ML Earth Sciences

Hydrology Paper of the Day @abhilashsingh.bsky.social suggested by @mlearthsciences.bsky.social on obtaining subsurface soil moisture from surface soil moisture observations: conditional generative modeling in the context of Fickian diffusion, and a reverse diffusion process from a neural network.

Can weak physics improve machine-learning generalization to new (or any) sites compared to hard-constraint physics-informed machine learning that requires site-specific details?

We address this question in our new paper in GRL.

doi.org/10.1029/2025...

#soilmoisture #machinelearning

Looking forward to the AI in Science Summit 2025,
📆 3–4 Nov in Copenhagen!
Markus Reichstein co-curates the Planet & Climate workshop with Sašo Džeroski.
Speakers: Jonas Peters, Gustau Camps-Valls, Florence Rabier & Christian Igel
🔗 ais25-summit.webflow.io/thematic-wor...

Hour by Hour PM2.5 Mapping Using Geostationary Satellites This study estimates ground-level fine particulate matter (PM2.5) concentrations using geostationary satellites-derived Aerosol Optical Depth (AOD) and radiance measurements and meteorological paramet...

🧪🛰️🌍
A great use of satellite platforms: ML applied to #geostationary native radiance observations (GOES and TEMPO) with High res model outputs to derive surface PM2.5

pubs.acs.org/doi/10.1021/...

@aerosolwatch.bsky.social @mparrington.bsky.social

Hydrology Paper of the Day @mlearthsciences.bsky.social on linkages between rainfall and landslides as predicted by machine learning: investigating how ANNs demonstrate a geography of risk; comparisons with known areas of landslides; and application to the Serra Geral region of southern Brazil.

Spatially distributed antecedent rainfall thresholds for landslide occurrence: a multitask machine learning modelling approach Landslide susceptibility and the amount of antecedent rainfall required to trigger landslides are conceptually tightly linked but usually modelled separately. We propose an approach for modelling b...

Spatially distributed antecedent rainfall thresholds for landslide occurrence: a multitask machine learning modelling approach

doi.org/10.1080/0262...
#machinelearning #landslide

Hydrology Paper of the Day @mlearthsciences.bsky.social on the use of frequency-domain neural networks for soil moisture imputation: weather station observations and model application by sliding windows and spatial convolution; a comparison of approaches; and rainfall magnitudes and sensitivities.

Leveraging Neural Operator and Sliding Window Technique for Enhanced Subsurface Soil Moisture Imputation Under Diverse Precipitation Scenarios Developed a novel Fourier Neural Operator (FNO) to enhance subsurface soil moisture imputation by employing a sliding window concept that seamlessly integrates rainfall, soil temperature, and norm...

Worried about missing value?

A novel imputation framework based on Fourier neural operators (FNO) for soil moisture. The FNO model outperforms traditional approaches, and incorporating temporal lag reduces error by up to 15% in the diverse climates in India and Zambia.
doi.org/10.1029/2025...

Recently published in #MachineLearningEngineering 🥳

Read the #openaccess article: 'Understanding interpretable patterns of Shapley behaviours in materials data'
👉 ow.ly/bmmN50WHE2X
From Tommy Liu and Amanda S Barnard, (Australian National University)
#DataAnalysis #ExplainableArtificialIntelligence

We are excited to announce openings for SIX new members to join the @ametsoc.org Committee on Hydrology! We are looking for 2 undergrad/grad student members and 4 regular members.

Applications are due September 16 (CV and letter of interest required): forms.gle/qKdsLTTK1k9e...

Hydrology Paper of the Day @pnas.org on how climate change, increases in stream temperature, and latitudinal gradients affect freshwater fish populations: the RivFishTime database paired with the Global Biodiversity Information Facility database for spatial analyses, and identifying decadal trends.

#CriticalZone at #AGU25

A flyer for the "B011" AGU session

BO11 - Advancing Biogeochemical Cycle Modeling with Artificial Intelligence (Al): Bridging Data-Driven Methods and Process-Based Approaches #AGU25

🌎 Are you attending #AGU25?
Consider submitting your abstract to our session on Challenges and Solutions for Hydrologic Scaling Across Multiple Processes and Scales.
More information below!!!

