The UK’s new RCS-UKC4 (Regional Coupled Suite – UK Coupled domain version 4) system combines atmosphere, ocean, waves, land, rivers, and biogeochemistry models to improve coastal weather and climate predictions. It offers better storm wave predictions, more accurate river flows, and captures rapid sea-level changes. These advances help predict multiple hazards more reliably, supporting safer communities and helping better planning.

AdaPBS (Adaptive Particle Batch Smoother) is a new algorithm to combine observations with cryospheric numerical models. AdaPBS is an iterative algorithm that automatically adjusts computing effort to the task, allowing the implementation of early stopping strategies. We tested AdaPBS at multiple sites with different models, matching or outperforming standard methods, when compared against more complex (computationally expensive) algorithms.

Based on the operational Global Ensemble Forecast System-Aerosols at the National Centers for Environmental Prediction, we developed an upgraded system using the Common Community Physics Package framework that allows aerosol particles to directly influence radiation and cloud formation, including how precipitation removes particles from the atmosphere. Evaluation against observations and reanalysis data demonstrates improved forecast skill for weather and subseasonal prediction.

Accurate operational forecasts of temperature and wind in the coastal marine boundary layer are important for a wide range of applications. Leveraging data that were collected along the United States northeast coast during a multi-year period for the Third Wind Forecast Improvement project, we investigated the performance of the operational forecast model and identified systematic errors in wind and temperature forecasts that are now being addressed by the model developers.

Soil organic matter models are characterised by equifinality, the phenomenon whereby multiple parameter sets yield similar results. This study shows that the number of identifiable parameters that can be optimised together is limited, even under data-rich conditions. As a result, overparameterised models show a large variability when simulating future changes. Optimising only identifiable model parameters is therefore necessary to avoid this hidden uncertainty in soil organic matter models.

The traditional weather prediction models improve slowly, while machine learning models struggle with extreme weather and fine details. To address these gaps, we developed an online  Spectral Nudging-based correction system that leverages a machine learning model's skillful large-scale circulation to guide a physical model. This hybrid model enhances large-scale skill while preserving small-scale features, providing a viable pathway for improving operational weather forecasting.

Clouds constitute a key uncertainty for climate change projections. The Cloud Feedback Model Intercomparison Project (CFMIP) aims to address this challenge by evaluating and understanding clouds and their impacts on atmospheric circulation, precipitation, and climate sensitivity. The present paper describes the CFMIP experiment protocol for the Coupled Model Intercomparison Project phase 7 (CMIP7), and discusses the accompanying science questions and opportunities for progress.

This study introduces a novel approach to incorporating city-specific EO (Earth Observation) data into an urban canopy model named BEP-BEM (Building Effect Parameterization-Building Energy Model) integrated in WRF (Weather Research and Forecasting). We highlight the added value given by BEP-BEM in the representation of the urban processes compared to a slab urban bulk approach. The implementation of the EO data improved specific aspects of the model performance.

ICONEval is a new framework to facilitate researchers to continuously monitor and check their climate models during the development phase. It builds on the ESMValTool software package to run tests verifying that the results are plausible, follow physical laws, and that the model has sufficient skill in reproducing the observed climate. An important aim is to make it easier to spot model errors early, for example when implementing new model components that use machine learning.

This study proposes a comprehensive uncertainty quantification framework that jointly evaluates data, model, and prediction uncertainties in deep learning-based mineral prospectivity mapping. By modelling and visualizing both data and model uncertainties, the framework transforms deep learning-based mineral prospectivity mapping from deterministic prediction to probabilistic decision-making, thereby enabling more reliable and trustworthy mineral exploration.

The cryosphere, including snow, ice, and melt ponds, strongly influences how sunlight is reflected and absorbed, affecting climate change. We developed a new version of radiative transfer model SCIATRAN, to better represent these complex surfaces and their interaction with the atmosphere. The model was validated using real measurements and showed good accuracy. These advances help scientists better understand climate processes in polar regions and improve satellite observations.

This paper uses the Neural Physics approach to determine parameters of a simple land-surface model. We show that we can only obtain a reliable parameter estimation using soil temperature measurements at more than one depth, and that latent and sensible heat fluxes cannot be differentiated. We then apply the inverse model to real urban flux tower data and show that parameters, as well as various heat fluxes, can be reliably estimated using an observed value for the effective surface albedo.

We developed a new way to model how planetary surfaces rise and sink as the deep interior slowly flows. Existing approaches are either costly or unstable. Our method represents the surface smoothly within a fixed grid, which avoids artificial air layers and numerical problems. Tests show it matches established results while running faster and working in more realistic settings, such as loaded surfaces and global models. This makes simulations of surface evolution more reliable and accessible.

The is a study of the plumes that rise from smokestacks. Knowing how these plume behave helps predict downwind pollutant concentrations. We use photos over a 2-year period to investigate how these plumes rise under different conditions and compare this to a commonly used model parameterization. It is found that the equations used to model plume rise in current models do well for some condition, but these equations can over-predict the plume rise, typically during the day when it is hot.

This study presents a new way to model how Earth’s surface changes over time as the deep interior moves. The method tracks the surface directly, allowing clearer and more detailed results worldwide while using less computing power. It improves accuracy compared to existing approaches and makes it easier to connect deep Earth processes with oceans, climate, landscapes, and life through time.

