Workshop on surface process coupling and its interactions with the atmosphere

The Earth System Modelling at the Weather scale (ESM-W) project, a collaboration between the German Weather Service (DWD) and GeoInfoDienst BW, aims to develop a coupled ocean-atmosphere forecasting system. This system integrates the ICON-O ocean model and the ICON-NWP atmospheric model for the NWP time scales, aligning with the ICON-Seamless development goal of a time scale independent, integrated modelling framework for both weather and climate applications.
We present a case study of Hurricane Fiona from September 2022 as an initial test for the coupled model, incorporating weakly coupled data assimilation. Hurricanes significantly im-pact both the atmosphere and the ocean, leading to noticeable changes in various physical quantities. Our primary focus is on comparing the modelled storm track with the real-world observations and analysing additional physical quantities to assess the model's performance. Additionally, we explore how the hurricane's behaviour is influenced by different technical configurations of the model. This poster presents our findings and outlines future development directions.

YAC (Yet Another Coupler) is a coupling library, which provides 2D interpolation for all grid types commonly used in coupled climate and weather simulations. Users can choose from a variety of different interpolation methods covering a diverse range of requirements. Its highly scalable online weight computation capabilities have proven to be able to handle simulations using globally defined km-scale grids with hundreds of millions of cells with ease. This presentation is an introduction to YAC and its abilities.

Machine learning models have emerged as powerful tools for simulating Earth system processes. Following their successful application in capturing atmospheric evolution for medium-range weather forecasts, attention has increasingly shifted towards other components of the Earth system, such as the marine and land environments. This interest is further driven by the potential to enhance forecasting capabilities beyond the medium-range. Machine learning frameworks offer remarkable flexibility in integrating these model components to achieve a coherent Earth system representation. At one end of the spectrum, model components can be trained jointly within a unified framework, optimized using a shared loss function. At the other end, components may be developed independently and coupled by exchanging physically relevant information at multiple interfaces, mirroring the traditional coupling strategies employed in numerical models. In this presentation, we will examine the advantages and challenges of these approaches, with a particular emphasis on coupling the atmospheric and marine components within the deterministic AIFS model, the machine learning-based forecast system developed at ECMWF. Furthermore, we will compare the coupling strategies of data-driven models with those of traditional numerical models, highlighting their respective strengths and limitations.

The goal of this project is to develop a Limited Area Mode (LAM) for the ICON-WAVES model, which is a important component of the "Earth System Model on the Weather Scale" (ESM-W) initiative, in collaboration with the GeoInfoDienst BW. LAM provides high-resolution weather forecasts for specific regions, offering a more detailed simulation compared to global weather prediction models, which cover the entire Earth's atmosphere. ICON-WAVES regionalization aims to enhance wave modeling capabilities, enabling more accurate and higher-resolution wave predictions for localized areas. This work focuses on coupling the ICON-WAVES system with atmospheric data to improve the precision of regional wave forecasts.

This study investigates and quantifies the relative contributions of shallow convective clouds, in relation to clear air entrainment and rainforest assimilation on the diurnal cycle and vertical distribution of atmospheric CO2 in the lower tropical troposphere. Using data from the comprehensive CloudRoots-Amazon22 campaign that ranges from the stomatal to cloud scales, we develop and validate a representative shallow convective numerical experiment above the Amazon with the turbulence-resolving Dutch Atmospheric Large Eddy Simulation model. The numerical experiments includes a bulk rainforest representation of photosynthesis and stomatal aperture at the canopy top. We assess the role of shallow convection through formulating the vertically integrated, domain-averaged CO2-budget, distinguishing between surface, cloud and environmental contributions. To isolate the role of shallow convective clouds, we perform an identical experiment excluding the dynamic effects of clouds. Our findings show that shallow convective clouds actively ventilate CO2 from the boundary layer at twice the rate of photosynthesis uptake. Shallow convective clouds transport CO2 from the surface to heights nearly twice the boundary layer height, significantly affecting its vertical distribution until late afternoon. Consequently, we can distinguish a three-regime diurnal cycle: entrainment-diluting, cloud-ventilation and CO2-assimilation. Without the presence of dynamically active shallow convective clouds, cloud-ventilation is absent, and entrainment alone drives the CO2 diurnal evolution and vertical distribution. Additionally, the mixing and transport by shallow convective clouds promote a vertically extended negative (< -0.3) to positively (> 0.8) correlated signal between the movement in CO2 and H2O from the surface upwards, linking the carbon and hydrological cycles. This research highlights key processes essential for accurately representing the lower tropical tropospheric CO2-budget including the role of shallow convection.

