Saturday, December 5, 2009

Surfing and Simulating Surf

First day of surfing this summer - hurray !! It was fun, but we lasted only a couple of hours and I bet I will be sore tomorrow.

Sea-clutter and simulating it has been a long fascination in the SAR community. People have flown planes around in circles recording waves at various incidence angles and sea states. Made anlytical expressions of the dynamic ocean surface including larger waves, capillary waves, breakers and even foam. Put these through electromagnetic simulators like HFSS and tried to relate the observations to the simulations. Theorists keep on borrowing from graphics and making ever so detailed waves.

On the other hands the experimenters and statisticians haven't been sitting on their hands. There are myriads of distributions trying to model clutter amplitude statistics like the classic k-distribution and the kk-distribution. These try to fit the observed values to a statistical pattern and predict the extrema in the sea-clutter which influences the probability of detection of a target in this environment.

Papers have been plenty, but as one of my colleagues (thanks for the long talks about your lack of belief in statistics Marion) points out. The target has to be detected at an instance, and statistics is not very good at working with small samples. All the work so far cannot predict whether we will see a crucial vessel in a single SAR image. The solution is to give the statistics a chance - build lots of sensors and observe repeatedly.

No comments: