Using a neural network model, county-level weather forecasts was achieved in the Western US. By combining traditional forecasting data with actual weather observations, the AI system achieved better temperature predictions at local scales. While showed promise for temperature forecasting, it still had difficulty accurately predicting extreme rainfall events. The research advances weather forecasting capabilities, potentially helping communities prepare for severe weather conditions.

Prescribing a fixed grassland density in the ORCHIDEE model limits its ability to capture grassland dynamics, leading to unrealistic mortality, especially in semi-arid grasslands. We proposed a dynamic density approach where a positive density-precipitation relationship emerges. This method improves spatial pattern, significantly reduces mortality, sustains productivity, and raises the aridity threshold above which frequent mortality events occur in grasslands.

Can we be sure that different computing environments, that should not change the model climate, indeed leave the climate unaltered? In this article, we present a novel methodology that answers whether two model climates are statistically the same. Besides a new methodology, able to detect significant differences between two model climates 60 % more accurately than a similar recent state-of-the-art method, we also provide an analysis on what actually constitutes a different climate.

Machine learning is increasingly used in environmental sciences for spatial predictions, but its effectiveness is challenged when models are applied beyond the areas they were trained on. We propose a Local Training Data Point Density (LPD) approach that considers how well a model's environment is represented by training data. This method provides a valuable tool for evaluating model applicability and uncertainties, crucial for broader scientific and practical applications.

Crystal growth and diffusion are common processes in geology. Our software MovingBoundaryMinerals.jl calculates compositional profiles in diffusion couples by simulating diffusion-growth processes for geometries with planar/cylindrical/spherical symmetries. Our software has been tested versus various benchmark cases and is provided as an open access software package. This package allows the further use of diffusion/growth phenomena in the calculation of the thermal histories of rocks.

Cities face growing heat challenges due to dense buildings, but predicting surface temperatures is complex because sunlight, airflow, and heat radiation interact. By simulating how sunlight bounces between structures and how heat transfers through materials, we accurately predicted temperatures on roofs, roads, and walls. The model successfully handled intricate city layouts thanks to GPU speed. By revealing which heat matters most, we aim to guide smarter city designs for a warming climate.

We propose an innovative climate modelling framework that combines statistical methods with climate simulations to study Earth's environmental systems. The model captures how orbital changes and carbon dioxide levels influence climate atmospheric dynamics, offering a detailed and efficient way to explore long-term processes. This tool provides new opportunities to investigate Earth's climate history and its implications for future changes.

We studied how pollution from cars and trucks contributes to tiny airborne particles that affect air quality and climate. These particles, called secondary organic aerosols, were often underestimated in global models. By improving how certain overlooked emissions from fuel use are represented in our model, we found that their impact is much larger than previously thought. Our results suggest that road traffic plays a far greater role in global air pollution than earlier estimates showed.

This study examines how radionuclides released from nuclear power plants are transported and transformed in the global ocean over time. Using an advanced ocean simulation model, it focuses on radionuclides released during the Fukushima accident and from planned wastewater discharges. The findings show that some radionuclides can travel across the Pacific within a few years and gradually spread to other ocean basins by mid-century, emphasizing potential long-term environmental impacts.

We set out to improve the accuracy of near-surface wind simulations in areas where buildings and tall vegetation have made the ground surface very rough. Through a clever use of differences between weather station measurements and reanalysis data, we estimated more realistic surface roughness values and created a new high-resolution map for China. This map greatly improves wind speed simulations and supports better decisions in wind-related fields.

A new configuration of the Australian Community Climate and Earth System Simulator coupled model, ACCESS-CM2, with a higher resolution ocean-sea ice component is introduced. The new version of the coupled climate model was designed to better capture smaller-scale ocean motions. While this configuration improves the representation of many aspects of the climate system, some biases from the existing lower-resolution version persist.

We present FRISIA version 1.0, a model for emulating sea level rise (SLR) and representing SLR impacts and adaptation in integrated assessment models (IAMs). FRISIA includes previously uncaptured coastal socio-economic feedback and a diverse set of impact strains, thereby improving the represenation of SLR impacts in IAMs. Here we describe the baseline behaviour of FRISIA, explore the effects of the additional feedback and showcase the coupling of FRISIA to an IAM.

This study developed a kilometre-scale sea-ice model in OpenFOAM that couples dynamic and thermodynamic processes for two types of ice, solid-like ice floes and fluid-like grease ice, under wave forcing. This model can help to improve data input for large-scale sea-ice models. Results show a linear relationship between the proportion of ice floes in the field and the overall viscosity. Additionally, we found that viscosity responds nonlinearly to the inclusion of thermodynamic sea-ice growth.

