Subject: SST IFMOS Date: Thu, 18 Feb 1999 13:06:25 GMT From: Santiago Arribas To: rfosbury@eso.org CC: pjakobse@estsa2.estec.esa.nl, lefevre@astrsp-mrs.fr, ghasinger@aip.de, sam@chantada.ll.iac.es Dear Bob, I have done a quick comparison between the Yan et al. and Thompson et al. data and they seem to agree rather well (in the plane half-light radius versus magnitude). Of course the data by Thompson et al, are prefereable as they contain fainter galaxies, and are less affected by surface brightness selection effects. In any case, before obtaining improved curves/tables for achiving density of objects and exposures times perhaps we should discuss about the general procedures to obtain these tables/curves. So let me summarize how I see this issue. The determination of density of sources is conceptually rather straightforward. The main problem until now is the disagreement (by a factor 2-3) among different observational data at the faintest end (K~23, see, for instance, Bershady et al, 1998, ApJ, 505, 50). However, the HDF NICMOS data (although with a limited spatial coverage) should allow to solve this well (e.g. Thompson et al.). Regarding the exposure time determination the problem is much more complex. One can use the NGST-ETC (Exposure Time Calculator) to have croase estimations. However now ETC has no much flexibility and requires some modelling. In fact for galaxies an Lstar absolute luminosity is always assumed and it is mandatory to specify the redshift. In addition is rather obscured for me the value of some parameters used in the calculations (pixels sizes, 'slit width', etc). For instance, it is not clear for me why exposures times for a point souce and an 'extended' source of 0.03'' are different by a factor 2 (for K with R=100, S/N=10, ST=G5V, M=22.74). Therefore, it may be dangerous to used it in a blind way especially for the faintest galaxies. The approach that I followed was to reduce the problem of the determination of exposure times for 'extended objects' to 'point sources', and then to use ETC (without any kind of hypothesis about the source, but still with uncertainties related to the parameters mentioned above). For that we need to characterize the extended source in terms of its surface brightness and to consider the size of the sampling element. For instance, if we have an object with 22mag/"^2 which is observed with a sampling element of 0.03"x0.03", it is assumed that this will require approximately the same exposure time as a point source of 29.6 mag (22-2.5log(0.03x0.03)). Here it is assumed that the energy collected by a sampling element is distributed over the same number of detector pixels as in the case of a point source observation. This is probably no true, but the needed correction can be approximately determined if we learn more about ETC. In any case, I do not expect a diffence larger than an order of magnitude in the computed exposures times due to this effect. >From the observational point of view we need to determine the surface brightness. I think a good reference value can be the 'mean surface brightness interior to a half-light radius'. This mean SB can be directly derived from measured magnitudes and half-light radii. However, the half-light radius determination may be bias for several effects. The most important one is due to the seeing (for the Keck observations by Bershady) and to the PSF in the case of the HST. This tend to increase the h-l radius making fainter the SB. This will be relevant for the smallest sources (at the faintest edge), but for well resolved sources I think this should not be a problem. Summaryzing, these are the specific things which I think should be fixed: 1) HDF: Determination of the counts density, magnitudes and h-l radii. The data by Thompson et al, seem good for that, but I have not analyzed them in detail. 2) ETC: We need to know more details about how this program works. I have tried to contact Massimo Stiaveli a few days ago but without success. Does he belong to ESA? Should be possible to ask him to participate in this more formally ? 3) SB determination: The most worrying issue is the effect of the PSF on the intrinsic SB for the faintest galaxies. Perhaps I can do some simulations convolving the HST-PSF with galaxies profiles (r-1/4), but I do not know how realistic this may be. Please, let me know your doubts/comments/suggestions on this before I go further. All the best , Santiago Subject: Re: ETC Date: Thu, 18 Feb 1999 14:36:57 GMT From: Santiago Arribas To: rfosbury@eso.org CC: pjakobse@estsa2.estec.esa.nl, lefevre@astrsp-mrs.fr, ghasinger@aip.de Dear Bob, I have just received this mail from Massimo Stiavelli, which seems relevant to understand the limitations of ETC, as well as his availability for participating with us in this part of the study. Regards, Santiago ----- Begin Included Message ----- >From mstiavel@stsci.edu Thu Feb 18 14:18:36 1999 Date: Thu, 18 Feb 1999 09:18:01 -0500 (EST) From: Massimo Stiavelli To: Santiago Arribas Subject: Re: ETC MIME-Version: 1.0 Dear Santiago, the tool you are referring about was written years ago as a zeroth order estimator and is now ENTIRELY INADEQUATE for any serious work on NGST. I do not know why it produces the results you are describing since I do not remember what I did. For point sources the assumption is that 69 per cent of the flux ends up in a box of either 4 pixels in size or one PSF FWHM in size, whichever is larger. The PSF was Nyquist sampled at 2 microns. There has been talk about replacing it but so far a replacement is not available. I have new tools that I am working on but these are not public since I am developing them as part of a NASA funded instrument study concept. Cheers, Massimo ----- End Included Message -----