Search

In this contribution, a statistical learning technique is used to quantify the response of cloud radiative effects to changes in a large number of environmental factors in spatial observation data.Clouds play a key role for the Earth’s energy balance; however, their response to climatic and anthropogenic aerosol emission changes is not clear, yet. Here, 20 years of satellite observations of cloud radiative effects (CRE) are analysed together with reanalysis data sets in a (regularised) ridge regression framework to quantitatively link the variability of observed CREs to changes in environmental factors, or cloud-controlling factors (CCFs). In the literature the meteorological kernels of such CCF analyses are typically established in regime-specific regression frameworks based on a low (2-8) number of CCFs. In our data-driven approach, the capabilities of the regularised regression to deal with collinearities in a large number of predictors are exploited to establish a regime-independent CCF framework based on a large number of CCFs. Using a reference 7-CCF framework, we show that ridge regression produces nearly identical patterns of CCF sensitivities when compared to the traditional regression. In the data-driven framework, however, the traditional regression fails at producing consistent results due to overfitting. The data-driven analysis reveals distinct regional patterns of CCF importance for shortwave and longwave CRE: Sea surface temperatures and inversion strength are important for shortwave CRE in stratocumulus regions, in agreement with existing studies. However, zonal wind speeds in the free troposphere and surface fluxes are also shown to be important. Free tropospheric meridional winds are important drivers of CRE in the subtropical belts (20°-40°) in both hemispheres, likely capturing aspects of Rossby Wave-related CRE variability. Aerosols are shown to be most important for shortwave CRE in the regions of stratocumulus to cumulus transition. While the multivariate method aims at limiting the influence of confounding factors on the estimated sensitivities, particularly the aerosol-CRE sensitivity may still be confounded to a degree. Future analyses of interactions between different CCFs and comparisons to global climate models are outlined.

This contribution presents spatially distinct fog and low stratus (FLS) formation and dissipation regimes derived from satellite and reanalysis data and their sensitivities to meteorological and land surface conditions over central Europe.FLS formation and dissipation processes are strongly governed by meteorological and land-surface conditions and vary geographically and across seasons. The timing of FLS formation and dissipation further has implications for traffic and solar energy production. While climatological analyses of FLS formation and dissipation exist, the influence of meteorological and land surface conditions on specific FLS formation and dissipation regimes is not clear yet.In this study, satellite-derived fog and low stratus life cycle information is explicitly linked to land surface and meteorology in Europe. Pixel-based correlations of FLS formation and dissipation times with environmental conditions were analyzed in a hierarchical clustering approach. Spatially distinct regimes of FLS formation and dissipation can be identified, and a dependency on the background geography and climate is apparent. FLS formation and dissipation regimes are analyzed on multiple hierarchy levels, i.e. various cluster numbers are explored, to investigate the spatial division of larger regional regimes to smaller sub-regional and local regimes. Monthly mean sensitivities for selected sub-regional regimes suggest a dependency of those sensitivities on FLS type. A case study of a regime covering the Po valley in northern Italy is used to showcase the quantification of FLS drivers regionally using explainable machine learning. Future work will expand this approach to analyze and compare FLS drivers in regional regimes with different background geography and climate.

This study applies explainable machine learning to near-global daily satellite and reanalysis data. We quantify and analyse the sensitivities of cloud fraction (CLF) to aerosol changes and their dependence on meteorological parameters. Aerosol-cloud interactions have prolonged influences on the Earth’s radiation budget but remain one of the most substantial uncertainties in the climate system. Marine boundary layer clouds (MBLCs) are particularly important since they cover a large portion of the Earth’s surface. One of the biggest challenges in quantifying aerosol-cloud interactions from observations lies in isolating the CLF adjustments due to aerosol perturbations from the covariability of local meteorology and quantifying the influences of meteorology on the aerosol-CLF relationship. In this study, 10 years (2011-2020) of near-global daily satellite cloud products are used in combination with reanalysis data of meteorological confounders. Using cloud droplet number concentration (CDNC) as a proxy for aerosol, MBLC CLF is predicted by region-specific gradient boosting machine learning models. By means of SHapley Additive exPlanation (SHAP) regression values, the predictions are explained by quantifying the sensitivities of CLF to the predictors. Furthermore, the meteorological influences on the CLF adjustments are analysed with SHAP interaction values to define an interaction index (IAI). Globally, the regional ML models are able to capture on average 32% and up to around 71% of the variability of CLF. Global patterns of CLF sensitivities show that CLF is positively associated with CDNC and lower tropospheric stability (LTS), strongest in low-cloud regions. Increased sea surface temperature (SST) on the other hand will lead to reduced CLF, probably by increasing the vertical moisture gradient. The CDNC-CLF sensitivities are especially strong in stratocumulus-to-cumulus transition regions. Negative CLF sensitivities to the u wind component at 700 hPa are found for most regions which may indicate an influence of facilitated turbulence at the cloud top. In terms of the interactive effects of meteorological parameters, a significant dependence of the CDNC-CLF relationship on LTS and SST is found in low-cloud regions, and the patterns coincide with sensitivities. Positive IAIs are shown globally for SST and LTS, indicating that the CDCN-CLF sensitivity is stronger with high SST/LTS values. The ongoing work shows that CDNC-CLF sensitivity is positive globally after accounting for meteorological covariations. Globally, SST and LTS can influence the positive CDNC-CLF relationship significantly, which is especially the case in stratocumulus regions. Detailed investigations will be carried out for not only SST/LTS but also other predictors to dig out the physics and causality behind the statistics.

