HYPERSPECTRAL IMAGE KERNEL SPARSE SUBSPACE CLUSTERING WITH SPATIAL MAX POOLING OPERATION

We propose a novel spatiotemporally self-supervised fusion method to establish long-term daily seamless global XCO₂ and XCH₄ products. Results show that the proposed method achieves a satisfactory accuracy that distinctly exceeds that of CAMS-EGG4 and is superior or close to those of GOSAT and OCO-2. In particular, our fusion method can effectively correct the large biases in CAMS-EGG4 due to the issues from assimilation data, such as the unadjusted anthropogenic emission for COVID-19.

The semi-empirical physical approach derives PM_2.5 with strong physical significance. However, due to the complex optical characteristic, the physical parameters are difficult to express accurately. Thus, combining the atmospheric physical mechanism and machine learning, we propose an optimized model. It creatively embeds the random forest model into the physical PM_2.5 remote sensing approach to simulate a physical parameter. Our method shows great optimized performance in the validations.

This study generated the first global 1-km continuous NDVI product (STFLNDVI) for 4-decades by fusing multi-source satellite products. Simulated and real-data assessments confirmed the satisfactory and stable accuracy of STFLNDVI regarding spatial details and temporal variations. STFLNDVI is an ideal solution to the trade-off between spatial resolution and time coverage in current NDVI products, which of great significance for long-term regional and global vegetation and climate change studies.

Acquired daily soil moisture products are always incomplete globally (just about 30 %–80 % coverage ratio) due to the satellite orbit coverage and the limitations of soil moisture retrieval algorithms. To solve this inevitable problem, we generate long-term seamless global daily (SGD) AMSR2 soil moisture productions from 2013 to 2019. These productions are significant for full-coverage global daily hydrologic monitoring, rather than averaging as the monthly–quarter–yearly results.

Estimating ambient PM_2.5 and PM₁₀ considering their precursors and chemical compositions instead of AOD products; Both remote sensing (Sentinel-5P) and assimilated data (GEOS-FP) are adopted; Sample-based Cross-Validation R²s and RMSEs are 0.93 (0.9) and 8.982 (17.604) μg/m³ for PM_2.5 (PM₁₀), respectively; Achieving better performance compared to the baseline (AOD-based) in different cases (e.g., overall and seasonal).