Propagation of electromagnetic signals in Earth's neutral atmosphere inflicts errors in space geodetic observations. To model these errors accurately, it is necessary to use a signal tracing algorithm which is informed of the state of the atmosphere. We developed such an algorithm and tested it by processing Global Navigation Satellite System network observations. Our algorithm's main advantage is lossless utilization of atmospheric information provided by numerical weather prediction models.

We model Earth's lithosphere as a thin elastic shell and present numerical methods of isogeometric finite-element analysis to simulate its deformation in isostatic equilibrium using technologies from computer-aided design. The simulations also serve as a basis for identifying parameters of the model that are most plausible to explain observed data. This research has been done to showcase the capabilities of isogeometric analysis in solving higher-order problems in geoscientific applications.

A special model called AMmonia–CLIMate (AMCLIM) has been developed to understand and calculate NH₃ emissions from livestock farming. It is estimated that about 30 % of excreted N from livestock is lost due to NH₃ emissions from housing, manure management and land application of manure. High NH₃ volatilization often occurs in hot regions, while poor management practices also result in significant N losses through NH₃ emissions.

Large-scale wave modeling is essential for science and society, typically relying on resource-intensive numerical methods to simulate wave dynamics. In this study, we introduce a rolling AI-based method for modeling global significant wave height. Our model achieves accuracy comparable to traditional numerical methods while significantly improving speed, making it operable on standard laptops. This work demonstrates AI's potential to enhance the accuracy and efficiency of global wave modeling.

In this study, a new model called FLAML-LUE was created by combining the Fast Lightweight Automated Machine Learning (FLAML) model with light use efficiency (LUE) models; the latter provides the key variables of vegetation growth for modeling. Such knowledge- and data-driven models aim to reduce the large uncertainty in estimating gross primary productivity (GPP).

We describe TROLL 4.0, a simulator of forest dynamics that represents trees in a virtual space at 1 m resolution. Tree birth, growth, and death and the underlying physiological processes such as carbon assimilation, water transpiration, and leaf phenology depend on plant traits that are measured in the field for many individuals and species. The model is thus capable of jointly simulating forest structure, diversity, and ecosystem functioning, a major challenge in modelling vegetation dynamics.

We evaluate the capability of TROLL 4.0, a simulator of forest dynamics, to represent tropical forest structure, diversity, dynamics, and functioning in two Amazonian forests. Evaluation data include forest inventories, carbon and water fluxes between the forest and the atmosphere, and leaf area and canopy height from remote sensing products. The model realistically predicts the structure and composition as well as the seasonality of carbon and water fluxes at both sites.

We made a new regional ocean model to assist fisheries and ecosystem managers in making decisions in the Northeast Pacific Ocean (NEP). We found that the model did well simulating past ocean conditions like temperature and nutrient and oxygen levels and can even reproduce metrics used by, and important to, ecosystem managers.

Hail forecasting using numerical models remains a challenge. In this study, we found that the commonly used graupel-to-hail conversion parameterization method led to hail overforecasting in heavy rainfall cases where no hail was observed. By incorporating the spongy wet growth process, we successfully mitigated hail overforecasting. The modified scheme also produced hail in real hail events. This research contributes to a better understanding of hail formation.

Predicting ice sheet contributions to sea level rise is challenging due to limited data and uncertainties in key processes. Traditional models require complex methods that lack flexibility. We developed PINNICLE (Physics-Informed Neural Networks for Ice and CLimatE), an open-source Python library that integrates machine learning with physical laws to improve ice sheet modeling. By combining data and physics, PINNICLE enhances predictions and adaptability, providing a powerful tool for climate research and sea level rise projections.

The greening of cities is recommended to limit the effects of climate change. In particular, green roofs can provide numerous environmental benefits, such as urban cooling, water retention, and carbon sequestration. The aim of this research is to develop a new module for calculating green roof CO₂ fluxes within a model that can already simulate hydrological and thermal processes of such roofs. The calibration and evaluation of this module take advantage of long-term experimental data.

Our research introduces GPTCast, a novel method for very short term precipitation forecasting using radar data. By applying advanced machine learning techniques inspired by large language models, we developed a system that generates accurate and realistic weather predictions. We trained the model using 6 years of radar data from northern Italy, demonstrating its superior performance over leading ensemble extrapolation methods.

This study provides a comprehensive evaluation of unstructured meshes using the integrated Atmospheric Model Across Scales (iAMAS) over Antarctica, encompassing both surface and upper-level meteorological fields. Comparisons with the fifth-generation reanalysis (ERA5) from the European Centre for Medium-Range Weather Forecasts and observational data indicate that iAMAS performs well in simulating the Antarctic atmosphere.

This paper presents a numerical method to assess the origin of air pollution. Combined with a numerical air pollution transport and chemistry model, it can follow the contributions from a large number of emission sources. The result is a series of maps that give the relative contributions from, for example, all European countries at any point on the map.

A new internal solitary wave (ISW) numerical model in the northern South China Sea (ISWNM-NSCS v2.0) improves ISW predictions by incorporating background currents and inhomogeneous stratifications. Additionally, viscosity and diffusivity coefficients are optimized to maintain stable stratifications, extending the forecasting period. Sensitivity experiments show that ISWNM-NSCS v2.0 significantly enhances predictions of various wave properties.