Identifying Regenerated Saplings by Stratifying Forest Overstory Using Airborne LiDAR Data

Area of Science:

• Forest Ecology
• Remote Sensing
• Geospatial Analysis

Background:

• Understory sapling identification is vital for understanding forest regeneration and informing sustainable forest management.
• Traditional 2D spectral remote sensing data struggles to accurately detect understory saplings due to canopy cover.
• 3D structural information offers a potential solution for overcoming the limitations of 2D spectral data in forest understory analysis.

Purpose of the Study:

• To develop and validate an automatic method for detecting regenerated understory saplings using 3D structural information from aerial laser scanning (ALS) data.
• To assess the accuracy of ALS-derived measurements of sapling height and crown width compared to field and terrestrial laser scanning (TLS) data.
• To provide a quantitative approach for understory sapling assessment to enhance forest ecosystem resilience.

Main Methods:

• Individual tree crown delineation using an improved spectral clustering algorithm to remove overstory canopy and trunk points.
• Segmentation of individual understory saplings employing an adaptive-mean-shift-based clustering algorithm.
• Validation of the method in a North China experimental forest farm, comparing ALS data with field and TLS measurements.

Main Results:

• Detection rates for understory saplings ranged from 94.41% to 152.78%, with matching rates increasing from 62.59% to 95.65% as canopy closure decreased.
• ALS-derived sapling heights showed strong correlation with field measurements (R² = 0.71) and TLS measurements (R² = 0.78).
• ALS-based sapling crown width measurements were comparable to TLS-based measurements (R² = 0.64).

Conclusions:

• The proposed automatic method effectively utilizes 3D ALS data for accurate detection and quantification of understory saplings.
• This approach provides a reliable solution for assessing understory vegetation, overcoming limitations of traditional 2D remote sensing.
• The findings support improved forest management strategies by enabling better utilization of light resources and enhancing ecosystem resilience.