Continuum removal in Hα extragalactic measurements

We point out an important source of error in measurements of extragalactic Ha emission and suggest ways to reduce it. The H alpha line, used for estimating star formation rates, is commonly measured by imaging in a narrow band and a wide band, both of which include the line. The image analysis relies on the accurate removal of the underlying continuum. We discuss in detail the derivation of the emission- line's equivalent width and flux for extragalactic extended sources, and the required photometric calibrations. We describe commonly used continuum- subtraction procedures, and discuss the uncertainties that they introduce. Specifically, we analyse errors introduced by colour effects. We show that the errors in the measured H alpha equivalent width induced by colour effects can lead to underestimates as large as 40 per cent and overestimates as large as 10 per cent, depending on the underlying galaxy's stellar population and the continuum- subtraction procedure used. We also show that these errors may lead to biases in results of surveys, and to the underestimation of the cosmic star formation rate at low redshifts ( the low z points in the Madau plot). We suggest a method to significantly reduce these errors using a single colour measurement.