This contribution presents early results on the effects of biomass burning aerosols (BBA) on fog and low clouds (FLC) in the Namib, using data from multiple satellite platforms and station measurements.Fog, which is the most relevant non-rainfall water source for plants and animals in the coastal parts of the Namib Desert, may become increasingly important for local ecosystems as regional climate simulations predict a warmer and drier climate for southern Africa in the future. Previous studies showed the role of BBA on cloud development over the ocean off the Namibian coast. The same processes are likely to influence Namib-region FLC formation and persistence as well. However, the potential effects of aerosols on FLC in the Namib Desert, a direct extension of the South-East Atlantic cloud system, have yet to be investigated.A clear seasonal cycle of FLC dissipation is found in a satellite-based product of FLC formation and dissipation times, with longer FLC persistence during the BBA season. Using a BBA reanalysis product in combination with the satellite data, it is found that during this season, FLC dissipation times are positively correlated to BBA loading (higher aerosol loading coinciding with later FLC dissipation). It is assumed that semi-direct and indirect BBA effects contribute to this pattern, with further analyses aimed at isolating aerosol effects from possible confounders.These findings are a first step in a better understanding of the Namib-region FLC system and will help in the development of a statistical model to quantify the sensitivities of FLC lifetime in the region in the next step of the project.

Knowledge of fog and low stratus (FLS) cloud patterns and life cycles is important for traffic safety, for the production of solar energy and for the analysis of cloud processes in the climate system. While meteorological stations provide information on FLS, a data set describing FLS formation and dissipation times on large spatial and temporal scales does not exist yet. In this study, we use logistic regression to extract FLS formation and dissipation times from a satellite-based 10-year FLS data set covering central Europe. The resulting data set is the first to provide a geographic perspective on FLS formation and dissipation at a continental scale. The patterns found show a clear dependency of FLS formation and dissipation times on topography. In mountainous areas, FLS forms in the night and dissipates in the morning. In river valleys, the typical FLS life cycle shifts to formation after sunrise and dissipation in the afternoon. Seasonal patterns of FLS for mation and dissipation show similar FLS formation and dissipation times in winter and autumn, and in spring and summer, with longer events in the former two seasons.

Changes in marine boundary layer cloud (MBLC) radiative properties in response to aerosol perturbations are largely responsible for uncertainties in future climate predictions. In particular, the relationship between the cloud droplet number concentration (Nd, a proxy for aerosol) and the cloud liquid water path (LWP) remains challenging to quantify from observations. In this study, satellite observations from multiple polar-orbiting platforms for 2006–2011 are used in combination with atmospheric reanalysis data in a regional machine learning model to predict changes in LWP in MBLCs in the Southeast Atlantic. The impact of predictor variables on the model output is analysed using Shapley values as a technique of explainable machine learning. Within the machine learning model, precipitation fraction, cloud top height, and Nd are identified as important cloud state predictors for LWP, with dynamical proxies and sea surface temperature (SST) being the most important environmental predictors. A positive nonlinear relationship between LWP and Nd is found, with a weaker sensitivity at high cloud droplet concentrations. This relationship is found to be dependent on other predictors in the model: Nd–LWP sensitivity is higher in precipitating clouds and decreases with increasing SSTs.

In this contribution geographic patterns of fog and low stratus (FLS) formation and dissipation over central Europe are presented using a novel satellite-based data set. Formation and dissipation of FLS are the results of complex interactions of meteorological and land-surface processes. Furthermore, the timing of FLS formation and dissipation has implications for traffic and the production of solar energy. Yet, little is known about the spatial and temporal patterns of both in central Europe. To improve this situation, this study analyzes these patterns, as well as the meteorological drivers and their modifications by the land surface. The basis of the analysis is a novel FLS formation and dissipation data set, derived based on satellite FLS products and logistic regression.Very distinct and contrasting spatial and seasonal patterns of FLS formation and dissipation are found across the study area. In large-scale river valleys, FLS forms most frequently in the morning and dissipates in the afternoon. In mountainous areas and on the coast, FLS forms during the night and dissipates in the morning. FLS persists longer in winter compared to other seasons. The quantitative analysis of meteorological drivers shows that the large-scale meteorological conditions, in particular mean surface pressure and wind speed, substantially influence the timing of FLS formation and dissipation. Local variations in topography modulate these conditions, leading to local differences in the observed patterns.

