The transit of a planet across HD 209458 lasts for about 3 hours, dimming the star by 1.5%. In this lab we are interested in whether or not we could be able to detect this change. Error calculations were done to find what the minimum change that we could detect is. We were only interested in the change in magnitude, and not with an absolute scale, so we compared the brightness of the star to a bright object of constant magnitude. This method of finding the ratio had the added benefit of cancelling out errors due to changes with time in the CCD, telescope efficiency, or atmospheric transmission. HD 210074 is one such object that's brightness is very constant; this is known from previous work (Henry et al. 2000).The Leuschner Telescope was used, with a 512x512 CCD, to collect this data. Data was taken in sessions, with a group of students working for about an hour; it was shared after it was analyzed.
The ratio in brightness of HD 209458 to HD 210074 was measured over a period of one hour. We took 18 images of HD 209458 "bookended" by images of HD 210074, for 19 images of the reference star in all. The reported ratio of HD 209458 to HD 210074 is actually the ratio of the DN of HD 209458 to the average DN of HD 210074 for the DN values that bookend the HD 209458 image. In this way changes in the DN of the reference star due to changes in the CCD, telescope efficiency, or atmospheric transmission. We calculated the flux from the star using a program that summed the DN within a given radius.
This table is of data that was taken by our group, Emily Deb and I, on Oct 19, 2002. Included are the errors associated with the DN counts and the calculated ratio.
Times of star exposures (sec) Starting at UT 5:23:00 | DN HD 209458 | Poisson Error in DN | DN HD 210074 | Poisson Error in DN | Ratio DN Star to Ref. Star | Propagated Uncertainty (dr/R) | Propagated Uncertainty (dr) |
---|---|---|---|---|---|---|---|
48 | 64421.2 | 253.813 | 328151. | 572.856 | 0.196510 | 0.00412893 | 0.000811375 |
272 | 63189.5 | 251.375 | 327490. | 572.267 | 0.190968 | 0.00416371 | 0.000795137 |
416 | 64317.8 | 253.610 | 334290. | 578.178 | 0.194330 | 0.00413019 | 0.000802621 |
629 | 61560.0 | 248.113 | 327654. | 572.411 | 0.186978 | 0.00421462 | 0.000788041 |
768 | 60635.3 | 246.242 | 330819. | 575.168 | 0.187520 | 0.00424716 | 0.000796427 |
1199 | 63190.7 | 251.378 | 315889. | 562.040 | 0.196290 | 0.00416873 | 0.000818280 |
1321 | 63764.9 | 252.517 | 327961. | 572.679 | 0.194252 | 0.00414799 | 0.000805757 |
1507 | 62276.1 | 249.552 | 328555. | 573.197 | 0.191189 | 0.00419435 | 0.000801912 |
1660 | 63871.3 | 252.728 | 322907. | 568.249 | 0.195778 | 0.00414597 | 0.000811692 |
1815 | 62195.5 | 249.310 | 329579. | 574.089 | 0.188433 | 0.00419577 | 0.000790623 |
2128 | 61320.0 | 247.629 | 330131. | 574.570 | 0.187613 | 0.00422347 | 0.000792377 |
2291 | 62409.0 | 249.818 | 323570. | 568.820 | 0.192232 | 0.00419088 | 0.000805622 |
2449 | 63162.9 | 251.341 | 325752. | 570.747 | 0.191738 | 0.00416501 | 0.000798590 |
2594 | 59091.1 | 243.043 | 333206. | 577.229 | 0.180020 | 0.00429567 | 0.000773308 |
3060 | 64607.0 | 254.179 | 323068. | 568.391 | 0.198986 | 0.00412531 | 0.000820879 |
3323 | 62383.9 | 249.768 | 326276. | 571.221 | 0.195948 | 0.00419527 | 0.000822056 |
3505 | 60765.1 | 246.506 | 310445. | 557.176 | 0.194704 | 0.00424958 | 0.000827411 |
3640 | 63147.6 | 251.292 | 313734. | 560.120 | 0.199573 | 0.00417326 | 0.000832873 |
- | - | - | 319092. | 564.882 | - | - | - |
Poisson Statistics: | Ratio | Star | Reference Star |
---|---|---|---|
mean | 0.192393 | 62572.7 | 325188. |
standard deviation | 0.00492553 | 1478.86 | 6445.30 |
standard deviation of the mean | 0.00116096 | 348.571 | 1478.65 |
Mean Ratio | Propagated fractional Error w/stdom | Propagated Error w/stdom | Propagated Fractional Error w/Poisson | Propagated Error w/Poisson |
---|---|---|---|---|
0.192420 | 0.00719083 | 0.00138366 | 0.00287565 | 0.00055333 |
Statistics for all 118 Ratios | Statistics for Ratios of data taken on October 26, 2002 | |
---|---|---|
Median Ratio | 0.195890 | 0.196604 |
Mean Ratio | 0.193735 | 0.197376 |
Standard Deviation | 0.0157511 | 0.00578906 |
Std. Dev./Mean | 0.0804079 | 0.0293301 |
Std. Dev of Mean | 0.00145001 | 0.000532927 |
The weather was horribly cloudy on the evenings of the transit; so nature decided for us: we couldn't detect the transit. The question of whether we would have been able to can be answered by analyzing the errors in the collected data. The propagated errors were much smaller than the error found by doing Poisson calculations on our final data. Error was introduced by changing weather conditions, and by the way we analyzed the data with our circle programs. Also, the reference star was a certain distance, a certain sky angle, away from HD 209458, so atmospheric conditions in one area of the sky would not necessarily affect both stars. This would change the ratios. Our group data has a smaller error than the class data as a whole. This is probably because of the varying weather conditions throughout the week.
Calculating the error, and the subsequently the minimum size of the planet detectable, raised a lot of debate in our class. Everyone used the equation:
Error/R = (radius(pl)/radius(star))^2
We disagreed on what error should be defined as. Our group initially used the standard deviation. Using this, and taking HD 209458 to be a star like our sun, we found that we would be able to detect a planet of radius 1.1005x10^8. This would make the planet just barely too small for us to see. An argument was made for using the standard deviation of the mean for the Error term. Doing the calculation over, the smallest detectable radius of the planet is .60 Jupiter radius', which means that we would be able to detect the planet.
Below are several simulations that create an array of points with a dispersion defined by the error in the ratio. The simulations are set up so that they show what data gathered during a transit would look like. The ratios are normalized. During transit, the ratio drops by 1.5%. The dotted line shows the true ratio value.
The second plot is similar to our group data plot, while the third gives a much larger distribution.
I hoped to find an answer to whether the standard deviation, or the standard deviation of the mean should be used to find the minimum radius. I drew how the standard deviation and the standard deviation of the mean would change with more data: the mean value is known more and more precisely, but the standard deviation approaches a finite value. Using the standard deviation of the mean is correct.
Maintained by Maureen Teyssier.