Here is a blue star.
The spectrum is high in the UV and dives down toward the IF.
But the colours in Skyserver show
u g r i z
17.16 15.95 16.13 16.30 16.40
I know bigger numbers mean dimmer light.
I know blue stars produce all colours of light, though mostly blue.
But I would have expected something like
17 18 19 20 21
Why don't the filter colours correlate with the spectrum?
587741601489420350
There's one step missing. Magnitude zero in various filters does not correspond to the same energy received (for historical reasons, often tracing back to using the A0 star Vega as the primary defining standard). The energy zero points (ZP) for the SDSS filters are approximately:
u 3.67e-9
g 5.11e-9
r 2.40e-9
i 1.28e-9
z 0.783e-9
all in units of ergs/cm^2 s Angstrom. So in each filter, the average energy received across the filter band will be F = ZP x 10^(-0.4 * magnitude), neglecting details like the precise filter transmission and the weighting by photons rather than by energy. This star has colors close to an A0 star which is why its griz magitudes are all so similar. Other magnitude systems have different zero points - constant flux per unit wavelength, constant flux per unit frequency, and so on.
Yes, as NGC3314 says, there's various magnitude systems. I'm not sure it works out quite as he said though - we'll maybe have to bash this out a bit. There's something here I'm not understanding.
The Vega system is most traditional and is the one referred to by him - the one where an A0 star basically has the same magnitude in each band.
The AB system is used by SDSS (there's actually tiny corrections but it's near enough AB for most purposes). This gives you the same magnitude if your flux measured in ergs/(cm^2 s Hz) is a constant. It's now quite a common system to work with.
The ST system comes from Hubble stuff I believe (ST for space telescope) and that works in ergs/(cm^2 s Angstrom).
There's also the CD system which works in photons/(m^2 nm) - this is what X-ray telescopes use too I believe, but we can ignore that here
Converting between the two has been the bane of much of my research at one time or another
They're all defined to be equal at 548nm though.
Now if you look at the spectrum plot you'll notice the y-axis is in the ergs/(cm^2 s Angstrom) system. That's the ST magnitudes, which aren't the same as the SDSS ones. What should happen though is that you divide through by something proportional to wavelength^2 to convert from ST to AB, and that shouldn't really throw things off much. So what I suspect is happening is that the u band is doing something funny - it's actually at a shorter band than is shown in the spectrum, so are there lots of big absorption features (like those we can see some of at the blue end already) that are making the u band result much fainter than we'd expect?
Another way of looking at it is that something with constant SDSS magnitudes has a constant flux in ergs/(cm^2 s Hz) which means that the flux in ergs/(cm^2 s Angstrom) must be proportional to 1/wavelength^2. The curve doesn't go quite like this but gets fainter faster, so the bluer magnitudes should be a bit lower than the red ones - and that's what you get, except for the anomalous u-band.
So erm... after that rather rambly post, back to Bill to figure out the anomaly