The present work discusses the results of applying a roughness sublayer parameterization to the land surface model SURFEX, which is extensively used both offline and coupled with different NWP atmospheric models and LES. By adding increased physical details in the classical similarity coupling scheme over a vegetated surface, the present work improves the surface-atmosphere coupling (the exchange of momentum and energy) in the NWP system above tall vegetation. However, it also reveals a few important questions (e.g. counter-gradient flow, effects of inaccurate physiography data, the coupling thickness) that deserve additional treatment.

The Met Office uses coupled atmosphere - ocean - sea-ice models across timescales from shortrange NWP to centennial climate predictions. Coupled NWP was the last system to go operational in May 2022.

In this presentation we will give an overview of the coupling methodology used in Met Office models. We will discuss the challenges of developing a model for coupled NWP and discuss the advantages of the seamless modelling approach. Finally, we will discuss some possible future developments for coupled NWP.

Several operational centres are either already using or planning to use a coupled modelling system with atmosphere, ocean, sea-ice and wave components for their numerical weather predictions (NWP). ECMWF implemented ocean/sea-ice coupling to all forecasting systems as part of the upgrade to CY45R1 on June 5 2018.
Experiences with ocean coupling for NWP time scales leading up to the operational implementation and lessons learned since then will be presented followed by an outlook of future developments.

The famous Craik-Leibovich eqautions are often used to incorporate sea state impacts in ocean circulation models. However, including the Stokes drift, in such phase-averaged equations for the Eulerian mean motion leads to an Eulerian energy budget which is physically difficult to interpret. Here, we show that a Lagrangian energy budget allows for a closed energy budget, in which all terms connecting the different energy compartments correspond to well known energy transfer terms. The Lagrangian energy budget is used to discuss an energetically consistent framework which can be used to couple a general circulation ocean model to a surface wave model. In this framework, the energy provided by the surface wave model is split-up into energy driving oceanic mean motions, inertial oscillations and turbulence in a consistent manner.

I will summarise recent progress on surface exchange over sea ice with a focus on using observations of turbulent fluxes of momentum, sensible and latent heat to constrain and develop parameterizations of surface exchange for numerical weather and climate prediction models. I will cover a recent parameterization development for scalar exchange. Plus the impacts of these new parameterizations on the atmosphere in operational weather forecasting models and state-of-the-art climate models.

Accurate tropical cyclone (TC) intensity forecasts remain a significant challenge due to the inadequate representation of upper-ocean dynamics and air-sea interaction processes. To address this issue, a regional Atmosphere-Ocean-Wave coupled model was developed, incorporating key ocean surface wave-related processes, including wave modulation of air-sea momentum flux, the sea spray effect on enthalpy flux, and non-breaking wave-induced mixing. Validated against airborne observations, our results demonstrate that ocean surface waves improve the representation of TC structure, size, and intensity. Furthermore, retrospective simulations of 21 TCs from 2013 were conducted to statistically assess the impacts of wave processes on TC simulations, showing a significant reduction in TC intensity simulation bias. This study highlights the critical role of ocean surface wave processes in modulating TC dynamics and improving intensity forecasts.

