Understanding the change in climate that would occur under different future pathways of greenhouse gas emissions and changes in land use is crucial. Here, we develop a new simple climate model to help study this. We reduce the number of inputs so that our model can be connected to a model of the human causes of climate change. This way, we can study the interaction between climate change and society, including climate impacts. Our model broadly agrees with historical observations.

Geophysically accurate model of mantle convection requires a mesh with a width on the order of ~1 km. Traditional codes represent the systems of equation by setting up the associated matrix. However, at the scales we want to operate, even forming this matrix is hardly possible. Thus, we use the matrix-free framework HyTeG to create the geophysical model and verify it with numerical experiments while assessing the scalability of the framework and laying out the difficulties involved.

Global Atmosphere (GA) configurations of the Unified Model (UM) and Global Land (GL) configurations of Joint UK Land Environment Simulator (JULES) are developed for use in any global atmospheric modelling application. We describe a recent iteration of these configurations, GA8GL9, which includes improvements to the representation of convection and other physical processes. GA8GL9 is used for operational weather prediction in the UK and forms the basis for the next GA and GL configuration.

In this work, we propose a novel approach, called chunkindex, that was designed to improve the access to time series from native NetCDF (Network Common Data Form) files in the cloud. The advantage of our approach is that it keeps existing data as they are without requiring any reformatting. The idea is to reduce the amount of data read from the NetCDF file by creating sub-chunks that allow extracting smaller portions of compressed data without reading the entire chunk.

Hydrogen (H₂) is an indirect greenhouse gas by increasing methane (CH₄) lifetime. Interaction between H₂ and CH₄ is important for hydrogen’s global warming potential (GWP). Global models do not represent this interaction well; H₂ or CH₄ are prescribed at the surface. We implement an interactive H₂ scheme into a global model coupled with interactive CH₄. We simulate scenarios demonstrating its capability, improving model performance and more accurately representing H₂-CH₄ interaction.

Rivers in cold climate regions such as Canada undergo freeze up during winters which makes the estimation forecasting of under-ice discharge very challenging and uncertain since there is no reliable method other than direct measurements. The current study explored the potential of deploying a coupled modelling framework for the estimation and forecasting of this parameter. The framework showed promising potential in addressing the challenge of estimating and forecasting the under-ice discharge.

The role of Arctic clouds in the regional climate remains uncertain due to insufficient understanding of the amount of liquid droplets and ice crystals present in these clouds. An aerosol-cloud model is employed to examine the role of different aerosol types and freezing parameterizations on the number of ice crystals. The choice of freezing parameterization significantly changes the number of ice crystals impacting the interpretation of the evolution and warming effect of Arctic clouds.

Phytoplankton are vital for marine ecosystems and nutrient cycling, detectable by optical satellites. Data gaps caused by clouds and other non-optimal conditions limit comprehensive analyses like trend monitoring. This study evaluated DINCAE and DINEOF gap-filling methods for reconstructing chlorophyll a datasets, including total chlorophyll a and five major phytoplankton groups. Both methods showed robust reconstruction capabilities, aiding pattern detection and long-term ocean colour analysis.

This paper presents recent advances in ICON-ART, a modeling system that simulates atmospheric composition – such as gases and particles – and their interactions with weather and climate. By integrating updated chemistry, emissions, and aerosol processes, ICON-ART enables detailed, scale-spanning simulations. It supports both scientific research and operational forecasts, contributing to improved air quality, weather and climate predictions.

Satellites can reveal greenhouse gas point sources, but current point source extraction methods rely on manual inspection. We developed a point-object-detection-based deep learning method for fast, automated detection and quantification of these sources. The model was trained on a large synthetic dataset and tested for generalization using two independent datasets, including simulations and satellite observations.

To support air quality decision-making in future emission scenarios, this study presents an agent model for a classic chemical transport model based on a transformer deep-learning framework. Addressing the long runtimes and input/output limitations of previous approaches, our agent model accurately reproduces simulations of fine particulate matter and ozone, enabling rapid air quality assessment.

Groundwater sustains vegetation and regulates land-atmosphere exchanges, yet most Earth system models oversimplify its dynamics. Our study develops an integrated framework coupling a dynamic vegetation model with the three-dimensional hydrological model ParFlow to explicitly represent groundwater-vegetation interactions. The results provide evidence that groundwater flow strongly regulates water exchanges and provides a powerful tool to improve simulations of water cycles in Earth system models.

Ice shelves are thinning and losing mass in West Antarctica because of interaction with warm water. The topography of the bedrock beneath the ice shelves is difficult to measure but important for understanding how quickly the ice shelves will melt. This study uses gravity data to infer the bedrock topography beneath the ice shelves. We use statistical methods to create an ensemble of bathymetry models that sample the uncertainty of the assumptions in the problem.

Ocean waves shape winds close to the surface and extend their impact throughout the atmospheric boundary layer. In this study, we built a new modeling tool that allows simulations to follow the moving wave surface itself. By testing different wave and wind conditions, we show how waves change air motion, turbulence, and energy exchange above the ocean. This approach improves our ability to represent air–sea interactions, with implications for weather studies and offshore wind energy.

Accurately simulating rainfall is essential to understand the impacts of climate change, especially extreme events such as floods and droughts. Climate models simulate the atmosphere at a coarse resolution and often misrepresent precipitation, leading to biased and overly smooth fields. We improve the precipitation using a machine learning model that is data-efficient, preserves key climate signals such as trends and variability, and significantly improves the representation of extreme events.