Nonlinear Kalman filtering of long-baseline, short-baseline, GPS, and depth measurements
Abstract
In a recent Moving Ship Tomography experiment
we deployed a vertical array of receiving hydrophones
300 times around a 1000 km diameter circle
south of Bermuda; six acoustic sources within the circle
transmitted every 3 hours.
We plan to use
the travel times measured along the source/receiver paths
to reconstruct the three dimensional sound speed field in the
interior of the circle.
Accurate location of the hydrophone receivers is crucial to the
execution of this plan.
A long baseline tracking system consisting of
floating buoys and the ship, acoustically measured the range to
each hydrophone
(~ 1km in depth and ~ 1km in horizontal range).
The range to orbiting satellites was measured with GPS.
A short base line acoustic system measured
range and direction from the ship to a beacon on the cable,
and to each buoy.
Depths, at various points near the hydrophones
were measured using three different pressure sensors.
We describe a Kalman filter that combines these primary measurements
and some supporting measurements,
to form an estimate of the buoy and hydrophone locations.
We use an iterated correction step
to adjust for the nonlinearities and an outlier detection scheme
to make the filter robust.
