A means of improving target detection in clutter has been presented. Two results concerning clutter estimation have been illustrated by this work. First, the use of frequency diversity improves adaptive detection performance to a greater extent than has been reported for optimum processors under an equal transmit energy constraint. Second, MPGLR will approach the performance of the optimum processor in the polarization domain, which is not usually possible with adaptive single band processing under severely limited observation conditions. These results are driven by the requirement to estimate the clutter covariance, which is inhomogeneous in range. We conclude that multiband processing improves performance by increasing the quantity of iid data vectors available for clutter estimation.