Philip Stier

Abstract:

The top-of-atmosphere net radiative effect of convective anvils is estimated to be close to zero and arises from a balance of significant short-wave cooling and long-wave warming over a complex diurnal cycle. When anvils are optically thick, the cooling due to daytime scattering of shortwave solar radiation dominates. In contrast, optically thin anvils have weaker scattering of solar radiation, so longwave warming becomes the dominant effect. Hence, it is essential to understand the controls of anvil radiative properties over the convective lifecycle, which arises from a complex interplay of convective cloud dynamics and microphysics. The convective mass flux modulates anvil extent, and changes in ice crystal size and morphology affect anvil lifetime and radiative properties. Convective anvils have been proposed to respond to global warming (cloud feedbacks) and anthropogenic aerosols (aerosol-cloud interactions). However, the associated uncertainties remain large and key relevant processes are not represented in the current generation of climate models. Emerging kilometre-scale climate models present new opportunities to examine these effects at the process level.In this work we bring together multiple research strands to quantify the controls of convective anvil clouds and associated radiative effects over the convective lifecycle towards understanding its sensitivity to climate and air pollution changes. We use the tobac cloud tracking framework to track convective cores and associated anvils in 4D across regional and global km-scale ICON model simulations which allows us to quantify the link between convective mass flux, anvil extent and anvil radiative properties. We apply this framework to regional high-resolution simulation of ICON coupled to HAM-lite, our reduced complexity aerosol model derived from the microphysical aerosol scheme HAM [Weiss et al., GMD, 2025], to explore the sensitivity of anvils and their radiative effects to aerosol perturbations in the context of the ORCHESTRA/EarthCARE Model Intercomparison Project (ECOMIP) as well as the TRACER campaign MIP. We find that an increase in aerosol increases cloud droplet numbers, suppresses warm rain formation, increases convective mass flux and thereby upper tropospheric ice water content and will discuss how these changes translate into anvil cloud radiative effects. Prototype next generation km-scale climate models are implicitly already including such anvil radiative effects; however, these currently remain unconstrained by observations. We develop novel observational constraints on the convective anvil cloud lifecycle through consistent tracking of convection using the tobac-flow cloud tracking framework [Jones et al., 2024] between MSG SEVIRI observations and forward simulated geostationary satellite radiances from ICON model output.  This reveals that deep convective systems in ICON grow too fast and show a faster dissipation of thick to thin anvils than observations, which affects their radiative effects. Our work provides novel approaches to improve our understanding of aerosol effects on convective clouds and climate.

Abstract:

Understanding the driving forces behind mesoscale cloud organization is fundamental to reducing uncertainties in cloud climate feedbacks. Traditional climate models cannot explicitly resolve mesoscale cloud structures due to their limited resolution, leading to large uncertainties in cloud climate feedback estimates. Storm-resolving models that simulate the atmosphere at kilometre resolution have the potential to reduce these uncertainties. Yet, these models are still biased in their organizational structure when compared to satellite observations. Approaches constraining cloud feedbacks directly from the satellite records are promising but often rely on manually chosen cloud controlling factors (CCFs) that do not necessarily capture all the information necessary to explain mesoscale organizational structures and generally only utilise linear models to predict cloud radiative properties from CCFs.We present CloudDiff, a probabilistic machine learning model that generates mesoscale cloud structures at kilometre resolution conditioned on environmental conditions in the atmosphere, namely the temperature and humidity profiles as well as vertical and horizontal winds. The model is trained on MODIS Level 1 satellite data and environmental conditions from ECMWF ERA5 reanalysis data. CloudDiff is able to reconstruct realistic MODIS observations from matching ERA5 environmental conditions and achieves a lower reconstruction error compared to generating MODIS observations solely from pre-defined CCFs. In CloudDiff’s generation stage, the environmental conditions are compressed into a latent representation using an attention mechanism. This latent representation can be interpreted as a set of CCFs that have been learned purely from data. We’ll discuss the properties of the learned CCFs including how they relate to existing CCFs, their geographical distribution, and their predictive power of the radiative properties of cloud fields.

Abstract:

The top-of-atmosphere radiative effect of tropical anvil clouds varies with cloud opacity, and can range from substantially negative to largely positive. Recent climate model assessments have found a decrease in the proportion of thick, or opaque, anvil cloud with warming, resulting in a positive climate feedback. However, the mechanism for this change remains obscure.Lifecycle analysis of deep convective clouds tracked using tobac in the convection-permitting global ICOsahedral Non-hydrostatic model (ICON) shows how anvil area and opacity respond to convection. We find that both properties increase in response to increased convective intensity and convective area, but that their sensitivity to each is not equal. To interpret these results, we independently develop a simple analytical model that links anvil expansion and opacity to convective mass flux (CMF). The model predicts that higher CMF leads to greater anvil expansion, increasing the area of thick anvil cloud. But when anvil opacity also depends on convective intensity, we find a strong, non-linear increase in thick anvil amount in response to increasing CMF, consistent with the response observed in ICON. This implies a strong sensitivity of thick anvil amount to changes in the upper tail of the distribution of CMF and illustrates a possible mechanism by which changes in the distribution of cloud CMF could drive anvil thinning in a warming climate.

Abstract:

Studies of aerosol-cloud interactions and estimates of the effective aerosol radiative forcing (ERF) of climate depend crucially on the vertical distribution of aerosol microphysical and radiative properties, but few reliable observations of such properties exist on a global scale. The 2024 launch of the EarthCARE mission provides new observations of aerosol extinction from the ATMospheric LIDar (ATLID) system. These observations are proving to be superior to past satellite-based lidar observations of aerosol extinction in accuracy because of the use of the high-spectral resolution lidar (HSRL) technique. These high-accuracy lidar observations can be used as input to machine-learning (ML) models to estimate cloud condensation nuclei (CCN at 0.4% supersaturation) and aerosol absorption (ABS at 532nm).We present novel ML-based CCN and ABS retrievals using the first full year of ATLID observations (September 2024 to August 2025) of aerosol backscatter, extinction, and depolarization as predictors. These higher-level aerosol properties are compared to retrievals of the same quantities derived from airborne HSRL observations by the WALES system (derived from WAter vapor Lidar Experiment in Space) during the ORCESTRA (ORganized Convection and EarthCARE STudies over the Tropical Atlantic) PERCUSION (Persistent EarthCARE Underflight Studies of the ITCZ and Organized Convection) campaign in the summer of 2024. We provide validation results of the ML-based CCN and ABS retrievals against ground-based in situ observations, which indicate relative errors less than 30% for all but the cleanest aerosol loading conditions. Based on the first year of ATLID observations, we present global maps of ML-derived CCN and ABS and suggestions for improvements in the ATLID observations. Finally, we discuss opportunities to study aerosol-cloud-climate interactions facilitated by these new retrievals and climatologies.

Abstract:

At kilometer-scale resolution, convective systems start to be explicitly resolved in atmospheric models, albeit coarsely. This allows a more process-based analysis of certain aspects of aerosol–cloud interactions in tropical regions. Convective clouds are a ubiquitous feature above the Amazon rainforest and develop under strongly contrasting aerosol conditions, with particle number concentrations during the dry season often exceeding those in the wet season by an order of magnitude.In this context, we explore aerosol and convective cloud processes over the Amazon rainforest by analyzing case studies that combine observations and km-scale cloud-resolving simulations with interactive aerosols in a limited-area configuration. Regional simulations are performed at approximately 1.6 km horizontal resolution using the Icosahedral Nonhydrostatic (ICON) model coupled to the one-moment aerosol scheme HAM-lite. The realism of the simulations is evaluated through comparison with a combination of ground-based, satellite, and aircraft observations.For the wet season, we analyze a case study based on flight RF15, conducted with the German research aircraft HALO during the CAFE-Brazil (Chemistry of the Atmosphere: Field Experiment in Brazil; CAFE-BR) campaign in 2022–2023. Three simulations are presented for this case: a best-estimate factual simulation and two counterfactual sensitivity experiments representing background “green ocean” conditions and heavy aerosol loading associated with biomass burning during dry season periods.  For the dry season, we also revisit two research flights from the ACRIDICON-CHUVA 2014 campaign, representing one clean and one polluted case, to further assess the representation of aerosol–cloud processes under different convective regimes. Combining these cases, we discuss the impact of changing aerosol environments on convective systems and draw conclusions relevant to a transition toward a post-fossil aerosol regime.