The efficiency of the numerical simulation of radiative transport is shown on modern server-class graphics cards (GPUs). The low-cost prefactor on GPUs compared to general-purpose processors (CPUs) enables future large retrieval campaigns for multi-channel data from infrared sounders aboard low-orbit satellites. The validated research software JURASSIC is available in the public domain.

Air transports moisture and heat, shaping the weather we experience. When and where was this air moistened and warmed by the surface? To address this question, atmospheric models trace the history of air parcels in space and time. However, their uncertainties remain unexplored, which hinders their utility and application. Here, we present a framework that sheds light on these uncertainties. Our approach sets a new standard in the assessment of atmospheric moisture and heat trajectories.

Wildfire is a devastating Earth system process that burns about 500 million hectares of land each year. It wipes out vegetation including trees, shrubs, and grasses and causes large losses of economic assets. However, modeling the spatial distribution and temporal changes of wildfire activities at a global scale is challenging. This study built a machine-learning-based wildfire surrogate model within an existing Earth system model and achieved high accuracy.

We have adapted machine learning techniques to build a model of the land surface in Great Britain. The model was trained using data from a very complex land surface model called JULES. Our model is faster at producing simulations and predictions and can investigate many different scenarios, which can be used to improve our understanding of the climate and could also be used to help make local decisions.

We modify the numerical model of forest growth MAIDENiso by explicitly simulating snow. This allows us to use the model in boreal environments, where snow is dominant. We tested the performance of the model before and after adding snow, using it at two Canadian sites to simulate tree-ring isotopes and comparing with local observations. We found that modelling snow improves significantly the simulation of the hydrological cycle, the plausibility of the model and the simulated isotopes.

We developed a technique to remap sea ice tracer quantities between circular discrete element distributions. This is needed for a global discrete element method sea ice model being developed jointly by Los Alamos National Laboratory and Sandia National Laboratories that has the potential to better utilize newer supercomputers with graphics processing units and better represent sea ice dynamics. This new remapping technique ameliorates the effect of element distortion created by sea ice ridging.

We present a global version of the Ecosystem Demography (ED) model which can track vegetation 3-D structure and scale up ecological processes from individual vegetation to ecosystem scale. Model evaluation against multiple benchmarking datasets demonstrated the model’s capability to simulate global vegetation dynamics across a range of temporal and spatial scales. With this version, ED has the potential to be linked with remote sensing observations to address key scientific questions.

Turbulent mixing is a vital process in 3D modeling of particle transport in the ocean. However, since turbulence occurs on very short spatial scales and timescales, large-scale ocean models generally have highly simplified turbulence representations. We have developed parametrizations for the vertical turbulent transport of buoyant particles that can be easily applied in a large-scale particle tracking model. The predicted vertical concentration profiles match microplastic observations well.

For realistically simulating the recently observed distributions of dissolved ²³⁰Th and ²³¹Pa in the ocean, we highlight the importance of the removal process of ²³¹Pa and ²³⁰Th at the seafloor (bottom scavenging) and the dependence of scavenging efficiency on particle concentration. We show that consideration of these two processes can well reproduce not only the oceanic distribution of 231Pa and 230Th but also the sedimentary ²³¹Pa/²³⁰Th ratios.

Climate-change-induced sea level rise is a major threat for coastal and low-lying regions. Projections of coastal sea level changes are thus of great interest for coastal risk assessment and have significantly developed in recent years. In this paper, the objective is to provide high-resolution (6 km) projections of sea level changes in the northeastern Atlantic region bordering western Europe. For that purpose, a regional model is used to refine existing coarse global projections.

The influence of climate on landform evolution has attracted great interest over the past decades. This paper presents a simple model for simulating the influence of topography on precipitation and the decrease in precipitation over large continental areas. The approach can be included in numerical models of large-scale landform evolution and causes only a moderate increase in the numerical complexity. It opens a door to investigating feedbacks between climate and landform evolution.

We introduce the first chain of computationally efficient Earth system model (ESM) emulators to translate user-defined greenhouse gas emission pathways into regional temperature change time series accounting for all major sources of climate change projection uncertainty. By combining the global mean emulator MAGICC with the spatially resolved emulator MESMER, we can derive ESM-specific and constrained probabilistic emulations to rapidly provide targeted climate information at the local scale.

