Development of a stream DTM generation method using vegetation and morphology composite filters with SfM point clouds

Area of Science:

• Geosciences
• Remote Sensing
• Photogrammetry

Background:

• Traditional stream surveys are time-consuming, costly, and lack continuous data.
• Accurate Digital Terrain Models (DTMs) are crucial for understanding hydrological processes and managing riparian zones.
• Limitations of existing methods necessitate the development of advanced, efficient DTM generation techniques.

Purpose of the Study:

• To develop and evaluate an advanced method for generating high-quality DTMs of streams using Structure from Motion (SfM) data.
• To identify optimal vegetation and morphological filter combinations for processing SfM data in riparian environments.
• To assess the efficiency and accuracy of the proposed SfM-based methodology for DTM generation.

Main Methods:

• Structure from Motion (SfM) photogrammetry was employed using Phantom 4 multispectral drone imagery.
• A leveling survey was conducted for ground-truthing on four cross-sections of the Bokha stream.
• Various vegetation (NDVI, NDI) and morphological (ATIN, CSF) filters were applied and evaluated using Mean Absolute Error (MAE) and Root Mean Square Error (RMSE).

Main Results:

• The integration of NDVI and CSF filters demonstrated superior performance in vegetated areas.
• Single NDVI application yielded the lowest MAE in bare ground areas.
• The SfM method effectively removed waterfront vegetation, achieving an overall MAE of 0.299 m and RMSE of 0.375 m.

Conclusions:

• The proposed SfM-based methodology provides an efficient and cost-effective solution for generating accurate DTMs of streams and riparian zones.
• This advanced technique overcomes the limitations of traditional field surveys, offering continuous data and reduced survey time.
• The findings support the broader application of drone-based SfM for hydrological and geomorphological studies in vegetated aquatic environments.