Answers are grounded in this paper and cite their sources.
Accurate remote sensing monitoring of intertidal zones is the core foundation for dynamic ecological and geomorphological assessment. Traditional manual surveys lack spatio-temporal continuity, while satellite remote sensing data are constrained by long revisit cycles and coarse spatial resolution. With the advancement of remote sensing and imaging technologies, multispectral imaging from unmanned aerial vehicles (UAVs) has seen widespread application. However, existing UAV multispectral methods predominantly rely on single spectral features, failing to fully exploit inter-channel correlations and thereby reducing identification accuracy. To address these issues, this study proposes a novel adaptive radiometric calibration and cross-Bayesian optimized random forest (CBRF) scheme with UAV mltispectral data for high-precision identification of intertidal flats. First, a six-band multispectral camera is calibrated using a combination of an indoor monochromator and an outdoor UAV. Outdoor calibration is achieved through absolute calibration using a standard grey chart and relative calibration via downward-looking light. The spectral characteristics of sandstones, waterfronts, and forests were analyzed. Second, the CBRF algorithm is designed through dynamically searching and cross-validating to mitigate overfitting risks. Finally, this algorithm is applied to measure and recognize intertidal flats in Zhanjiang, the uncertainty is reduced by 3.5% to 5.2% with a recognition accuracy of 95.8% (kappa coefficient: 93.7%). This study provides an operational Earth-observation solution for dynamic coastal wetland inventory and geomorphologic studies in intertidal regions.
No takes yet. Share an insight, caveat, or question.
Guo et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: