Understanding the vertical structure of deep convective systems is essential for assessing theirimpacts on the atmospheric energy budget and hydrological cycle and for evaluating their repre-sentation in models. However, because of limitations of current satellite observations, the verticalstructure of these systems remains poorly constrained. We use a novel ice cloud dataset calledIceCloudNet to study the temporal evolution of the vertical structure of tropical deep convectivesystems on the basis of ice water content as a marker for convective intensity and anvil devel-opment. IceCloudNet is the first 4D-consistent semi-observational ice cloud dataset covering thetropical belt between 30°S–30°N and 30°W–30°E, developed by Jeggle et al. (2025). The spatialresolution is 3 km in the horizontal and 240 m in the vertical. The temporal resolution is 15 min.The dataset is constructed by filling observational gaps using machine learning. By applying theTobac cloud tracking algorithm to the vertically integrated ice water content over the course of theyear 2010, we identify and track deep convective systems to diagnose systematic changes in thevertical distribution of ice water content during their lifecycle. We also assess the suitability ofIceCloudNet for a robust and physically coherent tracking and analysis of vertically resolved cloudproperties. This allows us to highlight both its limitations and its potential to enable, for the firsttime, a comprehensive four-dimensional analysis of the evolution of tropical ice clouds.

Land-surface representations in weather and climate models simplify the characterization of vegetation as a single layer with bulk environmental conditions. This approach overlooks the vertical variability in leaf traits and environmental conditions within the canopy. This research explores the vertical variability of plant ecophysiology and environmental measurements within the Amazon tropical rainforest during daytime, specifically at the ATTO site, during the late dry season. To characterize the canopy and its vertical variability, we categorized the canopy into three layers: the top layer (approximately the upper third of the canopy, 18-27 m), the medium layer (approximately the medium third of the canopy, 9-18 m), and the low layer (approximately the lower third of the canopy, 0-9 m) where leaf gas exchange measurements were conducted. Utilizing these layers, we developed a multi-layer model representation that calculates water and CO2 fluxes based on within canopy on-site observations. We conducted sensitivity analyses of the rainforest multi-layer representation to discern the significance of capturing vertical variability in leaf traits and environmental conditions for deriving net fluxes of water and CO2 of the forest. Current results show that measured leaf traits exhibit vertical variation within the canopy, indicating larger productivity in the top layer compared to the medium and low layers. Environmental conditions, such as incoming radiation in the top layer, fluctuate due to cloud presence. Temperature peaks in the top layer and reaches a minimum at mid-canopy. This results in a non-uniform mixing of the canopy air, maintaining a stable layer within the forest canopy that can potentially affect the distribution of scalars within the canopy. Ongoing analyses explore the similarities and differences between the CO2 exchange between the multi-layer representation and CO2 fluxes from eddy covariance systems, as well as the sensitivity of the former to vertical variability in leaf traits and environmental conditions. By doing so, we aim to gain knowledge on the relevance (or irrelevance) of characterizing vertical variability in land-surface representations and on important processes that may not be well captured yet by land-surface representations.

Surface solar irradiance variability is present under all broken clouds, but the patterns, magnitude of variability, and driving mechanisms vary greatly with cloud type. In this study, we performed numerical experiments to understand which main mechanisms drive surface solar irradiance (SSI) variations across a diverse set of observation-based cloud conditions. The results show that four mechanisms capture the essence. We find that for optically thin (τ<6) clouds, scattering in the forward direction (forward escape) dominates. In cloud fields with enough optically thin area, such as altocumulus, forward escape alone can drive areas of irradiance enhancement of over 50 % of clear-sky irradiance. For flat, optically thick clouds (τ>6), irradiance is instead scattered diffusely downward (downward escape), and (extreme) enhancements are thus found directly below the cloud. For vertically structured clouds, side escape dominates the domain-averaged diffuse irradiance enhancement until the sides become shaded by anvil clouds. Lastly, under optically thick cloud cover, surface albedo enhances radiative fluxes due to multiple scatterings between surface and cloud. This brightens shadows and contributes 10 % to 60 % of the total irradiance enhancement for low (0.2) to high (0.8) albedo. With these four mechanisms, we provide a framework for understanding the vast diversity and complexity found in surface solar irradiance and cloudiness. A next step is to apply this analysis to multi-layered cloud fields and non-isolated deep convective clouds.

