A Novel Graph-Regulated Disentangling Mamba Model with Sparse Tokens for Enhanced Tree Species Classification from MODIS Time Series

2026-05-06

AI Summary

Tree species classification faces significant challenges due to subtle spectral differences, strong coupling among spatial, spectral, and temporal information, and difficulties in modeling large-scale contextual dependencies. To address these issues, this work proposes the Graph-Regularized Decoupled Sparse Mamba (GDS-Mamba) model, which leverages graph structures to capture long-range spatial dependencies, decouples spatial-spectral-temporal features to mitigate correlation decay inherent in standard Mamba architectures, and incorporates an adaptive sparse token mechanism to enhance both computational efficiency and fine-grained detail recognition. Evaluated on MOD13Q1 time-series data from Alberta and Saskatchewan, Canada, the proposed model achieves overall classification accuracies of 93.94% and 80.19%, respectively, substantially outperforming twelve state-of-the-art methods.

Abstract

Although tree species classification from Moderate Resolution Imaging Spectroradiometer (MODIS) time series data is critical for supporting various environmental applications, it is a challenging task due to several key difficulties: the subtle signature differences among tree species, strong spatial-spectral-temporal information coupling, and the difficulty of modeling large-scale topological context information. To better address these challenges, this paper presents a novel Graph-regulated Disentangled Sparse Mamba model (GDS-Mamba) for enhanced tree species classification, with the following contributions. (1) First, to improve large-scale context modeling, we design a mini-batch graph-regulated approach that explicitly explores topological correlation effects among input images. (2) Second, to disentangle the high-dimensional spatial-spectral-temporal information coupling for improved feature extraction, we propose a novel disentangling Mamba architecture tailored for capturing independent spatial patterns, spectral signatures, and temporal phenology behaviors in MODIS time series. (3) Third, to improve efficiency and subtle feature learning, we design novel sparse token approaches that adaptively learn the optimum subset of tokens to better address the correlation decay problem that bottlenecks standard Mamba models. Extensive experiments using large-scale annual MOD13Q1 data across two Canadian provinces (i.e., Alberta and Saskatchewan) achieved an overall accuracy of 93.94\% in Alberta and 80.19\% in cross-provincial evaluations, outperforming twelve state-of-the-art classification models.

Problem

• tree species classification
• MODIS time series
• spatial-spectral-temporal coupling
• topological context
• subtle signature differences

Innovation

• Graph-regulated
• Disentangling Mamba
• Sparse Tokens
• MODIS time series
• Tree species classification