Priya et al. (2026) Optimal water depth bias correction in LiDAR data using deviation-level analysis and advanced deep hybrid models

Identification

• Journal: Remote Sensing Applications Society and Environment
• Year: 2026
• Date: 2026-01-01
• Authors: SVishnu Priya, N. Neelima, Vivek Venugopal, M. Eshwaraiah Raghunandan
• DOI: 10.1016/j.rsase.2026.101930

Research Groups

• Department of Electronics and Communication Engineering, Amrita School of Engineering, Amrita Vishwa Vidyapeetham, Bengaluru, India
• Monash Climate-Resilient Infrastructure Research Hub (M-CRInfra), School of Engineering, Monash University, Selangor Darul Ehsan, Bandar Sunway, 47500, Malaysia

Short Summary

This paper proposes an optimal model for water depth bias correction in bathymetric LiDAR point cloud data, specifically addressing the variability in bias during depth correction modeling. The developed Adaptive Weighted Bayesian – Linearly Scaled Hyperbolic – Long Short Term Memory (AWB-LSH-LSTM) model demonstrates superior performance in accuracy, computational efficiency, and data compression for large-scale LiDAR depth processing.

Objective

• To propose an optimal model for water depth bias correction from bathymetric LiDAR point cloud data, focusing on addressing the variability in water depth bias during depth correction modeling.

Study Configuration

• Spatial Scale: Large-scale processing of LiDAR depth data.
• Temporal Scale: Time-series data analysis using a sliding window technique.

Methodology and Data

• Models used: Hadoop map framework, Calinski–Harabaz Density-Based Spatial Clustering of Applications with Noise (CH-DBSCAN), Linear Interpolation (LI), Exponential Smoothing, Min-Max normalization, Independent Component Analysis (ICA), T-distributed Radial Basis Covariance Embedding (TRBCE), Adaptive Weighted Bayesian – Linearly Scaled Hyperbolic – Long Short Term Memory (AWB-LSH-LSTM).
• Data sources: LiDAR water depth dataset from US Geological Survey.

Main Results

• The proposed AWB-LSH-LSTM model significantly reduces correction time by 6.67% to 23.35%.
• It demonstrates substantial error reduction across all metrics, including MAPE, RAE, MAE, MSE, and RMSE.
• The model improved NSE scores ranging from 3% to 38% and R2 scores ranging from 3% to 47%.
• T-distributed Radial Basis Covariance Embedding (TRBCE) achieved a compression ratio of 10.12, outperforming T-SNE by 18.1% and PCA by 398.4%.
• The optimization strategy recorded the lowest fitness across all iterations, leading to shortened runtime compared with state-of-the-art methods.
• Overall, the method shows superior performance in clustering, data compression, bias correction accuracy, and computational efficiency for large-scale LiDAR depth processing.

Contributions

• Proposes an optimal model for water depth bias correction that specifically addresses the variability in water depth bias, a previously under-focused aspect in existing research.
• Introduces an advanced deep hybrid model (AWB-LSH-LSTM) that significantly enhances computational efficiency and accuracy in bathymetric LiDAR depth bias correction.
• Demonstrates superior performance in data compression using TRBCE and overall processing efficiency for large-scale bathymetric LiDAR data.

Funding

• Not specified in the provided text.

Citation

@article{Priya2026Optimal,
  author = {Priya, SVishnu and Neelima, N. and Venugopal, Vivek and Raghunandan, M. Eshwaraiah},
  title = {Optimal water depth bias correction in LiDAR data using deviation-level analysis and advanced deep hybrid models},
  journal = {Remote Sensing Applications Society and Environment},
  year = {2026},
  doi = {10.1016/j.rsase.2026.101930},
  url = {https://doi.org/10.1016/j.rsase.2026.101930}
}