We summarize the results of the Dynamical Core Model Intercomparison Project's idealized supercell test case. Supercells are storm-scale weather phenomena that are a key target for next-generation, non-hydrostatic weather prediction models. We show that the dynamical cores of most global numerical models converge between approximately 1 and 0.5 km grid spacing for this test, although differences in final solution exist, particularly due to differing grid discretizations and numerical diffusion.

To assess the effect of climate on vegetation, dynamic vegetation models simulate their response e.g. to climate change. Most currently used dynamic vegetation models ignore the fact that for colonization of a new area not only do the climatic conditions have to be suitable, but seeds also need to arrive at the site to allow the species to migrate there. In this paper we are developing a novel method which allows us to simulate migration within dynamic vegetation models even at large scale.

Mountain glaciers are one of the few remaining subsystems of the global climate system for which no globally applicable community-driven model exists. Here we present the Open Global Glacier Model (OGGM; www.oggm.org), developed to provide a modular and open-source numerical model framework for simulating past and future change of any glacier in the world.

During the last 2.6 million years, the Earth's climate has cycled between cold glacials and warm interglacials, causing the growth and retreat of ice sheets. These ice sheets can be independently reconstructed using numerical models or from dated evidence that they leave behind (e.g. sediments, boulders). Here, we present a tool for comparing numerical model simulations with dated ice-sheet material. We demonstrate the utility of this tool by applying it to the last British–Irish ice sheet.

We developed a new method for combining surface ozone observations from thousands of monitoring sites worldwide with the output from multiple atmospheric chemistry models. The result is a global surface ozone distribution with greater accuracy than any single model can achieve. We focused on an ozone metric relevant to human mortality caused by long-term ozone exposure. Our method can be applied to studies that quantify the impacts of ozone on human health and mortality.

In this research we developed a time-dependent dust source map for NMME-DREAM v1.0 model based on the MODIS Normalized Digital Vegetation Index (NDVI). Areas with NDVI < 0.1 are classified as active dust sources. The new modeling system is tested for the analysis of dust particle dispersion over SW Asia using a mesoscale model grid increment of 0.1° × 0.1° km for a period of 1 year. Simulated AOD increased compared to the static dust source approach and there was an increase in dust loads.

We present a description of a coastal ocean circulation model designed to work on variable-resolution meshes made of triangular and quadrilateral cells. This hybrid mesh functionality allows for higher numerical performance and less dissipative solutions.

Regional climate experiments coordinated throughout CORDEX aim to study and provide high-quality climate data over a given region. The data are used in climate change mitigation and adaptation policy studies and by stakeholders. CORDEX requires a list of variables, most of which are not provided by atmospheric models. Aiming to help the community and to maximize the use of CORDEX exercises, we create a new module for WRF models to directly produce them by adding generic and additional ones.

A robust validation of ice sheet models is presented using LIVVkit, version 2.1. It targets ice sheet and coupled Earth system models, and handles datasets and operations that require high-performance computing and storage. We apply LIVVkit to a Greenland ice sheet simulation to show the degree to which it captures the surface mass balance. LIVVkit identifies a positive bias due to insufficient melting compared to observations that is focused largely around Greenland's southwest region.

In this article, we describe the new ECMWF seasonal forecast system, SEAS5, which replaced its predecessor in November 2017. We describe the forecast methodology used in SEAS5 and compare results from SEAS5 to results from the previous seasonal forecast system, highlighting the strengths and weaknesses of SEAS5. SEAS5 data are publicly available through the Copernicus Climate Change Service's multi-system seasonal forecast.

Predicting future changes in ecosystem services is not only highly desirable but is also becoming feasible as several forces are converging to transform ecological research into quantitative forecasting. To realize ecological forecasting, we have developed an Ecological Platform for Assimilating Data (EcoPAD) into models. EcoPAD also has the potential to become an interactive tool for resource management, stimulate citizen science in ecology, and transform environmental education.

The Polar Amplification Model Intercomparison Project (PAMIP) is an endorsed contribution to the sixth Coupled Model Intercomparison Project (CMIP6). It will investigate the causes and global consequences of polar amplification through coordinated multi-model numerical experiments. This paper documents the experimental protocol.

This paper presents Devito, a Python-based software. The aim of this software is to provide a high-level simple interface to users for the description and discretization of the mathematical definition of the physics. This research initially started as an attempt to improve research time, portability, and performance in exploration geophysics. We present the latest version of the software that is already making an impact in academics and industry.

A simple novel technique for simulating multisite occurrence of precipitation is proposed. The proposed technique employs the nonparametric approaches k-nearest neighbor and genetic algorithms. We tested this technique in various ways and proved that this simple technique can be useful and comparable to the existing one.

Computer simulations of atmospheric chemistry are a central tool to study the impact of air pollutants on the environment. These models are highly complex and require a lot of computing resources. In this study we show that machine learning can be used to predict air pollution with an accuracy that is comparable to the traditional, computationally expensive method. Such a machine-learning-based model has the potential to be orders of magnitude faster.

Two developments are made to the United Kingdom Chemistry and Aerosols (UKCA) model to improve simulation of stratospheric ozone. The first is the addition of a solar cycle. The influence on ozone from the solar cycle is found to be 1–2 %, which is consistent with other studies. The second is to the heterogeneous chemistry, the most significant change being the addition of reactions involving bromine species. This was shown to reduce ozone biases relative to observations in most regions.

An operational multi-model forecasting system for air quality is providing daily forecasts of ozone, nitrogen oxides, and particulate matter for 37 urban areas of China. The paper presents the evaluation of the different forecasts performed during the first year of operation.