Diurnal temperature and carbon dioxide ranges are key metrics to quantify the impact of regional climate changes in forests. These ranges depend on biophysical processes, surface heat, water and carbon exchange, and boundary-layer dynamics. A crucial and elusive process is the entrainment of air from the free troposphere and residual air layers into the atmospheric boundary layer. Here we provide observational constraints on entrainment for two contrasting measurement sites: the Amazon Tall Tower Observatory (ATTO) in central Amazonia and the Loobos flux tower (NL-Loo) in a temperate forest in the Netherlands. We used radio soundings, air samples from tall towers and aircraft data in combination with surface air measurements and ecophysiological data. Fluxes and concentrations were measured for biophysical-process tracers CO2, O2/N2, δ13C, δ18O (in CO2) and δ18O (in water). These novel tracers are proposed to partition gross carbon and water fluxes and for estimating plant properties and we present a unique dataset with our interpretation. Our analysis enables us to unravel the role of entrainment on the diurnal ranges and how this is controlled by surface and entrainment fluxes. By means of a coupled forest-atmosphere model constrained by the comprehensive observations, we perform a sensitivity study on the surface flux partitioning (photosynthesis versus soil respiration; soil evaporation versus plant transpiration, sensible versus heat flux) under a wide range of leaf traits, surface and boundary-layer dynamic conditions. Our results are useful to assess the performance of carbon-climate models in tropical and temperate forests.

We analyze the diurnal variability of atmospheric , , and CO2 above the canopies of two contrasting ecosystems: the Amazon tropical forest and the Loobos temperate forest. Using a coupled forest-atmosphere model constrained by tower-based and aircraft observations, we quantify the role of atmospheric processes-including entrainment, subsidence, and cloud ventilation-in shaping the diurnal amplitude, or diurnal range (DR), of carbon-cycle tracers. Our results show that atmospheric processes can contribute more than twice as much as surface processes to DR. Misrepresenting these influences leads to substantial errors in interpreting observations and modeling tracer variability. We propose using DR as a metric to evaluate atmospheric tracer transport models and to compare site-level measurements. We present a roadmap to identify which atmospheric or surface processes are poorly represented when modeled and observed DR diverge.

We analyze the diurnal variability of atmospheric , , and CO 2 above the canopies of two contrasting ecosystems: the Amazon tropical forest and the Loobos temperate forest. Using a coupled forest‐atmosphere model constrained by tower‐based and aircraft observations, we quantify the role of atmospheric processes—including entrainment, subsidence, and cloud ventilation—in shaping the diurnal amplitude, or diurnal range (DR), of carbon‐cycle tracers. Our results show that atmospheric processes can contribute more than twice as much as surface processes to DR. Misrepresenting these influences leads to substantial errors in interpreting observations and modeling tracer variability. We propose using DR as a metric to evaluate atmospheric tracer transport models and to compare site‐level measurements. We present a roadmap to identify which atmospheric or surface processes are poorly represented when modeled and observed DR diverge.

This year marks the end of the Shedding Light On Cloud Shadows project (SLOCS, 2019-2024). SLOCS aims to understand temporal, spatial, and spectral variability in surface solar irradiance driven by individual clouds from field observations and 3D cloud-resolving large-eddy simulations. In this contribution, we would like to present the highlights of the project and the most important conclusions.The reason for initiating SLOCS is that clouds trigger large fluctuations in solar surface irradiance, and therefore in surface heat fluxes, but there is still much to be learned about these fluctuations. The incoming radiation in shadows is almost an order of magnitude less than under clear sky, while peaks near clouds shadows can sometimes reach a 50% increase with respect to clear sky, due to scattering of sunlight on clouds. Performing cloud-resolving simulations with realistic surface solar irradiance patterns under broken clouds remains therefore a challenge, and current cloud-resolving models do not capture the radiation-cloud interactions well. The Shedding Light On Cloud Shadows (SLOCS) project addresses this challenge by i) performing spatial observations in a spatial grid fine enough (~50 m, 10 Hz) to capture individual clouds using a newly designed instrument, and ii) developing 3D radiative transfer models for cloud-resolving models with optimal balance between detail level and performance. The FESSTVaL, LIAISE, and CloudRoots campaigns provided unique opportunities to measure surface solar irradiance around cloud shadows in different climates. In the campaigns, we performed grid measurements of radiation, while benefiting from complementary boundary-layer and cloud observations.The most important lessons learned from the field observations are:1. Scales as small as meters and seconds contribute significantly to fluctuations in surface solar irradiance2. All broken cloud patterns generate strong peaks, but the underlying mechanisms vary greatly amoung cloud types3. Spectral variations (in colors of light) are mostly significant under cumulus clouds.We used those observations to set up a series of cloud-resolving simulations with MicroHH and to evaluate two newly-developed radiative transfer solvers: i) a ray tracer fast enough to be coupled to our cloud-resolving model and ii) a solver that post-processes the outcome of a 1D two-stream solver to emulate 3D effects. Also, we studied the impact of periodic and open lateral boundary conditions. The most important conclusions are:1. Capturing 3D interactions between clouds and radiation accurately leads to larger clouds with more liquid water compared to those in simulations with conventional 1D methods2. Post-processing conventional 1D radiation computations allows for simulating surface solar irradiance fields with realistic probability density functions, but inaccurate cloud shadow shape and location.3. Open lateral boundaries in large-eddy simulations are at least as important as correct radiation-cloud interactions in producing realistic cloud shadows in the range from hectometers to kilometers.

