P REGUNTAS QUE SE FORMULAN

The photometric quality of the collected data is clearly of importance. For the detection of single transit events with a depth of one percent a high photometric precision of about 0.5% RMS is necessary. For three detections of the same depth in folded data this is reduced to about 1% RMS (see section 5.7.). The number of stars that provide these quality levels for the

Chapter 6: Results of the observations from the TLS

Figure 6.4.: Light curve RMS plotted vs. R magnitude for the 6 hours time series observations of Target field No. 2 from October 14/15, 2001. About 4,000 light curves have a photometric precision better than 1% for 240 sec exposures up to magnitude 14. The noise levels were calculated by equation 2.21, 2.22, 2.23 and 2.24. The light curves of the brighter stars are dominated by photon noise, but the majority of light curves (fainter stars) are dominated by background noise. Light curves of stars with an RMS well above the total noise can be affected by different additional noise sources: stellar variability, increased background noise due to crowding and pixel defects, passing satellites and aircrafts.

Figure 6.5.: The same as in Figure 6.4. but with an exposure time of 40 sec instead of 240 sec. 383 additional light curves with a photometric precision better than 1% were obtained (magnitude < 11). The precision of these light curves is photon noise limited. More light curves with RMS < 1% are obtained for fainter magnitudes, but light curves for these stars were obtained with higher precision in the time series assembled of 240 sec exposures.

photometry can be used as indicator for the photometric quality of the data.

How many high precision light curves are obtained is mainly dependent on the level of different noise sources. There are three major contributors to noise: Photon noise, background noise and scintillation noise. These contributions can be estimated according to equation 2.21, 2.22 and 2.23 of chapter 2. The total noise could be calculated as in equation 2.24. In Figure 6.4., 6.5. and 6.6. the RMS of light curves of an exemplary night for exposures of 240 sec, 40 sec and 15 sec is displayed. The different noise terms are overplotted.

Figure 6.6.: The same as in Figure 6.4.and 6.5. but with an exposure time of 15 sec. 38 additional light curves with a photometric precision better than 1% were obtained (magnitude < 9.2). For most observations this number is further reduced yielding to a low efficiency of the 15 sec observations. Therefore it was decided to cancel further 15 sec exposures to improve the sampling rate of the 40 sec and 240 sec time series.

A comparison of the theoretically obtainable total noise with the actually obtained noise for the light curves can be used to evaluate the photometric quality of the data. The light curves composed of longer exposures (240 sec and 40sec) are dominated by background noise (see Figure 6.4. & 6.5.); only the light curves of the brightest objects are dominated by photon (shot) noise. Scintillation noise is negligible low. Scintillation noise has only a mayor influence on the noise for light curves of bright stars in the 15 sec exposures (Figure 6.6.). Additionally the photon noise contributes in the same order as the scintillation noise. Background noise influences the light curves of the fainter stars (>10 mag) in the time series of the 15 sec exposures. Light curves of stars down to magnitude 14 with a photometric precision better than 1% were obtained during the darkest nights with 240 sec exposures (see Figure 6.4.). 1% precision was reached for time series of 40 sec exposures down to magnitude 12 (Figure 6.5.), for 15 sec exposures down to magnitude 11 (Figure 6.6.).

Only the reduction of the background noise will give a significantly higher number of high precision light curves of fainter stars. Analyses showed that the sky background noise for observations from TLS is linearly dependent from the sky signal, because readout noise and thermal noise are negligible low. Only the detected single hot pixels introduce additional noise, the noise of the general dark current is low enough to be not dominant for the total noise of the light curves.

The plots 6.4.- 6.6. represent observations of the star-rich target field No. 2 during the night of October 14/15, 2001. This night stands out from most of the nights when observations of

Chapter 6: Results of the observations from the TLS

star-rich target fields were carried out. The transparency of the atmosphere was high and stable, and the sky background was low yielding a high number of high precision light curves (about 4,000 light curves with precision < 1% for all three exposure times). The number of high precision light curves obtained in the time series of the shorter exposure times 15 sec and 40 sec which are not covered in the time series of 240 sec exposures with higher precision are limited. This number is of about one tenth of the number of light curves resulting from the time series of 240 sec exposures for good photometric conditions. For the common photometric conditions the number of additional light curves especially from exposures of 15 sec is further limited. In good nights only a few dozens additional light curves with a photometric precision just below 1% were obtained. Therefore it was decided to cancel the 15 sec exposures in early 2003 to improve the time sampling for the time series of 40 sec and 240 sec.

Figure 6.7. The number of stars that reach the photometric precision levels of 0.5%, 1.0% and 1.5% for the 240 sec observations of target field No.15. In the first half of the diagram observations during the late autumn and winter show a variable number of stars with distinct precision levels demonstrating the changeable conditions during this time (bad transparency, high background signal due to snow, …). The observations in the second half were made mainly during a 2-week period in September 2002 with more stable but not perfect weather conditions.

