Classifying ecosystems with metaproperties from terrestrial laser scanner data.

Area of Science:

• Ecology
• Remote Sensing
• Geospatial Analysis

Background:

• Terrestrial Laser Scanner (TLS) data generates millions of lidar pulses per scan.
• Properties within these large lidar datasets can reflect ecosystem attributes.
• Existing TLS applications primarily focus on object reconstruction.

Purpose of the Study:

• Introduce metaproperty analysis for ecological classification using TLS data.
• Develop and validate the Metaproperty Classification Model.
• Demonstrate the model's capability in identifying ecosystem conditions.

Main Methods:

• Developed the Metaproperty Classification Model based on pulse-level and spatial metrics from TLS data.
• Applied the model to classify scans from rooms and forests (proof-of-concept).
• Tested the model on temperate vs. tropical forests and inland vs. coastal tropical rainforests.

Main Results:

• Achieved 100% accuracy in a proof-of-concept classification of rooms and forests.
• Successfully differentiated temperate and tropical forests with 97.09% accuracy (N=224).
• Classified inland and coastal tropical rainforests with 84.07% accuracy (N=270).

Conclusions:

• Metaproperty analysis effectively utilizes TLS data for ecological classification.
• The method can identify subtle ecosystem conditions, including disease and disturbance.
• This approach complements existing TLS applications and characterizes regional heterogeneity.