Doppler imaging from artificial data - Testing the temperature inversion from spectral-line profiles

We present extensive numerical tests of our temperature-based image reconst ruction code TEMPMAP. Two test cases that represent a rapidly-rotating low- inclination star and a moderately-rotating high-inclination star are used f or the forward problem. Tests are then made to recover the original input m ap and include data errors and input-parameter errors. The former include v ariations of photon noise, continuum displacement, continuum slopes, scatte red light in the spectrograph, and phase gaps with and without continuum in formation. The input-parameter tests include the confusion in the recovery of hot versus cool spots, uncertainties of atmospheric parameters such as r adial-tangential macroturbulence and gravity, and the influence of line str ength and line damping. In general, we find that the geometric information is less dependent on photon noise and on input errors than the temperature information. Our test inversions also show that, at the v sin i of these te st cases, no significant gain in image quality is achieved once S/N of 300: 1 is surpassed.