Sensitivity analysis of a spatially-adaptive estimator for data fusion

We analyze the parametric sensitivity of a spatially-adaptive multiscale data fusion method. The fusion problem is formulated as a recursive estimation problem in scale and space using a set of 1D Kalman filters. The overall filter accommodates data acquired at different resolutions and missing data. The filter approaches optimal performance for data with spatially-varying statistics by adaptively updating the filter parameters using the innovation-correlation method. The contribution of this paper is the determination of the estimation error sensitivity to the process noise and measurement noise variances.