Análisis Semántico Contextual

Galaxy B R I

Leo A 2.027 1.177 2.238 GR 8 2.588 3.206 2.477

Table 3.2: Zeropoints obtained for the calibration using the zeropoints and color-terms listed in Equation. (3.10). The typical standard deviation of the fit is in the order of 0.04 mag.

3.4

Comparison of results

The main incentive to obtain the calibration at Calar Alto was, besides the need of a calibration for EGB 0427+63, the prospect of an independent comparison to the HST calibration. To achieve this we positionally matched sources from the calibrated WST DAOPHOT catalog and objects from the Calar Alto reference image and plotted the magnitudes against each other. The results are shown in Fig.3.4.

16 18 20 22 24 16 18 20 22 24 RCA [mag] RWST [mag] 16 18 20 22 24 16 18 20 22 24 RCA [mag] RWST [mag]

Figure 3.4: Comparison of the calibration in theR-Band between the HST calibrated Wendelstein data and the standard star calibrated Calar Alto observations using Ta- ble3.2. The left panel shows the results for Leo A, the right one for GR 8. The black points are the matched objects, the red ones those remaining after the sigma-clipping. The offset of the best fit for Leo A is 0.018 mag with a standard deviation of 0.002 mag and 0.062 with 0.003 mag standard deviation for GR 8. These results are in ex- tremely good agreement, considering that two completely different datasets calibrated using independent calibration methods.

For theBandI-band the situation looks different. A comparison of the zeropoints for the Calar Alto reference image in Table 3.1 and Table 3.2 shows differences in the order of 0.3 mag. To solve this discrepancy several possible reasons have to be evaluated.

First of all, a possible source for problems could be an unsound selection of stars for the transfer of the calibration onto the reference image. The use of red stars for the

B-band calibration or blue stars for the I-band calibration presents a possible source for a deviation of the final calibration zeropoint.

In case of theB-band, the stars used to transfer the calibration from the HST data show suitable blue colors in the range of (B−I) of −1 to 1. The stars employed as secondary standards for the calibration of the Calar Alto B-band obtained on the NGC2264 field however have relatively red colors of around 1 to 2 in(B−I). Unfor- tunately this in inevitable, as already all suitable stars (bright, isolated) are used.

TheI-band data calibration suffers from a similar problem, with the stars used for the HST calibration having suitable colors in the B−I range from 2 to 4. The stars available for the Calar Alto calibration have colors in between 1 and 2 in(B−I). This color range could still be sufficient, but in this case it is aggravated by another prob- lem. The observation were obtained using a JohnsonIfilter with a transmission curve significantly different from the Cousins filter used for the standard star observation by

Flaccomioet al.(1999). The Johnson filter transmits flux beyond the red cutoff of the Cousins filter.

Given these results of the comparison and of different empirical tests like a compar- ison of theI-band magnitude of the tip of the red giant branch (see Sec.6), we decided to use the HST based calibration in theBandI-bands.

Chapter 4

Catalog of variable sources

The reduced difference images produced by the reduction pipeline give the informa- tion about the relative variability of each pixel in the observed frame. From this vast amount of data candidates for periodic variability have to be selected. For these a period solution is derived at the same time further selecting sources with periodic vari- ability. The remaining objects are assembled into a catalog of variable sources and form the basis for further scientific studies.

4.1

Detection of the variable sources

To select candidates for variability in the difference images it was assumed that the difference magnitude of a regularly variable star would exceed its respective error on a sizable number of epochs. Capitalizing on this alone, i.e. just searching for deviant pixels, would still produce a far too large number of candidate sources, as all the pixels lying within the PSF of a variable star would show up as individual sources. To circumvent this problem a means to reassemble the stars is needed. To attain this an image was assembled with each pixel holding the number of epochs for which the difference flux of this specific pixel is larger than the corresponding error. This image could then be searched for local maxima over a given threshold to select candidates for variability. The exact number of epochs applied as threshold for each dwarf galaxy are given with the catalog at the end of this chapter. This approach can still lead to multiple detections, yet a lot fewer. To deal with these, the resulting variability catalog is position-matched against the object catalog obtained on the deep reference image. From all the variable sources matching against the same stellar source only the one with the most significant period signal is kept.