With the changing conditions in the Arctic, the need for improved forecasting of the meteorological and environmental conditions are increasing. Variability in surface conditions, especially over the ocean and sea ice, increases the uncertainties in heat, moisture and momentum fluxes between the atmosphere, ice and ocean, especially close to the marginal ice zone. These uncertainties increase the potential risks for operating in these harsh areas, both for aircraft and vessels.
Following the success of the Canadian Arctic Prediction System (CAPS) implemented at Environment and Climate Change Canada (ECCC) during the Year of Polar Prediction (YOPP), a second version of the Canadian Arctic Prediction System (CAPS) is currently being developed to address the growing need for accurate meteorological and environmental predictions in the pan-Arctic region. The GEM NWP model in CAPS covers the entire Arctic Ocean, from Bering Strait to south of Iceland (~60oN) at 3km resolution. The NEMO-CICE ice-ocean component follows the Regional Ice-Ocean Prediction System (RIOPS) covering Canada’s three oceans on a 1/12o ORCA grid. This presentation evaluates CAPS short-term 48h forecasts with a specific focus on the impact of coupling to the ice-ocean over and near the marginal ice zones.

This paper examines soil-moisture-atmosphere coupling "hotspots" in the Northern Hemisphere during the summer months and evaluates their representation in seasonal forecasts. Using hindcasts from the Copernicus Climate Change Service (C3S) multi-model seasonal forecast system the study explores the predictability of land-atmosphere interactions based on soil moisture anomalies. The results show that regions with strong soil-moisture-atmosphere coupling exhibit considerable potential for predicting seasonal temperature and precipitation patterns a season or more in advance. However, significant uncertainty exists in estimates of the soil-moisture initial conditions and soil-moisture persistence timescales. While some regions show realistic coupling, others, including the Central USA and Eastern Europe, display exaggerated coupling, leading to errors in temperature and precipitation forecasts. This study underscores the potential for predicting the atmosphere in summer based on memory of soil-moisture initial conditions whilst highlighting areas for further forecast system improvement.

Atlas is a library for Numerical Weather Prediction and Climate modelling developed at ECMWF which provides parallel distributed data structures for meshes and fields. Atlas has built-in support for various regional grids as well as global grids used in the community, such as the Gaussian grids used by the Integrated Forecasting System (IFS), ORCA grids used by ocean model NEMO, HEALPix grids, cubed sphere grids, and unstructured grids. Other grids can be supported via a plugin mechanism. Atlas provides domain decomposition strategies for meshes and fields and provides halo-exchange and gather-scatter parallel communication patterns. Atlas further contains several interpolation algorithms to enable remapping of fields defined in different meshes in a parallel distributed context. Currently following interpolation methods are supported: unstructured-linear, structured-linear, structured-cubic, structured-quasicubic, nearest-neighbour, k-nearest-neighbour, conservative-order-1 and conservative-order-2.
For parallel interpolation, the domain decomposition of the target grid must match that of the source grid. A new development started recently to enable different independent domain decompositions for each field, so that there is more flexibility in remapping fields defined in Earth System Model components with non-matching domain decompositions. This is e.g. the case for IFS ocean component using NEMO or FESOM2 and the IFS spectral atmosphere component. In this presentation we will present the current state of the art on remapping between the ocean surface and the atmosphere using Atlas, and the future plans on integration within the IFS model.

