What Are The Most Effective LiDAR Point Cloud Processing Techniques For Distinguishing Between Ancient Agricultural Terracing And Natural Terrain Features In Densely Vegetated Regions Of The Mediterranean, Specifically In Areas Where The Remains Of Hellenistic-era Settlements Are Known To Be Buried Beneath 2-3 Meters Of Sediment And Debris?

by ADMIN 343 views

To effectively distinguish ancient agricultural terracing from natural terrain features in densely vegetated Mediterranean regions using LiDAR, a comprehensive approach integrating multiple processing techniques is recommended:

  1. Data Preprocessing:

    • Noise Filtering: Apply statistical outlier removal and progressive morphological filters to clean the point cloud, enhancing clarity and reducing interference.

  2. Ground Classification:

    • Use algorithms like Cloth Simulation Filter (CSF) to separate ground points from vegetation, providing a clearer terrain model.

  3. Terrain Analysis:

    • Slope and Curvature Analysis: Examine the slope and curvature of the terrain to identify flat or gently sloping areas characteristic of terraces.
    • Digital Elevation Models (DEMs): Generate high-resolution DEMs and analyze topographical features using slope, aspect, and curvature rasters. Hillshade rendering can aid visualization.

  4. Feature Enhancement:

    • DEM of Differences: Subtract a smoothed DEM from the original to highlight subtle features, potentially revealing buried terraces.

  5. Point Cloud Segmentation:

    • Employ region growing or clustering based on slope and curvature to isolate potential terrace features.

  6. Intensity and Geometric Analysis:

    • Analyze laser intensity to detect variations under sediment. Use geometric analysis to identify uniform, aligned features typical of terraces, such as parallel lines and consistent spacing.

  7. Machine Learning and Classification:

    • Apply machine learning techniques, possibly using object-based image analysis (OBIA), to classify segments based on attributes. Consider integrating multispectral data for enhanced classification.

  8. Visualization and Integration:

    • Utilize 3D rendering and orthophotos for visual inspection. Combine LiDAR data with geophysical surveys (e.g., GPR) for detecting buried structures.

  9. Multidisciplinary Approach:

    • Integrate LiDAR with other data sources and techniques to improve accuracy, especially where sediment coverage is significant.

By combining these methods, the approach leverages the strengths of each technique to distinguish ancient terraces from natural features, even under challenging conditions.