In this contribution, statistical and machine learning techniques are applied to quantify the response of clouds to changes in environmental factors.Clouds play a key role for the Earth’s energy balance; however, their response to climatic and anthropogenic aerosol emission changes is not clear, yet. Here, 20 years of satellite cloud observations are analyzed together with reanalysis data sets in multivariate-regression and machine-learning approaches to quantitatively link the variability of observed cloud radiative effects to changes in environmental factors, or cloud-controlling factors (CCFs). In this data-driven approach, a large number of CCFs, including aerosol proxies, are used as predictors at a large spatial and temporal scale typical of CCF analyses. The analysis reveals distinct regional patterns of CCF importance for shortwave and longwave cloud radiative effects. In stratocumulus cloud regions, the main controls of shortwave CRE are the sea surface temperature and the estimated inversion strength, but also zonal winds in the lower free troposphere are relevant controls of CRE. Aerosol proxies are shown to be most important for shortwave CRE in the regions of stratocumulus to cumulus transition. Future analyses of interactions between different CCFs and comparisons to global climate models are outlined.

In this communication, we have found that the broadband longwave radiative effect of the cloud-aerosol transition zone conditions is of the order of 8.0 ±3.7 W m−2 by combining satellite measurements with radiative transfer modeling. It is often required to differentiate clouds and aerosols from each other in atmospheric studies, but the decision on where the boundaries of the clouds should be put is a point of debate. As a result, what detected as cloud by one method/instrument may be labeled differently by another. This is because 1) clouds and aerosols often co-exist and interact with each other, and 2) change in the state of sky from cloudy to cloudless (or vice versa) comprises an additional condition called “transition zone” (or “twilight zone”) at which the characteristics of the particle suspension lay between those corresponding to pure clouds and atmospheric aerosols [Koren et al. (2007) GRL, 34(8): L08805. 10.1029/2007GL029253]. Nevertheless, a vast area that potentially may represent the transition zone is usually neglected in the observations or assumed as an area that contains either aerosols or optically thin clouds. In this communication, we provide quantitative information about the broadband longwave radiative effects of the transition zone conditions at the top of the atmosphere based on the radiative observations made by the CERES and MODIS instruments onboard Aqua spacecraft and radiative transfer simulations. Specifically, we used the MODIS measurements to look for CERES footprints with homogeneous transition zone and clear-sky conditions over the Southeast Atlantic Ocean for August 2010. Then, CERES observations under homogeneous transition and clear-sky conditions were compared with the corresponding clear-sky radiances, which were simulated using the SBDART radiative transfer model, fed with ERA5 reanalysis atmospheric profiles. For the studied period and domain, transition zone broadband longwave radiative effect was on average equal to 8.0 ±3.7 W m−2 (heating effect; median: 5.4 W m−2), although cases with radiative effects as large as 50 W m−2 were observed. Furthermore, low-level transition zone conditions defined as those with suspension top height below 2 km (determined based on the difference between the layer top and surface temperature) on average produced a radiative effect of about 4.6 W m−2. The lowest layers (temperature difference less than 4 K) produced on average a radiative effect of 0.8 W m−2.

Continental boundary layer clouds play an essential role in the climate system and are driven by processes linked to the land surface. The observation of boundary layer clouds with high-resolution satellite data can provide comprehensive insights into spatiotemporal patterns of land surface-driven modification of cloud occurrence, such as the diurnal variation of the occurrence of fog holes and cloud enhancements attributed to the impact of the urban heat island. High-resolution satellite-based cloud masking approaches are often based on locally-optimized thresholds that are compared against satellite-observed visible and/or infrared radiances to separate cloudy from clear-sky observations that can be affected by the local surface reflectance. Therefore, spatial differences in surface albedo, as found in and around urban areas or forests, can introduce spatial biases in the detected cloud cover that may impede the analysis of spatial pattern changes due to land surface influences. In this study, two approaches for cloud masking using the High Resolution Visible channel of the Spinning Enhanced Visible and Infrared Imager aboard Meteosat Second Generation are developed and validated for the region of Paris to show and improve applicability for analyses of urban effects on clouds. Firstly, the Local Empirical Cloud Detection Approach (LECDA) uses an optimized threshold to separate the distribution of visible reflectances into cloudy and clear sky for each individual pixel accounting for its locally specific brightness. Secondly, the Regional Empirical Cloud Detection Approach (RECDA) uses visible reflectance thresholds that are independent of surface reflection at the observed location. Validation against in-situ cloud fractions reveals that both approaches perform similarly with a Heidke Skill Score of 0.69 and 0.71, respectively. While the LECDA is representative for the widespread usage of locally-optimized approaches, comparison against RECDA reveals that the cloud masks obtained from LECDA can result in regional biases of +−5 % that are caused by the differences in surface reflectance. This makes the regional approach RECDA a more appropriate choice for the high-resolution satellite-based analysis of cloud cover changes over different surface types and the interpretation of locally induced cloud processes.

