If astronomers don’t have a spectrum for a star, they use another tool to find its temperature: a “color-color diagram.” If you plot the wavelengths of SDSS’s five filters on a thermal radiation curve, you get a graph like this:
To find one of the star’s colors from this diagram, subtract the magnitude of the star seen through one filter from the magnitude seen through another filter. The thermal source graphed above has a curve that peaks in the green. Therefore, its g-r color will be negative (remember, brighter objects have lower magnitudes). Its u-g color will be positive. What about its r-i and i-z colors?