A Novel Semisupervised Contrastive Regression Framework for Forest Inventory Mapping with Multisensor Satellite Data

TL;DR

This paper introduces a semisupervised contrastive regression framework that enhances forest variable mapping accuracy using multisensor satellite data, addressing data scarcity issues in deep learning applications for Earth Observation.

Contribution

It presents a novel semisupervised regression framework combining contrastive and pseudo regression losses for improved forest variable mapping.

Findings

Achieved 15.1% relative RMSE in forest height prediction

Outperformed vanilla UNet and traditional regression models

Demonstrated effectiveness on boreal forest satellite data

Abstract

Accurate mapping of forests is critical for forest management and carbon stocks monitoring. Deep learning is becoming more popular in Earth Observation (EO), however, the availability of reference data limits its potential in wide-area forest mapping. To overcome those limitations, here we introduce contrastive regression into EO based forest mapping and develop a novel semisupervised regression framework for wall-to-wall mapping of continuous forest variables. It combines supervised contrastive regression loss and semi-supervised Cross-Pseudo Regression loss. The framework is demonstrated over a boreal forest site using Copernicus Sentinel-1 and Sentinel-2 imagery for mapping forest tree height. Achieved prediction accuracies are strongly better compared to using vanilla UNet or traditional regression models, with relative RMSE of 15.1% on stand level. We expect that developed framework can be used for modeling other forest variables and EO datasets.