Surface solar irradiance varies on scales as small as seconds or meters due to scattering and absorption by the atmosphere. Clouds are the main driver of this variability, but moisture structures in the atmospheric boundary layer and aerosols have an influence too, and depend on wavelength. The highly variable nature of solar irradiance is not resolved by most atmospheric models, yet it affects most notably the land-atmosphere coupling, which in turn can change the cloud field, and the quality of solar energy forecasting. Spatially and spectrally resolved observational datasets of solar irradiance at such high resolution are rare, but they are required for characterising observed variability, understanding the mechanisms, and developing fast models capable of accurately resolving this variability. In 2021, we deployed a spatial network of low-cost radiometers at the FESSTVaL (Germany) and LIAISE (Spain) field campaigns, specifically to gather data on cloud-driven surface patterns of irradiance, including spectral effects, with the aim to address this gap in observations and understanding. We find in case studies of cumulus, altocumulus, and cirrus clouds that these clouds generate large spatiotemporal variability in irradiance, but through different mechanisms and at difference spatial scales, ranging from 50 m to 30 km. Spectral irradiance in the visible range varies at similar spatial scales, with significant blue enrichment in cloud shadows, most strongly for cumulus, and red enrichment in irradiance peaks, particularly in the case of semi-transparent clouds or near cumulus cloud edges. Under clear-sky conditions, solar irradiance varies significantly in water vapour absorption bands at the minute scale, due to local and regional variability in atmospheric moisture.

Shedding Light On CloudRoots Solar spectral irradiance measurements made with the sensors produced within the Shedding Light On Cloud Shadows (SLOCS) project, deployed at the CloudRoots Amazonia 2022 campaign. Dataset contents Level 0 (raw): the raw data as it comes from the instruments Level 1 (L1): data in NetCDF format with metadata, quality control, homogenized factory calibration (counts bin-1 dt-1) Level 2 (L2): calibrated L1 data in W m-2 nm-1 extras: this folder includes reference calibration spectra and data quality quicklooks Data is available at 1 Hz (resampled) and 10 Hz (native) resolution. 10 Hz resolution is compressed using NetCDF compression with gzip level 5 (uncompressed is 1.13 GB per date). Data quality and uncertainty Please note this dataset is in version 0.1.0, meaning you should use the dataset with caution. Not all unphysical data may have been flagged as such, and spectral calibration is an estimate based on a simple modelled spectrum. This modelled spectrum is a standard tropical atmosphere without aerosols, and is not run with observed profiles except an ERA5 estimate of total column water vapour. Please refer to 'extras' for technical validation of the spectral calibration method, and LibRadtran input/output files. A production (1.0) version will be released as soon data is fully validated. Lower-end uncertainty can be estimated by looking at the sensor to sensor spread at wavelength level during the calibration measurements. In the calibration phase, all sensors were co-located and homogenized at wavelength level. The 13:50 to 14:10 UTC time on August 7 is the reference frame for spectral calibration. Other sources of uncertainty are difficult to quantify due to measurements taking place in a very heteregeneous forest. These uncertainties relate primarily to the less-than-perfect placement of sensors on the towers in comparison to the reference calibration phase. Sensor 18 is only available in raw data or calibrated data. Precalibration (homogenizing) is not possible given its deviating spectral filter set compared to the others (sensor version 3b vs. 3a). Technical information The NetCDF files comply with CF1.7 where applicable. Metadata include sensor location (altitude relative to ground and sea level, lat, lon). Code for processing raw data to NetCDF available at https://zenodo.org/records/10159129 Calibration of raw sensor units to spectral irradiance is done using a reference clear-sky spectrum simulated with LibRadtran. Settings and output is included in "extras". More information SLOCS project homepage CloudRoots project homepage 2022 campaign reference paper is in preparation See 'related works' for the instrument reference paper

