DFKI Earth & Space Applications
AI for Earth Observation and Space
Earth observation (EO) covers an ever-growing range of applications, from crop yield prediction and land-cover mapping to urban planning and disaster management.
Given the sheer volume of data being produced, simple manual analysis is hardly feasible anymore, so automated tools such as AI — or, more precisely, machine learning (ML) — are required.
This is what the AI4EO Solution Factory specializes in. Despite the wide range of application areas, the underlying technical methods often differ only slightly. When developing new applications, we can therefore build not only on our expertise but also reuse existing modules of the Solution Factory.
The AI4EO Solution Factory is funded through the ESA InCubed Programme.
Our CVPR 2026 paper presents YieldSAT — the first multimodal dataset for high-resolution crop yield prediction at field and subfield level, combining combine harvester data, Sentinel-2 time series, weather, soil and topography across 4 countries and 9 years.
Read more →
The Yield Consortium explores artificial intelligence approaches to reliably predict agricultural yields based on satellite data.
Read more →
A 2022 interview with then-DFKI researcher Marlon Nuske about the AI4EO Solution Factory and other DFKI projects.
Read more →
AI-based prediction of yield, quality and optimal mowing time for permanent grassland
SmartCut combines Earth observation, weather and field data with AI to predict the optimal mowing and fertilisation dates for permanent grassland – increasing productivity, forage quality and sustainable milk production.
AI-accelerated urban climate simulation for digital twins
KIB-UDZ develops ADAPT, an AI-based tool that lets municipalities simulate heat and heavy-rain adaptation measures in urban digital twins within seconds — combining the PALM-4U climate model with efficient machine learning.
Earth Observation for Agriculture
We use satellite imagery to predict agricultural yield reliably and early in the season, helping to use resources more efficiently and to make informed agronomic decisions.
„The AI4EO Solution Factory is a unique opportunity for our industry partners to expand their competence at the intersection of artificial intelligence and earth observation. Within the solution factory we develop cutting edge solutions for a range of different business cases. Within the current project on agricultural yield prediction, we have established a close collaboration with our industry partners which enables a fruitful environment for innovation.“
Prof. Dr. Prof. h.c. Andreas Dengel
YieldSAT: A Multimodal Benchmark Dataset for High-Resolution Crop Yield Prediction
M. Miranda, D. Pathak, P. Helber, B. Bischke, H. Najjar, F. Mena, C. Sanchez, A. Pai, D. Arenas, M. Valdenegro, M. Charfuelan, M. Nuske, A. Dengel
Deep learning emulators for large-scale, high-resolution urban pluvial flood prediction
D. K. Natarajan, R. Mukherjee, M. Charfuelan, M. Nuske and A. Dengel
Estimating Evapotranspiration using Physics-Informed Learning Enables Accurate Yield Prediction at Subfield Resolution
M. Miranda, M. Charfuelan, M. Valdenegro-Toro, and A. Dengel
An interdisciplinary team combining AI and earth observation expertise.
You would like to apply earth observation data to your business case or have another request? We look forward to hearing from you.
info@ai4eo-factory.deAI4EO Solution Factory · Trippstadter Str. 122, DE 67663 Kaiserslautern ·+49 (0) 631 20575 0