Recent improvements in the description of the surface-atmosphere interactions in the ICON atmospheric model of the German Meteorological Service are presented. The continental surfaces, including vegetation cover, represent an important component of the earth’s climate system. On the one hand, they are the habitat of humanity, which makes it important to try to understand the governing processes and living conditions at the land surface and how they may evolve in the future. On the other hand, from the point of view of atmospheric sciences, the land surface and biosphere interact with the lower atmosphere, and they have a significant impact on near-surface meteorological and climatological phenomena. During the recent years, the formulations of several processes in the ICON multi-layer land surface scheme TERRA were either deeply revised, or replaced by more realistic versions, for instance the formulation of the bare soil evaporation. A very persistent problem in atmospheric models is the computation of the surface temperature, in particular over vegetation. In TERRA, this was addressed by a simplified canopy approach, inspired by ECMWF’s IFS skin temperature formulation, embedded in the tile approach of ICON. Furthermore, with the trend of an increasing resolution of atmospheric models for numerical weather prediction or climate simulations, more fine-scale processes at the land surface can be resolved. Consequently, an urban canopy parameterization was developed and implemented in ICON, heading for km-scale or even hectometric scale applications. Beside further developing the physical parameterizations and using them with well-tuned, but static, parameter values, another possibility to improve the model skill is to apply an adaptive parameter tuning, based on the information from the data assimilation system. Meanwhile, this method is applied to several ICON land surface parameters. It is demonstrated that all these measures, and others, improve the interactions and feedbacks between land surface and atmosphere in ICON.

Windstorms associated with extratropical cyclones are destructive natural hazards. At the Laboratoire d’Aérologie (LAERO in Toulouse, France) we are interested in elucidating the processes involved in the formation of near-surface extreme winds and my focus is on wave and wave breaking related processes. Though crucial for their societal impact, these processes are not well understood and too small scale to be explicitly represented in numerical weather prediction models. Waves modulate air-sea exchanges, mix the upper ocean, and inject sea spray into the atmosphere when breaking. Air-sea fluxes of enthalpy and momentum greatly influence the dynamics of the marine atmospheric boundary layer (MABL). Waves increase the surface roughness but sea spray loading may act as a buffer layer reducing drag and stabilizing the MABL. Larger droplets increase air-sea enthalpy and decrease momentum transfers thus promoting the intensification of tropical cyclones, but what of extra tropical cyclones?

The role of waves and the ocean in extra tropical cyclones surface wind extremes will be illustrated by two case studies : 1) the North Atlantic storm Alex where wave coupling influences mesoscale jets and the downward momentum transport, and 2) the cold wake producing medicane (Mediterranean Hurricane) Ianos where the ocean induces a negative feedback similar to that seen in certain tropical cyclones. These were simulated using the coupled framework which consists of the atmospheric model Meso-NH, the 3rd generation wave model WAVEWATCH III®, and the oceanic model CROCO. As computing advances, such as GPU porting of Meso-NH, allow realistic storm scale LES-type simulations, we need to question 1) What do we gain from kilometer to sub kilometer scale coupling ? and 2) How should coupling strategies and air-sea parameterizations be adapted depending on resolution ?

Anomalous tidal waves triggered by atmospheric disturbances, occurring at frequencies similar to tsunamis, are called meteorological tsunamis or meteo-tsunamis. These waves typically form when atmospheric disturbances, often caused by mesoscale phenomena such as thunderstorms, atmospheric gravity waves, squalls, and cyclones, induce rapid changes in mean sea level pressure. Even small fluctuations can generate sea level oscillations of several centimetres. Typically, 1 mbar of pressure disturbance can cause 1cm sea surface change. When the speed of the atmospheric disturbance closely aligns with the wave speed (approximately equal to the phase speed, √gh), continuous amplification of the water level occurs through a mechanism known as Proudman resonance (Proudman, 1929; Renzi et al., 2023).
Although research on meteotsunamis in the US and Mediterranean is advanced, studies in the UK remain limited. This is largely because meteotsunamis are considered rare and of minimal risk in the UK (Lewis, 2023). However, recent events suggest that these phenomena are often missed by observations, as the frequency of tide gauge recordings is typically insufficient to detect them. Furthermore, feedback from the UK Environmental Agency (EA) during such events indicates that meteotsunamis can cause significant damage if actions are not taken.
This study explores the understanding and prediction of meteotsunami events using the atmosphere-ocean-wave (AOW) coupled model, which integrates the Unified Model (UM), NEMO, and WaveWatchIII, developed at the Met Office. The model was configured with a 10-minute coupling frequency to evaluate its capability for capturing meteotsunamis. We analyse two events: one associated with a slow-moving frontal system, and the other with a mesoscale convective system. Analysis of the first event on 31 October 2021, recorded by the Portsmouth tide gauge, demonstrates that the 10-minute coupling successfully captures the event, whereas the previously used 1-hour coupling does not. The model can also capture the seiche in the Channel.