We present DRIVE v1.0, a data-driven framework to estimate road transport emissions, their temporal profiles, and the associated uncertainties. The method was applied to the city of Munich, where we present bottom-up emission estimates for the years 2019 to 2022. The estimates are compared against official municipal reports as well as national and European downscaled inventories.

Each vehicle from road traffic is a source of heat and an obstacle that induce wind when it passes. It directly impacts the local atmospheric conditions and the road surface temperature. These impacts are included in the numerical model of the Town Energy Balance, used to simulate local conditions in urbanised environments. Simulations show that road traffic has a significant impact on the road surface temperature up to a few degrees, and on local variables.

LISFLOOD-FP 8.2 is a framework for running real-world simulations of rapid, multiscale floods driven by impact events like tsunamis. It builds on the LISFLOOD-FP 8.0 and 8.1 papers published in GMD: whereas LISFLOOD-FP 8.0 focussed on GPU-parallelisation, and LISFLOOD-FP 8.1 focussed on static mesh adaptivity of (multi)wavelets, LISFLOOD-FP 8.2 combines GPU (graphics processing unit)-parallelisation with multiwavelet dynamic mesh adaptivity to drastically reduce simulation runtimes, achieving up to a 4.5-fold speedup.

One of the main constituents of Particulate Matter at the surface are Secondary Inorganic Aerosols (SIA) which are influenced by both anthropogenic emissions and the acidity of clouds and aerosols. This study shows improvements in introduced into the IFS-COMPO (Integrated Forecast System-COMPOsition) simulating the surface concentrations of SIA and the resulting changes in the total wet deposition for Europe, the US and South-East Asia.

We present a new version of the landslide model r.avaflow. It includes a model where different materials move on top of each other instead of mixing; a model supporting the entire range from block sliding to flowing; a model for slow-moving processes; and an interface for virtual reality visualization. Based on the results for four case studies we conclude that, at the moment, our enhancements are very useful for visualization of landslides for awareness building and environmental education.

Mechanistic vegetation models serve to estimate terrestrial carbon fluxes and climate impacts on ecosystems across diverse conditions. Here, we demonstrate and evaluate the rsofun R package, which provides a computationally efficient implementation of the P-model for site-scale simulations of ecosystem photosynthesis. Bayesian model fitting to observed fluxes and traits and evaluation on an independent test data set indicated robust calibration and unbiased prediction capabilities.

We present the Potsdam Integrated Assessment Modeling validation tool, piamValidation, an open-source R package for validating climate scenarios. The tool enables comparison of model outputs with historical data, feasibility constraints and alternative scenarios. Designed as a community resource, validation configuration files can serve as a knowledge-sharing platform. The main objective is to improve the credibility of Integrated Assessment Models by promoting standardized validation practices.

Regional Climate Model version 5 (RegCM5) reasonably reproduced the dominant spatiotemporal pattern of annual precipitation over the basin using the Emanuel convective scheme with the Nogherotto–Tompkins (NoTo) large-scale scheme. Using the MOLOCH non-hydrostatic dynamical core is recommended in future research for improving precipitation simulation and assessing the impacts of land-use change, due to dam-induced reservoirs on local climate.

Global models of the atmosphere contain errors that lead to inaccurate simulations. A software tool ("QuadTune") is presented that attempts to mitigate errors related to suboptimal parameter values. It also displays diagnostic plots that provide hints about where structural errors might lie in the model.

Volcanic eruptions are natural hazards impacting aviation, the environment, and climate. Here, we improve the simulation of volcanic material transport using the Weather Research and Forecasting (WRF-Chem) version 4.8. Analysis of ash, sulfate, and SO₂ mass budgets was performed. The direct radiative effect of volcanic aerosols was implemented. A preprocessor, PrepEmisSources, was developed to streamline the preparation of volcanic emissions.

We developed a deep learning model based on Generative Adversarial Networks (GANs) to improve rainfall forecasts in northern China. Traditional models struggle with accuracy, especially for heavy rain. Our model merges data from multiple forecasts, capturing detailed rainfall patterns and offering more reliable short-term predictions.

We present MET-AICE, a sea ice prediction system based on artificial intelligence techniques that has been running operationally since March 2024. The forecasts are produced daily and provide sea ice concentration predictions for the next 10 days. We evaluate the MET-AICE forecasts from the first year of operation, and we compare them to forecasts produced by three physically-based models. We show that MET-AICE is skillful and provides more accurate forecasts than the physically-based models.