Surface solar irradiance varies on scales down to seconds, and detailed long-term observational datasets of this variable are rare but in high demand. Here, we present an observational dataset of global, direct, and diffuse solar irradiance sampled at 1 Hz as well as fully resolved variability until at least 0.1 Hz over a period of 10 years from the Baseline Surface Radiation Network (BSRN) station at Cabauw, the Netherlands. The dataset is complemented with irradiance variability classifications, clear-sky irradiance and aerosol reanalysis, information about the solar position, observations of clouds and sky type, and wind measurements up to 200m above ground level. Statistics of variability derived from all time series include approximately 185 000 detected events of both cloud enhancement and cloud shadows. Additional observations from the Cabauw measurement site are freely available from the open-data platform of the Royal Netherlands Meteorological Institute. This paper describes the observational site, quality control, classification algorithm with validation, and the processing method of complementary products. Additionally, we discuss and showcase (potential) applications, including limitations due to sensor response time. These observations and derived statistics provide detailed information to aid research into how clouds and atmospheric composition influence solar irradiance variability as well as information to help validate models that are starting to resolve variability at higher fidelity. The main datasets are available at https://doi.org/10.5281/zenodo.7093164 (Knap and Mol, 2022) and https://doi.org/10.5281/zenodo.7462362 (Mol et al., 2022); the reader is referred to the "Code and data availability" section of this paper for the complete list.

Clouds cast shadows on the surface and locally enhance solar irradiance by absorbing and scattering sunlight, resulting in fast and large solar irradiance fluctuations on the surface. Typical spatiotemporal scales and driving mechanisms of this intra-day irradiance variability are not well known, hence even one day ahead forecasts of variability are inaccurate. Here we use long term, high frequency solar irradiance observations combined with satellite imagery, numerical simulations, and conceptual modelling to show how irradiance variability is linked to the cloud size distribution. Cloud shadow sizes are distributed according to a power law over multiple orders of magnitude, deviating only from the cloud size distribution due to cloud edge transparency at scales below 750 meters. Locally cloud-enhanced irradiance occurs as frequently as shadows, and is similarly driven mostly by boundary layer clouds, but distributed over a smaller range of scales. We reconcile studies of solar irradiance variability with those on clouds, which brings fundamental understanding to what drives irradiance variability. Our findings have implications for not only for weather and climate modelling, but also for solar energy and photosynthesis by vegetation, where detailed knowledge of surface solar irradiance is essential.

Surface solar irradiance varies on scales down to seconds or meters due to clouds. This highly variable nature of irradiance is not resolved by atmospheric models, yet heterogeneity in surface irradiance impacts the overlying cloud field. The inability to resolve irradiance variability, aside from insufficient model resolution, is explained by our limited understanding of cloud-driven solar irradiance variability at short spatiotemporal scales and the lack of high resolution spatial observational data. Cloud resolving models utilizing ray tracing techniques are a useful research tool, but ultimately require validation against observations.In 2021, we gathered new observational data with a network of radiometers, specifically designed to gather data on cloud-driven surface patterns of irradiance. I will present results on various kinds of surface patterns in relation to cloud type and atmospheric conditions, based on these observations. Our radiometers sample surface solar irradiance at 10 Hz for 18 wavelengths, which we deployed in different setups in the FESSTVaL (Germany) and LIAISE (Spain) field campaigns. Our results highlight the complexity and wide range of regimes in spatiotemporal irradiance variability, but also provide insights into its driving mechanisms. These insights help guide the development of improved radiative transfer calculations, in order to move towards models that can accurately resolve irradiance variability in an operational setting.

