Our study was awarded the 2025 Outstanding Paper Award by the @bigearthdata1.bsky.social at @tandfresearch.bsky.social !
I am grateful for this recognition to the editorial board, reviewers and my co-athors!
Read the full paper here: doi.org/10.1080/2096...
@sciencecharles.bsky.social
• GEM-Forest offers a high potential for straightforward reproducibility for other years of AEF embeddings (2017-2025)
Explore the GEM-Forest in our GEE app: danielp-cuni.projects.earthengine.app/view/gem-forest
Some takeaways:
• Traditional ML models (linear SVM, ridge and logistic regression and kNN) performed as well as, or better than, more complex models (Random Forest, XGBoost, neural networks - MLPs).
• It suggests that the AEF embedding space already provides a well-structured representation.
We created a new global map of forests and tree crops for 2020 at 10 meter resolution using satellite embeddings. After testing several machine learning methods, we found that simple linear models performed as well as or better than more complex ones. The map identifies forests with 91 % accuracy and separates tree crops with low confusion with forests. This shows that satellite embeddings can support reliable and efficient global forest monitoring and inform international and national policies.
Introducing GEM-Forest, a 10m Global EMbedding-based #forest and tree crop map for 2020 that combines linear SVM and Google DeepMind's AEF satellite #embeddings.
Read the preprint: doi.org/10.5194/egus...
@valeriomarsocci.bsky.social @adamhastie.bsky.social #philab
#GISChat #GeoSky #EOChat
📢Evaluation of ten #satellite-based and #reanalysis #precipitation datasets on a daily basis for #Czechia (2001–2021) by Daniel Paluba et al. @palubad.bsky.social
👉https://doi.org/10.1080/20964471.2025.2592444
💌 #ERA5 #timeseries #meteorology #remotesensing #earthobservation #GIS #opendata
The front page of our article in Big Earth Data journal titled "Evaluation of ten satellite-based and reanalysis precipitationdatasets on a daily basis for Czechia (2001–2021)" by Daniel Paluba, Vojtěch Bližňák, Miloslav Müller and Přemysl Štych. Access the study using the DOI link: https://doi.org/10.1080/20964471.2025.2592444
Our new #OpenAccess paper in @bigearthdata1.bsky.social @tandfresearch.bsky.social on evaluation of 10 satellite-based and reanalysis #precipitation datasets on a daily basis for #Czechia
doi.org/10.1080/20964471.2025.2592444
Most accurate: GSMaP & ERA5-Land
Worst results: CHIRPS, GLDAS & PERSIANN
Now you can watch the recording of my presentation on estimating #NDVI, EVI, LAI and FAPAR using #SAR and ancillary data through #machinelearning approaches during the @pangeos-ca22136.bsky.social webinar, summarizing two studies: doi.org/10.14712/233... & doi.org/10.1080/2096...
#geosky #GEE #EO
‼️ Fully-funded #PhD #position on “Integration of Advanced Spectral and Thermal #RemoteSensing Methods and #ArtificialIntelligence for Assessing the State and Functioning of Ecosystems across Scales” at @sciencecharles.bsky.social 🇨🇿
🔗 Detailed info: shorturl.at/GKICI
🗓️ Deadline: December 15, 2025
Last Monday, during the @copernicusecmwf.bsky.social C3S User engagement workshop within the #CzechSpaceWeek, I got a chance to present our work on the evaluation of precipitation datasets in Czechia.
The paper is currently under review, but the preprint is available here:
doi.org/10.13140/RG....
The event was a joint effort of four #COSTActions: #3DForEcoTech (groud-based LiDAR), #PANGEOS (remote sensing), #CLEANFOREST (tree ecophysiology and ecology) and #ARiF - Augmented Reality In Forestry.
Huge thanks to #PANGEOS for supporting my participation!
#ConnectedByTrees
💡@markuseichhorn.bsky.social's keynote on interdisciplinarity reinforced my postdoc goal to step beyond Earth Observation and collaborate across disciplines: to learn more by knowing less.
I also briefly showed my peers how SAR time series can support forest monitoring during the PANGEOS section.
🌳 Last week I participated at the Interdisciplinary Summer School on Forest Ecosystems in Oxford with 40+ participants from 25 countries.
5 intensive days with a field trip, lectures, hands-ons & even keynotes in pubs.
Thanks @mokros.bsky.social and his team for the organization!
#ConnectedByTrees
In my poster, I also shared new improvements where I incorporated temporal features, applied snow masking and extended the training dataset timeline to five years, which significantly boosted the accuracy of our #MachineLearning models!
