Ensemble based atmospheric phase screen estimation using least squares

It is a matter of fact that atmosphere causes a significant effect in interferometric measurements. A stack of acquisitions is needed to reduce this nuisance influence. Another approach is to use numerical weather predictions (NWP) [1]. However, it is not straight forward to choose parameters of a NWP-system to get the best possible result. Also, different input data can be used. Ensemble methods are well established to get a most reasonable forecast which is a mixture of differently initialized forecasts. Based on these weather predictions (NWPs), a collection of possible atmospheric phase screens (APSs) is computed and a most reasonable atmospheric phase screen (APS) is estimated. The idea behind this ensemble is, that the difference between every two pixels in an interferogram should match the difference of the corresponding pixels of a restricted linear combination of different estimated APS which results from different NWPs. This paper describes the technique in detail and provides an example demonstration using real data.