Meet- up & Networking Session

Arrival at HIDA Hub

Welcome and Introductions

Earth System Models and AI - Can they work together?

• Maximilian Gelbrecht (PIK)

Comprehensive Earth System Models (ESMs) are the key tools to model the dynamics of the Earth system and its climate, and in particular to estimate the impacts of increasing atmospheric greenhouse gas concentrations in the context of anthropogenic climate change. ESMs couple general circulation models (GCMs) of ocean and atmosphere with models of land surface processes, hydrology, ice, vegetation, atmosphere and ocean chemistry and carbon cycle model. Despite their remarkable success in reproducing observed characteristics of the Earth’s climate system, such as the spatial patterns of the increasing temperatures of the last century, there remain many great challenges for state-of-the-art Earth System Models such as the representation of extreme events, the multistability of components and the reduction of climate uncertainties in the models. The recent years saw the advance of purely data-driven AI models. In Earth System Science, we have now AI weather models such as GenCast, Aurora or Pangu Weather that are competitive with process-based weather models. Despite their remarkable success in weather forecasting, they are ultimately limited on the longer time scales and scenarios we need for climate projections. Can we still make use of AI models to address some of the aforementioned challenges in ESMs? In this talk, we will get an initial glimpse into how ESMs work, and what both the potentials and challenges are when we want to integrate AI approaches into these models.

Machine Learning to Advance Climate Science

• Peer Nowack (KIT)

Global climate change projections, such as those from the Coupled Model Intercomparison Project phase 6 (CMIP6), are still subject to substantial modelling uncertainties. These uncertainties limit our ability to fully assess future climate risks and to effectively inform policy. At the same time, running high-resolution, global climate models remains computationally intensive, posing further challenges. In this talk, I will highlight three key ways in which machine learning (ML) can help advance climate science: (a) using observational constraints, i.e. to constrain uncertainty in CMIP6 projections on the basis of high-dimensional relationships in Earth observations, (b) developing ML parameterizations of complex Earth system processes that can be both more accurate and computationally efficient than traditional parameterizations or explicit process representations, and (c) building fast AI-driven emulators of climate models to enable rapid exploration of a variety of future scenarios. I will emphasize the importance of addressing core challenges such as extrapolation in a changing climate system and understanding causal relationships, which are central to consider when working with purely data-driven ML models.

The Importance of Data Standardization for Effective Implementation of CleanTech

• Ariel L. Furst (MIT)

As new technologies are introduced to support the clean energy transition, questions of their efficiency often arise. Evaluating such technologies based on data reported without established standards makes even direct comparisons between them akin to comparing apples and oranges. We will discuss case studies focused on the challenges associated with data comparisons for battery technologies as well as electrocatalysis. We will then compare these reports to the standardized reporting for photovoltaics, which is often promoted as the ‘gold standard’ of testing and data reporting.

Roundtable discussion. Pursuing an Academic Career

• Wolfgang zu Castell (GFZ)
• Ariel L. Furst (MIT)
• Ann Gregory (University of Calgary)
• Peer Nowack (KIT)
• Stefan Bauer (Helnholtz AI)

Featuring keynote speakers from the School, this roundtable discussion will discuss pursuing a career in academia, the practice of interdisciplinary research, and provide practical advice for early-career researchers. The panel will share their experiences in academia, collaboration across disciplines, and insights on navigating an academic journey.

Mentoring session

Transfer Berlin - Potsdam

Arrival and welcome

Exploring Virus Roles in Climate Change Through Genomics and Data Science

• Ann Gregory (University of Calgary)

Viruses are the most abundant biological entities on Earth and are ubiquitous across ecosystems, from oceans and soils to the atmosphere, and play critical roles in regulating microbial communities and biogeochemical cycles. As climate change accelerates, understanding the ecological functions of viruses becomes increasingly important. This talk will examine how genomics and data science converge to uncover the roles of viruses in climate-related processes. We'll explore methods such as viral metagenomics (viromes), machine learning, and ecological modeling to identify viral genomes and investigate viral contributions to biogeochemical cycling, microbial interactions, and ecosystem resilience. By using these approaches, we aim to shed light on the hidden influencers of our planet’s climate system.

Guided Tour Telegrafenberg

• Ludwig Grunwaldt (GFZ)

Introduction to Remote Sensing

• Sigrid Roesnner (GFZ)

Remote Sensing Research at GFZ aims at establishing remote sensing as a core methodology towards better understanding of Earth system dynamics. For this purpose, the full range of available remote sensing data acquired by satellites, airborne and ground based systems is used for area-wide mapping at different spatial and temporal scales. For this purpose, novel bio- and geo-physical information products are derived and a variety of computer-based methods is developed for their analytical exploitation. The goal is the spatiotemporal characterization of Earth's surface properties related to a large variety of application fields, such as observation of landscape and vegetation development, impacts of climate change and natural disasters and effects of human land use.

Workshop: Feature Detection in Remote Sensing Data

• Mike Sips (GFZ)

In the workshop, we follow Jim Gray’s approach to data-intensive science to answer scientific questions using Sentinel-2 satellite data. Jim Gray’s approach to data-intensive science involves four phases: capturing data, storing and managing data efficiently, exploring data through statistical and computational analysis, and finally visualizing and communicating results. The workshop thoroughly examines each phase and discusses practical methods with participants using widely adopted Python libraries. The hands-on activities focus on characterizing and monitoring wildfire activity in California. While the workshop focuses on this specific use case, participants can use Gray’s approach in future projects as a reusable template for answering a wide range of questions with Sentinel-2 data.

Transfer Potsdam - Berlin

Arrival at HIDA Hub

Causality as an Inductive Bias in AI for Science

• Stefan Bauer (Helnholtz AI)

Deep neural networks have achieved outstanding success in many tasks ranging from computer vision, to natural language processing, and robotics. However, even models trained on internet scale data pale in their ability to understand the world around us, as well as continuously adapting to new tasks or environments. One prevailing approach is to train on massive, internet-scale datasets to cover diverse distributions, while an alternative focuses on leveraging inductive biases to improve generalization. This talk will explore causality as an inductive bias in neural networks, examining its potential to enhance robustness and generalization, particularly in AI for Science applications including inference of gene regulatory networks or materials discovery.

Technical Set-up (Optional)

Group Work Phase 1

• Mike Sips (GFZ)

Assignment: Feature Detection in Remote Sensing Data

Group Work Phase 2

• Mike Sips (GFZ)

Assignment: Feature Detection in Remote Sensing Data

Group Presentations

Farewell and Networking

Arrival at HIDA Hub

Walk / Transfer BIMSB

Tour: Berlin Institute for Medical Systems Biology (MDC-BIMSB) with focus on climate-friendly construction

• Ralf Streckwall (Head of the MDC Planning & Construction Department)

Talk: Helmholtz Foundation Model Initiative & AqQua - building a foundation model on images from diverse aquatic environments

• Dagmar Kainmueller (MDC)

Walk / Transfer HIDA Hub

Goodbye