Whenever I've used or calculated ground speed in issues related to aircraft performance or particularly calibration of pitot-static system pressure error corrections, the definition used has always been horizontal component of absolute speed (although I don't recall any document where that is explicitly stated).
G