Increasing the complexity of both weather and climate forecasting modelling systems allows better prediction skills in coastal areas, where equilibrium assumptions between Earth system components break down. It also allows better consistence between earth system components, enabling multi-hazard forecasting. We present recent advances in the regional coupled environmental prediction system developed in the UK through the release of the Regional Coupled Suite – UKC4 configuration. The coupled system runs at km-scale over the northwest European shelf region, coupling a convection-permitting atmosphere with land, surface waves and shelf-enabled ocean models. The new UKC4 configuration is an upgrade to all model components, it includes a new Regional Atmosphere and Land configuration (RAL3.3), an online diagnostic of rivers, an option for higher frequency coupling, additional coupling to a biogeochemistry model, the possibility of running near-real time ensemble forecasts and to run climate hindcasts.

The new atmospheric configuration leads to a beneficial increase in shortwave radiation reaching the ocean in summer months and a beneficial reduction in wind speed over sea, which is further reduced with wave coupling. RCS-UKC4 has good skills in terms of ensemble wave forecasts during storms compared to the current Met Office operational wave ensemble because it captures tidal current/wave/wind interactions. We show coupling either inflates or deflates the ensemble spread in screen temperature depending on whether latent heat flux or radiative heat flux dominates the spread in near-surface fluxes. Finally, we show an example of use of the coupled system for increasing our understanding of regional marine heatwaves and their impacts on a multi-hazard storm.

Ocean surface waves can significantly influence the wind within the wave boundary layer, especially under swell conditions. However, current models fail to capture the wind profile in this layer, such as the swell-induced low-level wind jet. In this study, we propose a new turbulence closure model to estimate wind stress in the wave boundary layer by separately considering viscous stress, shear-induced turbulent stress, wind-sea induced stress, and swell-induced upward stress. The model also accounts for the misalignment between wind stress and wind direction. Single-column simulations indicate that (a) swell-induced upward momentum flux increases surface wind speed and alters wind direction, (b) the misalignment between upward momentum flux and wind has a more significant impact on the wind profile than downward momentum flux, and (c) the impact of swell-induced upward momentum flux decreases with atmospheric convection. The proposed closure scheme was implemented into an atmosphere-wave coupled model.

The fractional or percentage whitecap coverage of the ocean surface (W) is often parameterised in terms of wind speed. Datasets of W typically show order of magnitude scatter at a given wind speed value. Here we compare modelled values of W to measured values from the North Atlantic Ocean. The modelled W is forced by the spectrally-integrated whitecap dissipation source function in the European Centre for Medium-Range Weather Forecasts spectral wave model, ecWAM. Without tuning, best agreement is found for mature sea states, with an average modelled to measured
ratio of 0.87. This ratio approaches unity with the introduction of a dissipation rate threshold value and an explicit wave-age dependence. The study suggests that accurate estimates of W can be routinely produced by ecWAM and opens new opportunities to model bubble-mediated fluxes of CO2 and sea spray aerosol with ecWAM.

Sofar Ocean has developed a coupled atmosphere-wave-ocean forecast system to improve forecast skill of surface winds and waves over the ocean. Using data from ECMWF and modeling components from NOAA, along with the U.S. Earth System Modeling Framework (ESMF) National Unified Operational Prediction Capability (NUOPC) coupling software, Sofar runs this forecast system containerized and on the cloud using virtual HPC.

