Tree Species Classification using Machine Learning and 3D Tomographic SAR -- a case study in Northern Europe

TL;DR

This research focuses on evaluating multiple tabular machine-learning models using the height information derived from the tomographic image intensities to classify eight distinct tree species, comparing the performance of various tabular machine-learning models and optimizing them using Bayesian optimization.

Abstract

Tree species classification plays an important role in nature conservation, forest inventories, forest management, and the protection of endangered species. Over the past four decades, remote sensing technologies have been extensively utilized for tree species classification, with Synthetic Aperture Radar (SAR) emerging as a key technique. In this study, we employed TomoSense, a 3D tomographic dataset, which utilizes a stack of single-look complex (SLC) images, a byproduct of SAR, captured at different incidence angles to generate a three-dimensional representation of the terrain. Our research focuses on evaluating multiple tabular machine-learning models using the height information derived from the tomographic image intensities to classify eight distinct tree species. The SLC data and tomographic imagery were analyzed across different polarimetric configurations and geosplit configurations. We investigated the impact of these variations on classification accuracy, comparing the performance of various tabular machine-learning models and optimizing them using Bayesian optimization. Additionally, we incorporated a proxy for actual tree height using point cloud data from Light Detection and Ranging (LiDAR) to provide height statistics associated with the model's predictions. This comparison offers insights into the reliability of tomographic data in predicting tree species classification based on height.

Figures (5)

• Figure 1: From left to right: the first image shows the single-look complex (SLC) image of the study area; the second depicts the intensity of the Tomographic SAR image; the third presents the cloud points from the Light Detection and Ranging (LiDAR) data; and the final image displays the tree species map for the study area.
• Figure 2: Geographic data splits used for training and testing: The left figure illustrates the swath split, while the right figure depicts the square split.
• Figure 3: Result for Classification and Prediction
• Figure 4: The figure is read from left to right. Initially, the tomographic image is converted into a tabular format with X and Y coordinates. The data is then allocated to its corresponding split, either swath or square. The data can be processed either by separating it by polarimetry channels or by combining them. Finally, this processed data is integrated with tree species classification information in a tabular format and submitted to AutoGluon for analysis.
• Figure 5: Each graph displays violin plots, with the kernel density estimation shown on both sides and the corresponding box plot in the center. The top left and bottom left corners depict the LiDAR heights for each actual tree class in the testing and training partitions, respectively. The top right and bottom right corners illustrate the LiDAR heights for each predicted tree class in the testing and training partitions, respectively.