Predicting trace organic compound breakthrough in granular activated carbon using fluorescence and UV absorbance as surrogates