Ground and space based coronagraphs have been proposed to suppress the light of the star so a planet nearby can be imaged. But even when starlight has been suppressed by $10^{10}$, the residual starlight is as bright as the planet, and must be subtracted to $2*10^{-11}$ for a 5 sigma detection of the planet. For a ground based AO coronagraph, the problem is even more severe. Typically suppression of starlight to $10^{-5}$ of the star is possible and the residual speckle pattern must not have any “bumps” that mimic a planet at $10^{-7}-10^{-8}$. This paper describes a speckle calibration approach that measures the electric field of the light after it exits the coronagraph, in order to estimate the speckle pattern in the image plane. This technique makes use of the coherence of star light or rather the incoherence of starlight to planet light, and has very significant advantages compared to other techniques.

For a space based coronagraph, an alternative approach is to rotate the telescope /coronagraph and subtract two images. The calibration interferometer described here has the advantage that the temporal stability of the system can be relaxed by several orders of magnitude. For a ground based AO coronagraph system this approach has none of the serious limitations of the techniques based on the radial expansion of the speckle pattern with wavelength and enables ground based AO coronagraphs to approach the photon limit rather than the atmospheric limit. The calibration interferometer is being built for a NASA sounding rocket experiment by BU, JPL, MIT, and GSFC (PICTURE) with a 50cm telescope and a nulling coronagraph to be launched in 2007. It is also part of a design study for an extreme AO coronagraph for the Gemini Telescope, and a conceptual study of an extreme AO coronagraph for the TMT.