Initial results indicate that improvements to the atmospheric model planetary boundary layer (PBL) scheme, coupled interactions with the ocean mixed layer, improved representation of interactions between the atmosphere and wave components, and influence of surface currents on waves all help to improve the skill of wind and waves forecasts over the ocean surface. We also apply a Bayesian Optimization Machine Learning approach to calibrate key coupled model parameters to improve the fit of forecasts with Sofar's global network of Spotter buoys, which collect measurements of directional wave spectra that are processed to provide wind and wave observations.

Vegetation plays a fundamental role in shaping Earth's energy, water, and carbon cycles. Plants consume water resources through transpiration and interception, regulate atmospheric carbon dioxide concentration, alter surface roughness, and determine both net radiation and its partitioning. This influence propagates through the atmosphere, from micro-climate scales to the atmospheric boundary layer, subsequently impacting meso-scale and large-scale circulation, as well as the planetary transport of heat and moisture. Understanding biophysical feedbacks across scales is crucial for predicting hydro-climatic extremes, such as droughts or heatwaves, which are believed to be jointly exacerbating due to these feedbacks. While this finding is concerning, it also opens the door to improving sub-seasonal weather forecasts and leveraging land geoengineering as a strategy to mitigate extremes occurrence.

This presentation will explore biophysical feedbacks focusing on extreme events, particularly droughts and heatwaves, but also investigating their role over climatological scales. Key questions will be addressed: How do extreme meteorological conditions impact ecosystem evaporation? In what ways does vegetation regulate the atmospheric boundary layer, affecting the intensification and propagation of these extremes? How do these biophysical feedbacks enhance the inflow of heat and reduce the inflow of moisture to downwind regions, potentially leading to the propagation of extreme events, and even causing dryland expansion at longer time scales? What are the consequences of land feedbacks for human heat stress during compound drought–heatwaves? How can information on land conditions be used for the timely prediction of these events? The goal of this presentation is not to provide definitive answers to these questions, but rather to concentrate on recent results from my team's work that may help advance our collective understanding of biophysical feedbacks and the role they play in climate.

Earth system models (ESMs) couple many submodels in time and space. We focus on atmosphere-ocean-sea ice coupling. Standard coupling approaches in state-of-the-art ESMs can be classified as the first iteration of so-called Schwarz waveform relaxation (SWR) algorithms. Not iterating is computationally cheap but produces a numerical coupling error in time. With an SWR algorithm, the submodels exchange time-dependent boundary data, successively updated from one iteration to the next. In the converged limit, one obtains a consistent numerical solution to the underlying coupled problem. We use SWR to produce reference solutions and quantify the coupling error of standard coupling algorithms. Past studies demonstrated that the SWR solution eliminates phase errors, reduces ensemble spread, and can indicate whether current coupling setups are mathematically consistent.

Since the boundary conditions at the air-sea interface are part of vertical physics parameterizations, we study the coupling error in a coupled single column model, the EC-Earth AOSCM. This simulates a single vertical column of the atmosphere and ocean, including sea ice. The EC-Earth AOSCM uses the same set of physics parameterizations as its host model, EC-Earth. We have implemented an SWR algorithm based on the OASIS3-MCT coupler, treating the other model components as black boxes. This allows us to compute the coupling error of multi-day simulations. Our results show that the AOSCM is sensitive to changes in the coupling algorithm, particularly with respect to the vertical turbulence parameterization and when sea ice is present. We discuss how this behavior is affected by the major version change from EC-Earth 3 to EC-Earth 4. Finally, we show some new theoretical results on analytically describing atmosphere-ocean-sea ice coupling.

Bare ground evaporation has a strong impact on near-surface weather and soil moisture forecasts. In North America, this is especially true during the transition seasons, because vegetation is less present and less active.

Environment and Climate Change Canada (ECCC) currently relies on the Soil, Vegetation and Snow (SVS) land-surface scheme for surface and river prediction, using an offline system, but uses a legacy version of the Interaction Sol-Biosphère-Atmosphère (ISBA) for weather forecasting. SVS evolved from this version of ISBA. It includes, among other improvements, a multiple energy balance and a multi-layer soil moisture scheme.