A quarter of the surface of the Earth is covered by marine sediments rich in calcium carbonates, and their dissolution acts as a giant antacid tablet protecting the ocean against human-made acidification caused by massive CO₂ emissions. Here, we present a new model of sediment chemistry that incorporates the latest experimental findings on calcium carbonate dissolution kinetics. This model can be used to predict how marine sediments evolve through time in response to environmental perturbations.

WRF model accuracy relies on numerous aspects, and the model parameters are one of them. By calibrating the model parameters, we can improve the model forecast. However, there exist hundreds of parameters, and calibrating all of them is unimaginably expensive. Thus, there is a need to identify the sensitive parameters that influence the model output variables to reduce the parameter dimensionality. This study addresses the different methods and outcomes of parameter sensitivity analysis.

Biogeochemical models used by the scientific community can support society in the quantification of the expected environmental impacts caused by global climate change. The Biome-BGCMuSo v6.2 biogeochemical model has been created by implementing a lot of developments related to soil hydrology as well as the soil carbon and nitrogen cycle and by integrating crop model components. Detailed descriptions of developments with case studies are presented in this paper.

Earth-observing satellites provide routine measurement of several ocean parameters. However, these datasets have a significant amount of missing data due to the presence of clouds or other limitations of the employed sensors. This paper describes a method to infer the value of the missing satellite data based on a convolutional autoencoder (a specific type of neural network architecture). The technique also provides a reliable error estimate of the interpolated value.

Reservoir hydrodynamics is often described in numerical models differing in dimensionality. 1D and 2D models assume homogeneity along the unresolved dimension. We compare the performance of models with 1 to 3 dimensions. All models presented reasonable results for seasonal temperature dynamics. Neglecting longitudinal transport resulted in the largest deviations in temperature. Flow velocity could only be reproduced by the 3D model. Results support the selection of models and their assessment.

There is growing interest in data-driven weather forecasting, i.e., to predict the weather by using a deep neural network that learns from the evolution of past atmospheric patterns. Here, we propose three components to add to the current data-driven weather forecast models to improve their performance. These components involve a feature that incorporates physics into the neural network, a method to add data assimilation, and an algorithm to use several different time intervals in the forecast.

The GOBLIN (General Overview for a Backcasting approach of Livestock INtensification) model is a new high-resolution integrated bottom-up biophysical land use model capable of identifying broad pathways towards climate neutrality in the agriculture, forestry, and other land use (AFOLU) sector. The model is intended to bridge the gap between hindsight representations of national emissions and much larger globally integrated assessment models.

This paper provides metadata and first analyses of the volc-pinatubo-full experiment of CMIP6-VolMIP. Results from six Earth system models reveal significant differences in radiative flux anomalies that trace back to different implementations of volcanic forcing. Surface responses are in contrast overall consistent across models, reflecting the large spread due to internal variability. A second phase of VolMIP shall consider both aspects toward improved protocol for volc-pinatubo-full.

The 4D ensemble Kalman filter data assimilation system for the whole neutral atmosphere has been updated. The update includes the introduction of a filter to reduce the generation of spurious waves, change in the order of horizontal diffusion of the forecast model to reproduce more realistic tidal amplitudes, and use of additional satellite observations. As a result, the analysis performance has been greatly improved, even for disturbances with periods of less than 1 d.

The Green's functions methodology offers a systematic, easy-to-implement, computationally cheap, scalable, and extendable method to tune uncertain parameters in models accounting for the dependent response of the model to a change in various parameters. Herein, we successfully show for the first time that long-term errors in earth system models can be considerably reduced using Green's functions methodology. The method can be easily applied to any model containing uncertain parameters.

A dynamic vegetation model, designed to estimate potential impacts of early vascular vegetation, namely, lycopsids, on the biogeochemical cycle at a local scale. Lycopsid Model (LYCOm) estimates the productivity and physiological properties of lycopsids across a broad climatic range along with natural selection, which is then utilized to adjudge their weathering potential. It lays the foundation for estimation of their impacts during their long evolutionary history starting from the Ordovician.

To better understand the effects of surface waves on global intraseasonal prediction, we incorporated the WW3 model into CFSv2.0. Processes of Langmuir mixing, Stokes–Coriolis force with entrainment, air–sea fluxes modified by Stokes drift, and momentum roughness length were considered. Results from two groups of 56 d experiments show that overestimated sea surface temperature, 2 m air temperature, 10 m wind, wave height, and underestimated mixed layer from the original CFSv2.0 are improved.