As described above, if the total flux density in the MEM image is left unspecified then the value found may be seriously biased if the signal-to-noise ratio is low. There is no way around this at the moment, except to guess a value and then adjust it to get an image that looks ``reasonable''--for example, possessing a flat baseline. For bright objects, only an order-of-magnitude estimate is required to set the flux density scale. The estimated flux density is not then fitted, but is used only to set a reasonable default image. Guessing low by about an order of magnitude often works well.
1996 November 4