Asteroseismic radii of dwarfs: new accuracy constraints from Gaia DR2 parallaxes

Precise stellar masses and radii can be determined using asteroseismology, but their accuracy must be tested against independent estimates. Using radii derived from Gaia DR2 parallaxes, we test the accuracy of asteroseismic radii for a sample of 93 dwarfs based on both individual frequency fitting and the seismic scaling relations. Radii from frequency fitting are about 1 per cent smaller than Gaia radii on average; however, this difference may be explained by a negative bias of 30 mu as in the Gaia parallaxes. This indicates that the radii derived from frequency fitting are accurate to within 1 percent. The scaling relations are found to overestimate radii by more than 5 percent, compared to the Gaia radii, at the highest temperatures. We demonstrate that this offset is reduced to 3 per cent after applying corrections based on model frequencies to the scaling relation for Delta nu, but only when the model frequencies are corrected for the surface effect. With corrections to Delta nu, the scaling relation gives radii accurate to about 2-3 per cent for dwarfs in the temperature range 5400-6700 K. The remaining offset at the highest temperatures may indicate the need for a correction to the scaling relation for nu(max).