Less suitable photometric conditions for most of the observations reduced the number of high precision light curves. The numbers of light curves resulting from the 240 sec exposures with precisions better than 0.5%, 1.0% and 1.5% are plotted for comparison in the Figures 6.7. (target field No. 15), 6.8. (field No. 2) and 6.10.(field No. 8).

Some general trends can be detected. During the summer time the photometric precision decreases because of the higher sky background, during the winter the same problem can be observed when reflection from snow increases the background level significantly. Only in stable phases of the weather is a constant number of high-precision light curves obtained as seen for observations of field No. 15 and field No. 2.

Obtaining constant transparency is another major issue to reach high photometric precision as it can be seen for the third night of observations of target field F2 (October 14/15, 2001) in.

Figure 6.8. The same as Figure 6.7. but for field No. 2. The number of stars with the same precision is relatively constant during the winter season when this target field was observed. Only night No.3 (October 14/15, 2001) shows an exceptionally high photometric quality. Βoth transparency and seeing were very stable during that night, the background signal was low.

Figure 6.9. The same as Figure 6.8. but for 40 sec exposures. 877 light curves had a photometric precision better than 1% on average. About 530 of these light curves were covered by light curves of 240 sec exposures with higher precision. Thus 350 additional light curves (precision < 1%) were recovered from 40 sec exposures for the transit search.

Chapter 6: Results of the observations from the TLS

Figure 6.10. The same as for Figure 6.7. but for field No. 8. Note the trend of a decreasing photometric precision in both halves of the diagram corresponding to the two observed seasons 2002 and 2003. The number of stars that reach a photometric precision level is mainly influenced by the increasing sky background signal. Night No. 25 featured an exceptionally high background signal due to a close and bright moon three days before full moon.

The influence of the photon and scintillation noise increases with reduced exposure times. The number of high precision light curves obtained with 40 sec exposures is influenced mainly by background noise (see Figure 6.9.) and varies in a similar way as for light curves of 240 sec exposures (compare Figures 6.8. and 6.9.). Photon noise is of the same order for stars with light curves that are not covered with higher precision by 240 sec exposures. Thus the number of additional high precision light curves obtained with 40 sec exposures is nearly constant.

Going to shorter exposure times (15 sec, Figure 6.6.) increases the noise especially for the brighter unsaturated stars that are not covered by longer exposures (40 sec, 240 sec.). This additional noise further reduces the number of light curves with high precision in the observed target fields with a limited number of stars in the magnitude range from 8-9 that is not covered by longer exposures. Additionally it is known that the objects in this magnitude range are mostly giants or large main sequence stars. The low efficiency of the 15 sec observations for transit search was the main reason for the decision to cancel further 15 sec exposures to improve the sampling rate of the 40 sec and 240 sec time series.

In Table 6.1. the average number of light curves with distinct precision levels per target field is given. The highest number of high precision light curves is reached for target field No. 2 despite the fact that the total number of stars that is higher in target field No.15. A higher sky background in the summer and unstable weather conditions in the autumn (changing transparency) are the reason for this. Field No. 2 was mainly observed under such stable conditions in the winter season when cold but dry air was dominating. But high humidity during the winter season made it impossible to carry out observations more frequently.

Another important aspect for the number of high precision light curves is the crowding situation of the target fields. Even though target field No.8 has about 10 times (V<14) less target stars than field No.15 and seven times less stars than field No.2 a relatively higher number of high precision light curves was nevertheless reached: only 4.5 times less high- precision light curves than field No. 15 and 6 times less for field No. 2 (1% precision). Higher

Table 6.1.: Average number of stars with distinct light curve precision levels listed for the individual target fields. The number of additional light curves obtained by 40 sec exposures and not covered by 240 sec exposures is given in parenthesis.

Target field No.15 Target field No.2 Target field No. 8 RMS < 0.5% 427 ± 243 (78 ± 40) (105 632 ± 247 ± 67) (37 138 ± 12) ± 76 RMS < 1.0% 1849 ± 703 (261 ± 69) 2468 (337 ± 492 ± 36) 405 (68 ± 113 ± 9) RMS < 1.5% 3285 ± 831 (366 ± 49) 4012 (372 ± 640 ± 19) 647 (75 ± 130 ± 5) RMS < 2.0% 4660 ± 884 5328 ± 798 853 ± 156 RMS < 2.5% 5935 ± 1002 6540 ± 906 1023 ± 180

background levels can partly explain the large discrepancy for field 15. For field 2 with nearly the same background levels as field 8 additional noise caused by crowding is responsible for the observed discrepancy. Crowding plays an important role decreasing the photometric quality and has been analyzed more accurately in chapter 7.

Nevertheless, the obtained number of high-precision light curves with RMS < 1.0% and better will allow the detection of Jupiter-sized planets in short-period orbits according to the determined detection limits for one detection (section 5.7.1.) or three detections (see section 5.7.2.) in folded data.

6.2. Observations of transits of the exoplanet orbiting the star HD