Mon, 04/15/2024 - 14:04
I'm just finding my way around Transformation. I've done a series of observations on two standard fields Melotte 11 and SA29. I've got 19 stars measured. When I plot the Colour Transform chart (b-v against B-V) I get a nice straight line with a slope of 0.9432. But the Magnitude transforms have a huge amount scatter. I'm using B and V filters so the plots are B-b against B-V and V-v against B-V. You'd have a hard job seeing any kind of trend line at all. Any suggestions?
Steve
Yes, the plots for the magnitude coefficients look poor. And the regression coefficient confirms this: Where the regression coefficient for the color transform is 0.99 the magnitude transform is often 0.3
Some of the bad optics is the auto-scaling of plots. The magnitude graph is nominally slope 0 so the vertical scale is expanded.
They are what they are. You need at least one magnitude coefficient in any transform group.
George
As George says, the magnitude coefficient is very near zero, and so doesn't show a strong correlation with color. You can sometimes get a better curve if you use a wide range of colored stars. This is one reason why Arlo Landolt went out of his way to find stable, very red, stars for his standards papers. These tended to be heavily reddened main sequence stars, as most intrinsically red stars are variable. Examples of the reddened stars are in the SA110 and SA111 selected areas near the galactic plane. The AAVSO standard clusters have a pretty limited range of color, which makes it difficult to obtain good, low amplitude, coefficients.
Arne
Steve,
I think George was onto it.
I'm assuming you are using your own spread sheet and also using auto-scaling for the Y axis. If so, experiment with manual scaling of the Y axis. Leave the X axis alone. Keep experimenting with the Y axis until you are smiling. Then look at the slope of the linear trend line.
Of course, the way the plot is displayed does not change the slope, but some are so dismayed by the way the plot looks with auto-scaling that they give up.
Phil
Thanks everyone. I'll keep at it.
Phil's suggestion is the way to go.
I use spreadsheets for transforms, and scale the Y axes for the filter terms (e.g, Tv_bv) to display the same range of values as those for the colour terms (e.g., Tbv).
Since the plots for the colour terms typically show a substantial slope, the automatic scaling will usually provide a good visual impression of the slope and the residuals. Using the same range of units for the filter terms gives you similar good visual impressions.
Roy
Thanks Roy, I've started using manual scaling for the magnitude graphs and they look a lot better . I've also added an extra 20 stars from a different standard field and that's also improved the picture.
Steve