Accurate daily climate maps are essential for environmental monitoring, yet many regions lack dense weather-station networks. We developed a deep-learning method that converts sparse station measurements into high-resolution gridded temperature and precipitation fields over complex terrain. Tested over Romania, it outperforms traditional geostatistical interpolation by learning how topography shapes local climate, while providing reliable uncertainty estimates for each prediction.

Satellite measurements of sea surface temperature are often blocked by clouds, leaving large gaps in the data used to study the ocean and climate. We developed a new computer model that fills in these missing areas while also estimating how uncertain the reconstruction is. Tests on several ocean regions show more accurate results and more realistic patterns, helping scientists better analyze and model ocean conditions.

Ice thickness models based on laminar flow theory often rely on conventional assumptions regarding basal sliding parameterization when studying alpine glaciers. This paper presents the Ice Thickness Model considering Sliding Law (ITMSL) model, which integrates a basal sliding law with laminar flow equation, with the objective of simulating basal sliding to enhance the accuracy of ice thickness inversion.

We developed a strongly simplified mechanism for isoprene oxidation that reduces the computational cost of atmospheric chemistry while retaining the key processes controlling formaldehyde formation. Its parameters were optimized using box-model simulations, and the mechanism reproduces the full chemistry well in global simulations. It also gives similar isoprene emissions when constrained by satellite formaldehyde observations, enabling faster large-scale simulations and emission inversions.

The Coupled Model Intercomparison Project (CMIP) is a large collaborative project to better understand the Earth’s climate system. The data produced through this project is downloaded by users around the world. In this paper, we analyze the patterns of downloads and the usage of this massive dataset. From this analysis, we make some recommendations for future data production and usage tracking.

This study presents a new 12 km CORDEX (Coordinated Regional Downscaling Experiment )-CMIP6 (Coupled Model Intercomparison Project) WRF (Weather Research and Forecasting) simulation for North America, compared against an earlier 25 km CMIP5 version and a 4 km convection-permitting simulation. The updated run reduces temperature and precipitation biases, improves precipitation timing, and better captures tropical cyclone intensity. Extreme precipitation at 12 km closely matches convection-permitting results, offering a practical resolution for regional climate assessment.

Computer models used to simulate forest canopy microclimates rely on radiative transfer simulations. These models commonly only consider tree leaves and exclude wood structures. Here we present modifications to a commonly used model to consider stem absorption. We compared model estimates to detailed simulations using 3D data from ground lidar at four deciduous forest sites. We also compared simulations against field observations of sunlight reflected by canopies.

Accurate winter road surface temperature prediction prevents icy‑road accidents. Existing methods either need rare pavement parameters or miss local weather patterns and long‑term trends. This study combines two deep‑learning models to analyse four winters of Jiangsu data. It achieves lower prediction errors in general and outperforms common models at 1, 3  and 6 h horizons. Station‑based features beat external reanalysis data.

High resolution models of the earth system exhibit some disagreement and uncertainty on future methane emissions from natural sources, in particular wetlands, with some studies predicting wetlands alone could be very significant sources over the 21st century. Modelling these emissions as a response to global temperature is one option for simple models to approximate the climate impact of wetlands. The effect is a small increase in overall temperatures and a widening of the uncertainty range.

We present a three-stream radiative transfer model within the NEMO (Nucleus for European Modelling of the Ocean) framework and a use case for the Black Sea. The model is able to simulate in-water irradiance and sea surface reflectance in a wide spectral range. When coupled with an ecosystem model, the simulated irradiances are used to compute water temperature and primary production. A stochastic version of this model is also proposed to inform on uncertainties in the optical properties of seawater.

Radar altimeters measure ice-sheet elevation and sea-ice thickness. To improve their accuracy, we developed a model that simulates altimeter waveforms based on the physical properties of snow and sea ice. It computes the power, time, and Doppler shift of radar echoes as waves travel to the surface, interact with snow and ice, and return to the satellite. The model was verified internally, validated against other models and applied in Antarctica using in-situ snow measurements.

This paper describes simulations of the present climate using the NASA Goddard Institute for Space Studies Earth System model. Major highlight is the capability to include complex interactions of the different components of the Earth System, including biology, and compare well with a wide variety of observations. We demonstrate that this model can be used to address present, past and future changes in the Earth system and their impacts and interactions with ecosystems.

It is challenging to assimilate land surface temperature (LST) owing to its fast temporally varying nature. This study proposes a scheme by jointly updating the soil temperature and soil moisture. Results show marginal enhancement in LST, yet soil temperature bias over Northeast Asia (NA) drops sharply. Snow temperature and snow depth over NA, and soil moisture in the humid tropics also improve significantly. These consistent improvements demonstrate the effectiveness of the proposed scheme.

The spatial variability in the land surface properties is often not captured by the resolution of land surface models. To overcome this limitation, most models subdivide the grid cells into fractions with homogeneous characteristics, for which the land processes are calculated separately. In reality, the fractions interact via the lateral exchange of water and heat, and the present manuscript details an approach to include these fluxes in the land component of the ICON modeling framework.

This study represents the first application of CorrDiff (Corrective Diffusion), a diffusion-based model, to achieve 3 km resolution statistical downscaling over China by integrating various variables at multiple levels. Through structural improvements to CorrDiff, our methods produce forecasts that outperform the China Meteorological Administration Mesoscale Model for almost all target variables. Our results show that for radar reflectivity prediction, CorrDiff yields more realistic outputs than regression models.