Multipath track association for over-the-horizon radar using a bootstrapped statistical ionospheric model

Abstract

Over-the-horizon (OTH) radar uses the refractive properties of high-frequency radiowave propagation through the ionosphere for wide-area surveillance at long ranges. Ionospheric propagation often gives rise to multiple raymodes between the OTH radar and a target which results in multiple slant tracks from a single target. Multipath and multiple target ambiguities are typically resolved by assuming that the down-range ionosphere is known precisely and then using ray tracing to determine the coordinate registration (CR) transformations from slant coordinates to target locations in ground coordinates. To achieve greater robustness to the uncertainty in down-range ionospheric conditions, this paper presents a maximum a posteriori (MAP) mode linking method for track association that employs statistical modeling of the down-range plasma frequency profile and corresponding multipath slant track measurements. To determine the statistical model parameters from quasi-vertical incidence (QVI) ionogram and wide-sweep backscatter ionogram (WSBI) measurements, the plasma frequency profile is approximated as a homogeneous random process over the region near the mid-point between the radar and the dwell illumination region. Using samples of a 3-D ionospheric model fitted to the and QVIs and WSBIs, statistical propagation model parameters are obtained by smoothed bootstrap resampling combined with Monte Carlo evaluation of a ray tracing propagation model. Simulation results indicate that the MAP mode linking method can achieve nearly a 3:1 improvement in ground coordinate accuracy over conventional mode linking methods with much higher probabilities of correct raymode identification and slant-track-to-target assignment. Real data results from roughly 90 minutes of OTH radar slant track data demonstrate that the MAP mode linking method can provide as much as a 4:1 improvement in ground coordinate accuracy over conventional methods.