Vegetation and atmosphere processes are coupled through a myriad of interactions linking plant transpiration, carbon dioxide assimilation, turbulent transport of moisture, heat and atmospheric constituents, aerosol formation, moist convection, and precipitation. Advances in our understanding are hampered by discipline barriers and challenges in understanding the role of small spatiotemporal scales. In this perspective, we propose to study the atmosphere–ecosystem interaction as a continuum by integrating leaf to regional scales (multiscale) and integrating biochemical and physical processes (multiprocesses). The challenges ahead are (1) How do clouds and canopies affect the transferring and in‐canopy penetration of radiation, thereby impacting photosynthesis and biogenic chemical transformations? (2) How is the radiative energy spatially distributed and converted into turbulent fluxes of heat, moisture, carbon, and reactive compounds? (3) How do local (leaf‐canopy‐clouds, 1 m to kilometers) biochemical and physical processes interact with regional meteorology and atmospheric composition (kilometers to 100 km)? (4) How can we integrate the feedbacks between cloud radiative effects and plant physiology to reduce uncertainties in our climate projections driven by regional warming and enhanced carbon dioxide levels? Our methodology integrates fine‐scale explicit simulations with new observational techniques to determine the role of unresolved small‐scale spatiotemporal processes in weather and climate models.

Description Measurements from the spatial network of 15 radiometers deployed at the LIAISE field campaign in 2021. 14 July 2021 until 28 July 2021. Dataset Contents rsds: Total shortwave downwelling solar irradiance @ 10 Hz, 1 sec, and 1 minute resolution (Level 2, derived dataset) prw: Total column integrated water vapour @ 1 sec resolution (Level 2, derived dataset) spectrum: Pre-calibrated solar spectral irradiance measurements @ 10 Hz resolution (Level 1, source dataset) raw data: straight from the sensors (Level 0) Dataset Quality All data is quality controlled and completed with metadata and quality flags. Level 2 data is calibrated against high quality references, and derived from Level 1 data. Methodology, performance, and usage all described in detail in an upcoming pre-print. References and more info Radiometer reference paper (FROST) LIAISE campaign website, LIAISE database Dataset description paper: coming soon Code to produce these data: coming soon Version History v1.1: added raw (level 0) data, fixed typo in file name for the 10 Hz spectrum zip (L2 -> L1).

We developed a cost-effective Fast-Response Optical Spectroscopy Time-synchronized instrument (FROST). FROST can measure 18 light spectra in 18 wavebands ranging from 400 to 950 nm with a 20 nm full-width half-maximum bandwidth. The FROST 10 Hz measurement frequency is time-synchronized by a global navigation satellite system (GNSS) timing pulse, and therefore multiple instruments can be deployed to measure spatial variation in solar radiation in perfect synchronization. We show that FROST is capable of measuring global horizontal irradiance (GHI) despite its limited spectral range. It is very capable of measuring photosynthetic active radiation (PAR) because 11 of its 18 wavebands are situated within the 400-to-700 nm range. A digital filter can be applied to these 11 wavebands to derive the photosynthetic photon flux density (PPFD) and retain information on the spectral composition of PAR. The 940 nm waveband can be used to derive information about atmospheric moisture. We showed that the silicon sensor has undetectable zero offsets for solar irradiance settings and that the temperature dependency as tested in an oven between 15 and 46 degrees C appears very low ( 250 ppmK(-1)). For solar irradiance applications, the main uncertainty is caused by our polytetrafluoroethylene (PTFE) diffuser (Teflon), a common type of diffuser material for cosine-corrected spectral measurements. The oven experiments showed a significant jump in PTFE transmission of 2% when increasing its temperature beyond 21 degrees C. The FROST total cost (< EUR 200) is much lower than that of current field spectroradiometers, PAR sensors, or pyranometers, and includes a mounting tripod, solar power supply, data logger and GNSS, and waterproof housing. FROST is a fully standalone measurement solution. It can be deployed anywhere with its own power supply and can be installed in vertical in-canopy profiles as well. This low cost makes it feasible to study spatial variation in solar irradiance using large-grid high-density sensor setups or to use FROST to replace existing PAR sensors for detailed spectral information.

