Climate AI Nordics

Climate AI Nordics

Nordic network for collaborations around tackling climate change using AI and machine learning.

Postdoc: Weakly supervised ML-based Earth observation for climate extreme impact quantification

RISE will soon be officially opening applications for a postdoc position focused on developing an advanced Earth observation (EO) machine learning (ML) framework. The role involves identifying, geolocalizing, and quantitatively estimating the physical and socio-economic impacts of climate extremes such as floods, storms, wildfires, and landslides, using satellite image sequences.

A primary bottleneck in climate adaptation is that historical impact data is plagued by annotation sparsity: detailed pixel-level ground truth of damage is rare, while rich textual data from news, insurance disclosures, and humanitarian reports is unstructured, noisy, and spatially coarse. Furthermore, models trained in data-rich regions may face data distribution shifts when deployed across different geographic, infrastructural, or socio-economic contexts.

To address these challenges, the postdoc will develop annotation-efficient, weakly supervised ML-based Earth observation models that leverage unstructured data as a source of scalable distant supervision. The framework will incorporate uncertainty quantification techniques to ensure that the AI-derived impact estimates are reliable and actionable for decision-makers.

Core research task: Architecting data-efficient change detection for geospatial foundation models

Design and develop specialized, data-efficient architectural modules, built on top of frozen geospatial foundation models. The goal is to automatically detect and isolate large, extreme-event-related physical changes resulting from floods, storms, or wildfires, by modeling the contrast between pre- and post-event observations in heterogeneous, multi-sensor image sequences (Sentinel-1/2, Landsat/HLS, VIIRS/MODIS).

The development will focus on two major machine learning challenges:

  • Weak supervision under annotation sparsity: Engineering models to learn robust representations of rapid environmental transitions using minimal labeled instances and sparse, noisy text reports treated as distant supervision.
  • Distribution shifts and uncertainty quantification: Integrating strict geometric constraints directly into the adapter layers so the model can inherently flag high epistemic uncertainty when encountering out-of-distribution geographical features or unprecedented extreme-event severities.

About Climes

The Swedish Centre for Impacts of Climate Extremes (climes) is a research and training platform building an interdisciplinary field on how climate extremes affect people, ecosystems and infrastructure in a changing world. The Centre brings together physical, medical, social and engineering sciences to advance knowledge, shape the Swedish research landscape and strengthen societal resilience. Our work centres on three themes: compiling high-quality data on the impacts of climate extremes; analysing the physical–societal interactions that drive consequences; and developing policy-actionable scenarios to prepare for future extremes.

Who You Are

Beyond the specific requirements for each role, we are looking for curious, analytical individuals with a genuine passion for artificial intelligence and its potential to solve real-world problems, often in cross-disciplinary settings. You are a proactive problem-solver who thrives in a dynamic, collaborative environment. You are motivated by the opportunity to contribute to impactful research with tangible outcomes. Excellent communication skills in English (written and spoken) are required. Knowledge of Swedish is a plus but not a strict requirement.

Are We a Good Match?

At RISE, we strive to create a workplace where diverse expertise and backgrounds converge to solve societal challenges. We offer a dynamic research environment where you will work on exciting projects with strong connections to industry and societal needs. With us, you’ll have ample opportunities to learn, develop professionally, and contribute to meaningful innovation alongside experienced colleagues both within machine learning and application areas.

Key details and information:

  • Duration: 1.5 Years
  • Hosted by: RISE
  • Location: Gothenburg
  • Status: Position upcoming. While the formal application link is not yet live, interested candidates are highly encouraged to get in touch early!
  • Contact: For questions about the upcoming role or to express early interest, please contact Dr. Olof Mogren (olof.mogren@ri.se)

You May Also Enjoy

Understanding the complex Earth system with Machine Learning and Hybrid Modelling

Event date:

Webinar with Markus Reichstein, Max-Planck-Institute for Biogeochemistry and ELLIS Unit Jena. The Earth system is a complex, dynamic and strongly interconnected system, shaped by interactions between climate, ecosystems, biogeochemical cycles and human activities. Rapidly growing streams of satellite, in-situ and experimental observations, together with advances in machine learning, offer new opportunities to detect patterns, infer processes and improve prediction across scales. Yet purely data-driven approaches often lack physical consistency and interpretability, while classical process-based models remain limited by uncertain parameterizations and incomplete representations of complex feedbacks. In this talk I will discuss how machine learning and hybrid modelling can help bridge this gap. By combining the versatility of data-driven methods with the constraints and explanatory power of mechanistic understanding, hybrid approaches can support more robust, interpretable and physically consistent models of the Earth system. Examples from the terrestrial biosphere, land-atmosphere exchange, carbon and water cycles, and climate extremes will illustrate how such approaches can contribute not only to improved prediction, but also to deeper scientific understanding of Earth system dynamics.

EarthShift: A benchmark for real-world ditribution shifts in Earth observation

Event date:

Webinar with Kelsey Doerksen, University of Cape Town and Arizona State University. Geospatial Foundation Models claim to offer powerful solutions to simplify and accelerate real-world problems, enabling the capabilities to monitor, analyze, and predict changes on our planet. Current Earth Observation benchmarks to quantify the performance of these models focus on measuring performance on diverse tasks and applications, typically measuring generalization in-distribution. However, when models are deployed, they must generalize to many out-of-distribution scenarios, such as new time periods, geographies, and sensors; and in many contexts, these models are brittle. We introduce EarthShift: the first public testbed for benchmarking robustness across multiple realistic distribution shifts encountered in remote sensing. EarthShift enables users to measure distributional robustness by comparing performance in- and out-of-distribution using datasets from paired data sources, temporal windows, geographic locations, and sensors. EarthShift provides a testbed to guide future work to create foundation models that are robust and reliable in real-world applications.