The Remote Sensing Gazette

Nov 16 – Nov 22, 2026 · Read one paper outside your usual orbit; the map gets clearer when you step off the familiar path.

WeatherDreamer: Learning a World Model from Multimodal Earth Observations

WeatherDreamer learns to understand and predict global weather from raw satellite and ground observations without complex pre-processed data. It can generate accurate short-term forecasts and global analyses faster and more efficiently than current operational methods.

WeatherDreamer: Learning a World Model from Multimodal Earth Observations · Kate Duffy, Elise Penn, Tristan Ballard et al. · arXiv:2611.18393

Towards GPROF V08: A Machine-Learning Framework for Simulating Passive Microwave Observations of Precipitation

A machine learning framework simulates how satellites see rain, creating more accurate training data than traditional methods. It may lead to better rain estimates from many satellites and help train future Earth observation tools.

Towards GPROF V08: A Machine-Learning Framework for Simulating Passive Microwave Observations of Precipitation · Simon Pfreundschuh, Christian D. Kummerow, Paula J. Brown · arXiv:2611.19966

GPROF V08: Improving Satellite Precipitation Estimates Across a Heterogeneous Constellation of Passive-Microwave Sensors

An upgraded algorithm uses machine learning to turn raw satellite data into more accurate global rainfall maps across many satellites. More reliable rain data can help daily weather forecasts and long-term climate studies.

GPROF V08: Improving Satellite Precipitation Estimates Across a Heterogeneous Constellation of Passive-Microwave Sensors · Simon Pfreundschuh, Paula J. Brown, Christian D. Kummerow · arXiv:2611.19978

Machine Learning for Near Real-Time Identification of Wildfire Burned Areas

An AI method turns satellite images into a map of burning probability for each location, to identify wildfire burned areas across the globe. It offers a faster automated way to track wildfires, helping emergency efforts and environmental assessment.

Machine Learning for Near Real-Time Identification of Wildfire Burned Areas · Khoi Nguyen, Adam W. Bargteil, Milton Halem · arXiv:2611.20036

Observing Earth's ionospheric irregularities from the Moon

A method studies Earth's turbulent upper atmosphere by analyzing satellite navigation signals, like GPS, received directly from the Moon. It gives a distant view of how these disturbances affect radio waves, especially over polar regions, for space weather and long-range communication.

Observing Earth's ionospheric irregularities from the Moon · Rong Yang, Zhihong Li, Andrea Nardin et al. · arXiv:2611.21062

Advancing global water cycle science with the upcoming MAGIC satellite mission through data assimilation

Simulations show that data from the upcoming MAGIC satellite mission, combined with computer models, can improve predictions of groundwater, river flow and how plants absorb carbon. Far more accurate predictions of groundwater, river flow and how plants absorb carbon are crucial for managing water resources and adapting to climate change.

Advancing global water cycle science with the upcoming MAGIC satellite mission through data assimilation · Helena Gerdener, Yorck Ewerdwalbesloh, Anne Springer et al. · arXiv:2611.15421

When Do Histology Foundation Models Need Geometric Test-Time Augmentation?

The standard trick of testing an AI model on rotated and flipped images still gives some accuracy gains, but the gains are much smaller for the latest foundation models. It offers ways to gain performance more efficiently with large foundation models for satellite and aerial imagery.

When Do Histology Foundation Models Need Geometric Test-Time Augmentation? · Lars Chen, Valentino F. Foit, Dorry Segev et al. · arXiv:2611.21565

All Gazettes · Remote Sensing