As part of an initiative aimed at implementing SVS in its short-term, high-resolution numerical weather prediction system, ECCC is revisiting the parameterization used for bare ground evaporation. Currently, both SVS and ISBA rely on the “alpha” method, in which relative humidity of the surface is parameterized as a function of surface soil moisture, using a relationship proposed by Jacquemin and Noilhan (1990). Based on a literature review, we compared this existing approach to the “beta” method of Lee and Pielke (1992) and to the soil resistance method of Albergel et al. (2012), which is used in the ecLand model. Objective verification showed that the soil resistance method performed better for wetter soil conditions, whereas the “beta” method performed better for dryer soil conditions. A new hybrid approach was thus developed and implemented. It behaves like the “beta” method in dry conditions, and like the soil resistance method for wet conditions, leading to significant improvements to weather forecasts in both conditions.

Although this new hybrid method improves weather forecasts in our specific context, it remains empirically based and may require tuning for a specific land-surface and atmospheric model combination. The sensitivity of the land-surface and atmospheric model predictions to these parameters is thus explored in this presentation.

With coupled environmental NWP systems, the traditional approach of developing specific schemes or components in parallel with other developments before integrating them together has reached its limits. Since models’ equilibrium are highly dependent on the existing balance of parameterizations, many centers have identified the new added complexity, or even limits, of advancing model developments without degrading results in the whole system. Environment and Climate Change Canada (ECCC) has experienced the same limits when trying to integrate a major upgrade in the land-surface scheme and assimilation together with a modernization of the atmospheric physical parameterizations. Since the connections between the land-surface and lower atmosphere are rapid and highly coupled, a new collaborative approach was deemed necessary to improve both the land-surface and the atmospheric physical parameterizations altogether.
A new coordinated effort is now in place for the high-resolution deterministic prediction system (HRDPS) at ECCC with the aim of replacing the current land-surface model with the Soil and Vegetation Scheme (SVS) and to upgrading the Canadian Land Data Assimilation System (CalDAS) to integrate satellite remote-sensing while modernizing the atmospheric physics parameterizations.
This project includes researchers from all groups, and necessary steps include revisiting the fundamentals behind parameterizations of both surface and atmospheric processes. It helps strengthen the understanding between the groups while validating the coherence between the physical schemes. More in-depth validation techniques are also evolving since the system is looked at from different angles, particularly for surface-atmosphere interactions. Promising results and examples of strategies to tackle specific surface-atmosphere interactions will be presented.

Subseasonal forecasts rely on accurate representations of land surface processes in general and of vegetation dynamics in particular to improve temperature predictability through land-atmosphere interactions. This work examines the potential of integrating novel Earth observations and data-driven approaches to enhance subseasonal forecast skill.
Results demonstrate that soil moisture and vegetation play a key role in explaining temperature forecast errors due to their persistent anomalies, which influence evaporative cooling and energy fluxes. Relatively high forecast errors are found in regions with strong land-atmosphere coupling, suggesting deficiencies in model data assimilation and/or process representation. Incorporating up-to-date land cover data and near-real-time vegetation indices, such as leaf area index, improves simulations of water and energy fluxes. Thereby, enhanced model performance is dependent on re-calibration which yields even better results when performed locally rather than globally. Finally, using a machine-learning approach, we demonstrate the value of novel Earth observations so far not considered in data assimilation for informing weather forecasts. Using a flexible machine learning approach we show that including data on Enhanced Vegetation Index, Sun-Induced Fluorescence, and Land Surface Temperature, in addition to traditional initial condition information, we achieve a 6–7% reduction in temperature forecast errors, especially at short-to-medium lead times.
Overall, our findings highlight the critical role of land surface dynamics in subseasonal temperature forecasts and emphasize the potential of combining novel Earth observations and advanced modeling techniques to improve prediction accuracy.