👉 My poster is available here: github.com/palubad/SAR-...
During the ESA #LPS2025 in Vienna last week, I had the chance to present our work on estimating time series of vegetation indices for Central European #forests using Sentinel-1 #SAR, ancillary data and #ML approaches.
🔗 More details in our paper published earlier this year: doi.org/10.1080/2096...
⏳ Reminder: The application deadline for the 14th #ESA Advanced Land #Training course with the focus on #agriculture has been extended to 1st July 2025!
🗓️ 29 Sept. - 3 Oct. 2025
📍 Thessaloniki, Greece
👇 Find out more directly on the course website: landtraining2025.esa.int
#EOchat #GISchat
‼️ The deadline for applications has been extended to 1st July 2025!
Don't miss out the ESA Land training focusing on EO in Agriculture in Thessaloniki, Greece!
Register here: landtraining2025.esa.int
#EOTrainingAcademy #GISchat #EOchat #GeoSky #EarthObservation #RemoteSensing #Agriculture
📢 Last 3 days to apply for #ESA 14th Advanced #Training Course on Land #RemoteSensing with a focus on #Agriculture!
📅 When: 29 September - 3 October 2025
📍 Where: Thessaloniki, Greece
👉 Deadline: 12 June 2025
🔗 More info and registration: landtraining2025.esa.int
#GISchat #EOchat #geosky #eo
I have developed a user-friendly interface for using the MMTS-GEE. It is available here: code.earthengine.google.com/64a4db31218a... (you need a GEE account).
You can find it also on GitHub: github.com/palubad/MMTS...
The full-quality poster is available here: storage.mlcdn.com/account_imag...
Last week, I participated the #CLMSGA2025 in Kraków, Poland, where I presented the MMTS-GEE tool to generate spatially and temporally aligned SAR, optical, topographic and meteo. data. I also showed two case studies using MMTS-GEE:
1️⃣ doi.org/10.14712/233...
2️⃣ doi.org/10.1080/2096...
#GisChat #GeoSky
🌍 Join us for a presentation by Dr. Francesco Sarti from ESA/ESRIN on "Our planet from space: ESA’s satellite missions & education"
📅 Wednesday, 28 May 2025 14:00–15:30
📍 Kampus Hybernská, A.3 Velká zasedací místnost, Hybernská 998/4, Prague,
Register: evening-closet-d98.notion.site/1ff8779ec7b7...
🌧️ Which #precipitation dataset is the most accurate on a daily basis in Central European conditions?
We tried to answer this question evaluating 10 satellite-based and #reanalysis precipitation datasets on a daily basis for #Czechia (2001-2021).
🔗More in our #preprint: dx.doi.org/10.13140/RG....
@tandfresearch.bsky.social
Best-performing models available on Hugging Face: huggingface.co/palubad/SAR-...
Demo-code for running the models on GitHub: github.com/palubad/SAR-...
MMTS-GEE to generate multi-modal time series data: github.com/palubad/MMTS... , developed in our previous work: karolinum.cz/data/clanek/...
✅ Achieved temp. resolution of up to 240 measurements/year using SAR-based VIs with 2 S1 satellites in 2021 [soon to be restored with Sentinel-1C]
✅ Early disturbance detection: within 4 days vs 35 days with optical indices
✅ Fine-tuned #XGBoost and #RandomForest regressors outperformed #AutoML
👉 Temporally and spatially aligned SAR, DEM-based & meteo. features trained to estimate time series of #NDVI, #EVI, #LAI & #FAPAR
👉 Trained and tested in Czechia for healthy deciduous and coniferous forests, as well as disturbed areas ⏩ transferability tested for the whole Central Europe
https://doi.org/10.1080/20964471.2025.2459300
Can we estimate optical vegetation indices and biohpysical parameters for forests using #Sentinel1 #SAR & ancillary data? We tried to answer this in our latest paper in @bigearthdata1.bsky.social With the help of #machinelearning we generated SAR-based NDVI, EVI, LAI & FAPAR
doi.org/10.1080/2096...
Great to see it in action. Looks cool!
It reminded me that for larger countries, I should add a selection for smaller administrative units. I will try to implement it soon.
🌍 I have created an interactive #GEE app to generate #Sentinel2 cloud-free mosaics for any country and timeframe, defined by the user.
Generate your mosaics: danielp.users.earthengine.app/view/sentinel2-cloudfree-mosaic-generator
GitHub: github.com/palubad/S2-Mosaic-Generator
#gischat #dataviz