Numerical weather prediction models operate on grid spacings of a few kilometers, where deep convection begins to become resolvable. Around this scale, the emergence of coherent structures in the planetary boundary layer, often hypothesized to be caused by cold pools, forces the transition from shallow to deep convection. Yet, the kilometer-scale range is typically not resolved by standard surface operational measurement networks. The measurement campaign Field Experiment on Submesoscale Spatio-Temporal Variability in Lindenberg (FESSTVaL) aimed at addressing this gap by observing atmospheric variability at the hectometer-to-kilometer scale, with a particular emphasis on cold pools, wind gusts, and coherent patterns in the planetary boundary layer during summer. A unique feature was the distribution of 150 self-developed and low-cost instruments. More specifically, FESSTVaL included dense networks of 80 autonomous cold pool loggers, 19 weather stations, and 83 soil sensor systems, all installed in a rural region of 15-km radius in eastern Germany, as well as self-developed weather stations handed out to citizens. Boundary layer and upper-air observations were provided by eight Doppler lidars and four microwave radiometers distributed at three supersites; water vapor and temperature were also measured by advanced lidar systems and an infrared spectrometer; and rain was observed by a X-band radar. An uncrewed aircraft, multicopters, and a small radiometer network carried out additional measurements during a 4-week period. In this paper, we present FESSTVaL’s measurement strategy and show first observational results including unprecedented highly resolved spatiotemporal cold-pool structures, both in the horizontal as well as in the vertical dimension, associated with overpassing convective systems.

Radiation in the atmosphere provides the energy that drives atmospheric dynamics and physics on all scales, from cloud particle growth to global weather and climate. Radiation schemes in global weather and climate models make assumptions to simplify the complex interaction of radiation with the Earth system. Capturing cloud-radiation interactions is particularly challenging since clouds vary strongly on small spatial and temporal scales not resolved in the models, and interact strongly with radiation. Uncertainties in these assumptions in the radiation scheme and the cloud, aerosol, gas and surface inputs lead to uncertainties in multiple weather and climate processes, such as energy balance, cloud development and dynamics. The modular radiation scheme ecRad (Hogan and Bozzo, 2018, Rieger et al. 2019) is operational in ICON at DWD since April 2021 and provides the opportunity to vary parametrisations and assumptions individually to determine their impact. Several options are available for the radiation solver, cloud vertical overlap and horizontal inhomogeneity treatment and cloud hydrometeor optical property parametrisations. The solver SPARTACUS is the only radiation solver in a global model that can treat 3D radiative effects. Using global satellite and surface data and high-resolution surface radiation measurements gathered during the FESSTVaL campaign (https://fesstval.de), we evaluate the radiation and cloud parametrisations on local to global scales and investigate the sensitivity of radiation results to model assumptions and cloud properties and the role of cloud-radiation interactions. In ICON, ecRad improves the global radiation balance, model physics and forecast performance as evaluated against observations. References: Hogan, R. J., & Bozzo, A. (2018), A flexible and efficient radiation scheme for the ECMWF model. Journal of Advances in Modeling Earth Systems, 10, 1990-2008. https://doi.org/10.1029/2018MS001364 Rieger, Daniel, Martin Köhler, Robin J. Hogan, Sophia A. K. Schäfer, Axel Seifert, Alberto de Lozar and Günther Zängl (2019). ecRad in ICON - Implementation Overview, Reports on ICON

Clouds cast shadows and locally enhance solar irradiance through absorbing and scattering sunlight, resulting in fast and large solar irradiance fluctuations on the surface. The resulting spatiotemporal variability poses a challenge for solar energy production amidst increasing need for reliable renewable energy. It furthermore influences biological processes and the exchange of water and energy. Yet, no numerical weather prediction model is able to reproduce the observed local properties of irradiance, due to the complexity of radiative transfer and its dependence on accurately resolved cloud fields. Improving the radiative transfer models, whether it involves running full Monte Carlo raytracing in academic setups or simplified paramerizations, ultimately requires observations for validation. However, dense spatial observation of irradiance on the scale of cloud shadows are rare. Even single 1D time series are rarely available at high enough temporal resolution to capture irradiance variability. In ongoing work, we provide those missing observations using a dense network of our custom, low-cost radiometers that we deployed at two field campaigns in summer 2021, FESSTVaL (Germany) and LIAISE (Spain). I will present our gathering and analyses of these new and detailed observations of surface irradiance to address knowledge gaps in our physical understanding and provide validation datasets for models. The instruments, which sample at 10 Hz, are able to closely match expensive conventional instruments, and combined with skyview imagery, the spatial observations are directly linked to observed clouds. Information about atmospheric water content can be retrieved using the information from water vapour absorption bands. To complement these short term spatial data, long-term statistics of irradiance variability are derived from a 10-year 1 Hz resolution dataset from the Baseline Surface Radiation Network station in Cabauw, the Netherlands. Distributions and typical spatio-temporal scales of cloud shadows and irradiance peaks can be related to cloud type and meteorological conditions. The gathering and study of these datasets will lead to a better understanding of the physics. E.g., whether the dominant mechanism driving irradiance peaks is either forward scattering in transparent parts of clouds or 'reflections' from cloud sides. Furthermore, these datasets will help validate models, and ultimately improve our ability to accurately forecast irradiance variability at the small scales.