Open source, open science for earth, climate and geospatial science? Coming to #AGU25? Build tools in #Python @jupyter.org?

Submit an abstract for this session and come meet us and like minded scientists!

First post !!!

Happy to share our new paper is out in Engineering Applications of AI!
"Overcoming Data Scarcity" uses transfer learning + satellite fusion to predict soil moisture with 55% less in-situ data
🔗 doi.org/10.1016/j.enga…
�� abhilashsingh.net/codes.hthtml
#RemoteSensing #ML #AI

Publish in AI for Science for:

🔬 Broad scope for AI-driven scientific breakthroughs
🌍 Open access publishing at no cost to you
✅ Review by top international experts
⚡ Fast publication
📝 Formatting your way and we’ll handle the rest

🔗 Submit now: iopscience.iop.org/journal/3050...

#AIResearch

A new paper is out in Engineering Applications of AI!
"Overcoming Data Scarcity" uses transfer learning + satellite fusion to predict soil moisture with 55% less in-situ data
🔗 doi.org/10.1016/j.enga…

✊ From observations to actionable information: Take a look at our mission in our digital flyer!

It gives an overview of 40+ years of satellite data from a growing number of Essential Climate Variables to strengthen climate understanding and inform effective decision-making:

t1p.de/lbgyl #LPS25

Wrapping up your PhD? I’m planning to hire a postdoc in the next few months (start date flexible). Will start formally advertising soon, but you heard it here first!! Possible research topics include - critical zone hydrology, agricultural water quality, SW-GW interactions. Reach out if interested!

Hydrology Paper of the Day @rarakihydro.bsky.social on a model of soil moisture loss that utilizes a nonlinear function: model fitting to SMAP remote sensing data; an examination of global-scale patterns; how aridity, sand fraction and landcover affects outputs; and the role of evapotranspiration.

⏳ Time is running out!

Apply by 15 June to join the JRC as a hydrologic modeller and help shape the future of early warnings in Europe and beyond.

Make an impact with #CEMS, #EFAS, #GloFAS.
🔗 recruitment.jrc.ec.europa.eu/vacancy/1885

#Floods #Drought #Hydrology

PC: Yifang Ban, KTH

Hydrology Paper of the Day @bioclimatology.bsky.social on how more than six years of data from a beech forest in Germany indicates changes in Critical Zone hydrology due to climate change: rainfall trends and canopy-scale partitioning; seasonal changes; and identifying storage and drivers.

Calling all river scientists & engineers! #RiverFlow2026 is now accepting abstracts. Share your latest research in fluvial hydraulics and join us in Thessaloniki, Greece! Submit your abstract by August 15, 2025.
🔗 riverflow2026.web.auth.gr 
#Hydraulics #RiverScience

Interesting readings very recently published on 𝘿𝙞𝙜𝙞𝙩𝙖𝙡 𝙏𝙬𝙞𝙣𝙨 𝙤𝙛 𝙩𝙝𝙚 𝙀𝙖𝙧𝙩𝙝

Digital Twins of the Earth Between Vision and Fiction
lnkd.in/dYQTs5wM

and also:
Bring digital twins back to Earth
lnkd.in/dih-JvGT

Hydrology Paper of the Day @ianabrown.bsky.social on how landscape affects drone-based photogrammetry: comparisons with LiDAR DEMs and CHMs for the Kronängen area of boreal forests and meadows; differences in topography; and accuracy comparisons with respect to weather, location and vegetation.

Hydrology Paper of the Day @okke-batelaan.bsky.social on determining ET and temperature beneath the canopy of a forest by drone and airplane thermal imaging, LiDAR scanning and modelling: two catchments in Australia; satellite optical sensing; upscaling; validation; and quantifying uncertainties.

Hydrology Paper of the Day @geospatialtao.bsky.social on a novel framework for UAS remote sensing of rangelands: the Reynolds Creek Experimental Watershed in southwest Idaho; machine learning image classification; comparison with existing data products; and the effects of spatial resolution.

🌍 AI is transforming climate research!

Catch us at #EGU25 presenting cutting-edge work on applying #MachineLearning, hybrid models, and causal discovery in climate research.

⚡️ Don’t miss: Camps-Valls, Reid, Ouala, Beucler + more!

#AI4Climate #ML4Science #EGU25 #ClimateAI