WFI performances.

All images discussed below have been obtained with seeing 0.6'' or 1'', airmass 1.2, sky brightness of 3 days after new moon. Long exposure images have been assumed to result from the sum of several shorter exposures, and a flat-fielding accuracy of 0.5% has been assumed within each exposure.

Figure 17 (lower panel) presents the comparison between the faint galaxy counts in the V band obtained in the original HDF and those obtained in simulated LBT-WFI images of different exposure times. A somewhat pessimistic seeing of 1'' has been chose in this case. As expected, a significant bending of the counts from the HDF happens at progressively fainter magnitudes as the integration time increases. The counts show that in the V band, a $5\sigma$ limit of about 26, 27 and 27.5 mags can be obtained at integration time of 600s, 3600 and 10000 sec, respectively. The middle and upper panels show the same results for NTT-SUSI2 and the VLT Test Camera. As expected, LBT-WFI is more efficient. In particular, it is interesting to see that the advantage in efficiency over the VLT Test Camera (which is at the Cassegrain focus and has less efficient chips) is not degraded by the poorer sampling (the VLT-TC has a pixel scale of about 0.15''), at least under moderate seeing.

Figure 18 shows the results of a similar simulation in the U band (magnitudes are in the AB system, the transformation to standard Johnson is UJ = UAB - 0.85). In this case, $5\sigma$ limit of about UAB = 25.5, 26 and 27 mags can be obtained at integration times of 600s, 3600 and 10000 sec, respectively. Here the advantage over VLT-TC is significantly higher, thanks to the improved UV efficiency of the EEV chips.



Figure 17: Lower: A comparison between the galaxy counts in the HDF F606W image (solid line) with those extracted from LBT-WFI simulated images (in the V band) of different integration time. Middle: The same as lower panel, for NTT-SUSI2; Upper: The same as lower panel, for VLT Test Camera.



Figure 18: Lower: Same as figure 17, in the U band. Magnitudes are in the AB system,
UJ = UAB - 0.85)



Figure 19:The original HDF image. Encircled objects are not detected in the simulated LBT frame.



Figure 20:The corresponding image simulated at LBT-WFI. Circles show the position of HDF obects not detected in this image.

A closer scrutiny to figure 18 shows that some incompleteness in the counts holds at magnitudes fainter than 26 even in the deepest exposures, well before the detection limit. To investigate the origin of this effect, we have cross-correlated the HDF and LBT catalogs, looking for objects at 26 < V < 27 that are not detected in the LBT simulated images. As shown in figures 19 and 20, the undetected objects are typically faint objects located near a brighter one. In some cases, the failure to detect the object is due to the inability of the software to deconvolve the two objects. We have verified that a modification to the optimal SExtractor configuration parameter may lead to an improved deblending of some object, although at the price of a higher rate of spurious detections in the frame. The efficiency if the deblending efficiency depends obviously on the seeing conditions. Figure 21 shows a comparison between simulated LBT counts at different exposures time taken under 0.6'' and 1'' of seeing, respectively, showing that - as expected - the number of blended objects decrease when a lower seeing is used.



Figure 21:A comparison between the differential counts in LBT simulated images of the HDF at 1'' (see fig. 17) and 0.6''.



Figure 22:A simulation of the same HDF field as observed with LBT-WFI in 3600 s of integration, with a seeing of 0.6''.
Next: Summary