Climate AI Nordics

Climate AI Nordics

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

Featured member, May 2026: Alouette van Hove


Hi Alouette! Could you tell us a bit about yourself and your work?

I am a postdoctoral researcher at the University of Oslo. I’m part of the ERC-funded ACTIVATE project, where I work on quantifying methane emissions from Arctic landscapes using a combination of machine learning, Bayesian inference, and drone-based field measurements. I am particularly interested in adaptive experimental design – the idea that we can use probabilistic models to actively guide where and when to deploy sensors or fly drones to maximize information value, rather than relying on fixed sampling schemes.

What kinds of research opportunities or collaborations are you excited to be part of in the future?

Methane surface fluxes are highly spatially and temporally variable, yet field data is expensive and sparse — which makes it a compelling problem for both machine learning and experimental design. I’m currently interested in machine learning models that turn sparse ground-based observations into spatially continuous flux maps using landscape and environmental predictors. I am also excited about applying Bayesian experimental design in practice here. The challenge is to develop general sampling policies that remain effective across varying conditions such as landscapes, domain sizes, and weather. I’d love to connect with others working on similar challenges.

Is there anything else you would like to share with the Climate AI Nordics network?

I am looking forward to meeting many members of the network at the Climate AI Nordics workshop in Copenhagen in June (registration is still open)!

What is the best way for people to get in touch with you?

You can best reach out to me by email at a.van.hove@geo.uio.no.

You May Also Enjoy

Featured project: BoquilaHUB - SOTA AI models for biodiversity

Published:

BoquilaHUB is an open-source desktop application that makes state-of-the-art AI for biodiversity accessible without complex setup or coding. It enables researchers to efficiently process large-scale camera trap and bioacoustic data using leading models like SpeciesNet and Perch 2. It also supports easy local deployment and shared GPU access, helping democratize AI tools for conservation worldwide.

Several climate related workshops at ECCV 2026

Published:

ECCV 2026 in Malmö features as many as 93 workshops, several of which are related to climate and environment and Earth observation, highlighting a growing momentum for environmental applications in computer vision. Climate AI Nordics is proud to co-organize the second edition of the AI for Climate and Conservation (AICC2) workshop as a full-day event. The conference also features two specialized Earth Observation workshops and two additional climate and ecology-focused tracks, bringing together leading global researchers to address urgent planetary challenges.

2nd Workshop on AI for Climate and Conservation (AICC-2) at ECCV 2026

Event date:

Climate AI Nordics is glad to announce that the [2nd Workshop on AI for Climate and Conservation (AICC-)](https://sites.google.com/g.harvard.edu/aicc2eccv26/) has been accepted at [ECCV 2026](https://eccv.ecva.net/)! The AICC-2 workshop will take place in Malmö, Sweden, Sep 8th or 9th (TBD).

EarthShift: A new testbed for benchmarking robustnessGeospatial Foundation Models to real-world distribution shifts

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.