Ocean–Land Interface Determination from Mixed Waveform of Airborne Oceanic LiDAR

Laser waveforms in airborne oceanic LiDAR (AOL) data are classified as either ocean or land waveforms. However, ocean and land may simultaneously exist along the ocean–land interface (OLI) in the large footprint of the AOL, producing mixed ocean–land waveforms. Conversely, if we can identify mixed ocean–land waveforms, the position of the OLI can be determined. This study aims to identify mixed infrared (IR) ocean–land waveforms and further proposes a novel method for determining the OLI using the identified mixed IR ocean–land waveforms. First, a novel fuzzy convolutional neural network is proposed to classify IR waveforms and output a predicted probability vector indicating the likelihood of the waveform being classified as either ocean or land waveforms. Second, this predicted probability vector is used to identify the mixed IR ocean–land waveforms. Finally, the position of the OLI is determined by using mixed IR ocean–land waveforms and the corresponding laser point clouds. The proposed method is applied to a raw AOL dataset collected via the Optech coastal zone mapping and imaging LiDAR system. Compared with the other traditional AOL-based OLI determination methods, the proposed mixed waveform method reduces the standard deviation of distance biases by 27.59% and improves the structural similarity index by 0.017. The low standard deviation and high structural similarity index indicate the effectiveness and correctness of the mixed waveform method for OLI determination via AOL.