Bridge Parameter Inversion Using Generalized Compact Polarimetric SAR: A Feasibility Study With Intensity-Based Stripe Localization

Abstract

Bridge parameter inversion from synthetic aperture radar (SAR) data has traditionally relied on fully polarimetric (FP) observations, where $\mathrm {Re}(\langle S_{HH} S_{VV}^{*}\rangle)$ is used to localize double-bounce scattering stripes. This letter demonstrates, for the first time, the feasibility of bridge multiparameter inversion using SAR data in a generalized compact polarimetry (GCP) configuration. Since this FP discriminator is unavailable from GCP observations and no unified scattering-mechanism discriminator exists across arbitrary GCP transmit states, we propose an intensity-based strategy based on standalone local-peak localization in the total-power range profile. Four GCP modes synthesized from ALOS-2 PALSAR-2 data over the Golden Gate Bridge show stripe-position deviations within ±2 pixels, yielding a clearance-height error of 2.60% and deck-top-height errors of 10.71%–14.97%. A native RISAT-1 circular-transmit linear-receive (CTLR) acquisition further reproduces the three-stripe localization without algorithmic modification. Stokes-discriminator analysis confirms strong discrimination for CTLR and $\pi /4$ but not for standard DP modes, while a search over the ALOS-2 transmit-state space shows only 0.6-dB variation in weighted stripe signal-to-clutter ratio (SCR). The results support GCP bridge inversion for bridge–water geometries in which the single-, double-, and triple-bounce stripes remain sufficiently visible and spatially separable.