Boundary-layer clouds trigger large fluctuations in solar surface irradiance and in surface heat fluxes. The incoming radiation in shadows is almost an order of magnitude less than under clear sky, while peaks near clouds shadows can sometimes reach a 50% increase with respect to clear sky, due to scattering of sunlight in cloud edges. Performing large-eddy simulation (LES) with realistic surface solar irradiance patterns under broken clouds remains a challenge. First, this is due to the absence of spatial radiation observations that capture individual cloud shadows at a typical LES resolution (~50 m), second, because cloud fields need to be accurate up to a very high detail level and, third, the 3D aspects of radiative transfer needs to taken into account. The Shedding Light On Cloud Shadows (SLOCS) project aims to overcome these challenges by i) gathering spatial observations in a spatial grid fine enough to capture individual clouds with a novel instrument, and ii) further developing 3D radiative transfer models for LES with optimal balance between detail level and performance. The FESSTVaL campaign in Lindenberg, Germany in early summer 2021 provided a unique opportunity for the SLOCS team to address those challenges. In FESSTVaL, we performed grid measurements of radiation, while benefiting from complementary boundary-layer and cloud observations of other research groups. In addition, FESSTVaL brought a boundary-layer modelling community together ranging from people working on NWP models to people working on fine-scale LES. This permitted comparison of the ability of different modelling techniques to capture surface irradiance variability driven by clouds. Here, we will present the design and the results of LES of four selected case studies based on FESSTVaL data: one clear sky case, two shallow cumulus cases with different cloud depths, and one deep convection case with cold pools. First, we will show how we have extended and accelerated the RTE+RRTMGP radiation model to take into account 3D interactions between clouds and radiation, to enable comparison against our grid observations near the Falkenberg measurement tower. Second, we will present the outcome of the LES of the four cases and evaluate simulated solar and turbulence surface fluxes, boundary layer structures and cloud properties against FESSTVaL observations. Third, we will show a comparison between our own 25 m resolution large-eddy simulations with GPU-accelerated MicroHH against 100 m resolution simulations of the ICON large-eddy model and 2 km resolution simulations of the ICON NWP model. In our analyses, we compare the different modelling techniques in their ability to reproduce the FESSTVaL cases. This comparison will address the balance between grid resolution and domain size as well as the necessity for lateral inflow and outflow boundary conditions in modelling accurate surface solar irradiance fluxes.

Surface irradiance variability is present on many spatio-temporal scales, but most strongly on the scale of minutes to seconds due to low broken clouds. Fast and large fluctuations, or spatial heterogeneity, of irradiance affects solar energy production. In idealised settings, let alone in operational forecasts, the modelling of realistic fields of surface irradiance in the presence of clouds is challenging. It relies on realistic cloud fields, is computationally demanding due to the nature of 3-d radiative transfer models, and ultimately requires observations for validation. Dense spatial observation of irradiance on the scale of cloud shadows or solar energy parks are rare, however. Even 1-d time series are often not available at high enough resolution. In ongoing work, we provide those missing observations. I will present our gathering and analyses of new and detailed observations of surface irradiance to address knowledge gaps in our physical understanding and provide validation datasets for models. In 2021, we deployed a dense network of custom, low-cost radiometers at two field campaigns, FESSTVaL (Germany) and LIAISE (Spain), to observe spatial patterns of irradiance driven by clouds. The instruments are able to closely match expensive conventional instruments, and combined with skyview imagery, the spatial observations are directly linked to observed clouds. To complement these short term spatial data, long-term statistics of irradiance variability are derived from a 10-year 1 Hz resolution data from the Baseline Surface Radiation Network station in Cabauw, the Netherlands. Distributions and typical spatio-temporal scales of cloud shadows and irradiance peaks can be related to cloud type and meteorological conditions. The gathering and study of these datasets will lead to a better understanding of the physics, help validate models, and ultimately improve our ability to accurately forecast irradiance variability at the small scales.