A 3D maser model has been used to perform an inverse problem on the light curves from three high-amplitude maser flares, selected on the basis of contemporaneous infra-red observations. Plots derived from the model recover the size of the maser cloud, and two parameters linked to saturation, from three observational properties of the light curve. Recovered sizes are consistent with independent interferometric measurements. Maser objects transition between weak and moderate saturation during a flare.

We simulate maser flares by varying either the pump rate or the background level of radiation in a 3D model of a maser cloud. We investigate the effect of different cloud shapes, saturation levels and viewpoints. Results are considered for clouds with both uniform and internally variable unsaturated inversion. Pumping and background variations are represented by several different driving functions, some of which are light curves drawn from observations. We summarise the pumping variability results in terms of three observable parameters, the maximum flux density achieved, a variability index and duty cycle. We demonstrate typical ranges of the flux density that may result from viewing an aspherical object from random viewpoints. The best object for a flare is a prolate cloud, viewed close to its long axis and driven from unsaturated conditions to at least modest saturation. Results for variation of the background level are qualitatively different from the variable pumping results in that they tend to produce short intervals of low flux density under conditions of moderate saturation and sufficient variability to be consistent with strong flaring. Variable background models typically have a significantly higher duty cycle than those with variable pumping.