Leveraging Multi-task Deep Learning for Quantitative Monitoring of Fugitive Methane Plumes from Spaceborne Hyperspectral Imagery
A deep learning framework is proposed for the spaceborne retrieval of methane emissions based on data simulation using Large Eddy Simulation (LES), Radiative Transfer Equation (RTE), and data augmentation techniques. The framework includes an instance segmentation algorithm for isolating and identifying methane emission sources, and multi-task learning models that outperform single-task models in methane concentration inversion, plume segmentation, and emission rate estimation.
Enhancing Sentinel-2 Satellite Image Resolution: Evaluating Advanced Techniques Based on Convolutional and Generative Adversarial Neural Networks
Advanced super-resolution techniques based on convolutional and generative adversarial neural networks can effectively enhance the spatial resolution of Sentinel-2 satellite imagery.
Large-Scale Masked Autoencoding Reduces Label Requirements for Satellite-Based Monitoring of Climate Change Effects
Self-supervised pretraining using masked autoencoding on large-scale Synthetic Aperture Radar (SAR) data significantly reduces the labeling requirements for downstream tasks crucial to climate change monitoring, such as vegetation cover prediction and land cover classification.
Hyperspectral Unmixing of Agricultural Images Captured by UAV Using an Adapted U-Net Architecture
This paper proposes a hyperspectral unmixing algorithm based on an adapted U-Net network architecture to achieve more accurate unmixing results on existing and newly created hyperspectral unmixing datasets from agricultural UAV imagery.
SEN12-WATER: A Comprehensive Multisource and Multitemporal Dataset for Drought Analysis and Water Resource Management
This study introduces a new multisource and multitemporal dataset, SEN12-WATER, which integrates Sentinel-1 SAR, Sentinel-2 optical, elevation, and slope data to enable detailed analysis and forecasting of water dynamics for drought monitoring and water resource management.
High-Resolution Satellite Imagery Dataset for Deep Learning-Based Forest Wildfire Detection
The development and application of a large, labeled satellite imagery dataset (CWGID) for training deep learning models to accurately detect forest wildfires.
Estimating Anisotropic Surface Reflectance from Sparse Satellite Imagery using Neural Radiance Fields and the RPV BRDF Model
BRDF-NeRF can successfully estimate the Rahman-Pinty-Verstraete (RPV) BRDF model parameters from as few as three or four satellite images, enabling high-quality novel view synthesis and digital surface model generation.
Evaluating the Transferability of Remote Sensing Foundation Models for Global Crop Type Mapping
Recent remote sensing foundation models can improve the transferability of crop type classification across different geographic regions, especially in data-scarce developing areas.
Automating the Creation of Thermal Hotspot Datasets from Sentinel-2 Raw Multispectral Imagery for Onboard Artificial Intelligence
This work presents a novel methodology to automate the creation of datasets for the detection of thermal hotspots, such as wildfires and volcanic eruptions, directly from Sentinel-2 raw multispectral data. The proposed approach leverages existing algorithms designed for processed Level-1C data to efficiently identify and annotate the corresponding raw data granules.
High-Resolution Time-Series Observations for Detecting Earth Surface Anomalies
The core message of this article is to propose the anomaly change detection (AnomalyCD) technique, which can process an unfixed number of time-series observations to distinguish anomalous changes from normal changes, without the need for human supervision.
© 2024 by Linnk AI