Marine low clouds cool the Earth's climate, with their coverage (LCC) being controlled by their environment. Here, an observed significant decrease of LCC in the northeastern Pacific over the past two decades is linked quantitatively to changes in cloud‐controlling factors. In a comparison of different statistical and machine learning methods, a decrease in the inversion strength and near‐surface winds, and an increase in sea surface temperatures (SSTs) are unanimously shown to be the main causes of the LCC decrease. While the decreased inversion strength leads to more entrainment of dry free‐tropospheric air, the increasing SSTs are shown to lead to an increased vertical moisture gradient that enhances evaporation when entrainment takes place. While the LCC trend is likely driven by natural variability, the trend‐attribution framework developed here can be used with any method in future analyses. We find the choice of predictors is more important than the method., Plain Language Summary: Marine low clouds efficiently cool the Earth's climate, and their prevalence is controlled by environmental factors. Here, a decrease of the cover of marine low clouds in the northeastern Pacific over the past 20 years is analyzed to attribute the trend to changes in environmental factors known to be important for low clouds. A decrease in the strength of the temperature inversion and an increase in sea surface temperatures (SSTs) are shown to be the main causes of the low‐cloud trend. The decreased inversion strength leads to more mixing in of dry air from above the clouds, leading to cloud evaporation. The increasing SSTs increase the atmospheric moisture levels near the surface more than above the cloud, enhancing evaporation when the mixing takes place. While the trend in low clouds is likely driven by natural variability rather than climate change, the analytical framework developed here can be deployed to attribute causes for trends with any statistical or machine learning method in the future. The analysis shows that the choice of environmental factors used for the analysis has a larger impact on the results than the method., Key Points: Significant decrease of low cloud cover in northeastern Pacific in last two decades. Increased vertical moisture gradient, decreased inversion strength, and winds drive low cloud trend. Good agreement between statistical and machine‐learning methods, predictor choice more important., Horizon 2020 research and innovation program, https://doi.org/10.24381/cds.f17050d7, https://doi.org/10.24381/cds.6860a573, https://dx.doi.org/10.5067/MODIS/MOD08_M3.061, http://dx.doi.org/10.5067/MODIS/MYD08_M3.061, https://doi.org/10.5067/TERRA-AQUA/CERES/EBAF-TOA_L3B004.1, https://doi.org/10.5281/zenodo.5747221

An intensive observation period was conducted in September 2017 in the central Namib, Namibia, as part of the project Namib Fog Life Cycle Analysis (NaFoLiCA). The purpose of the field campaign was to investigate the spatial and temporal patterns of the coastal fog that occurs regularly during nighttime and morning hours. The fog is often linked to advection of a marine stratus that intercepts with the terrain up to 100 km inland. Meteorological data, including cloud base height, fog deposition, liquid water path, and vertical profiles of wind speed/direction and temperature, were measured continuously during the campaign. Additionally, profiles of temperature and relative humidity were sampled during five selected nights with stratus/fog at both coastal and inland sites using tethered balloon soundings, drone profiling, and radiosondes. This paper presents an overview of the scientific goals of the field campaign; describes the experimental setup, the measurements carried out, and the meteorological conditions during the intensive observation period; and presents first results with a focus on a single fog event.

This study presents an approach for quantification of cloud-aerosol transition zone broadband longwave radiative effects at the top of the atmosphere (TOA) during daytime over the ocean, based on satellite observations and radiative transfer simulation. Specifically, we used several products from MODIS (Moderate Resolution Imaging Spectroradiometer) and CERES (Clouds and the Earth’s Radiant Energy System) sensors for identification and selection of CERES footprints with horizontally homogeneous transition zone and clear-sky conditions. For the selected transition zone footprints, radiative effect was calculated as the difference between the instantaneous CERES TOA upwelling broadband longwave radiance observations and corresponding clear-sky radiance simulations. The clear-sky radiances were simulated using the Santa Barbara DISORT Atmospheric Radiative Transfer model fed by the hourly ERA5 reanalysis (fifth generation ECMWF reanalysis) atmospheric and surface data. The CERES radiance observations corresponding to the clear-sky footprints detected were also used for validating the simulated clear-sky radiances. We tested this approach using the radiative measurements made by the MODIS and CERES instruments onboard Aqua platform over the south-eastern Atlantic Ocean during August 2010. For the studied period and domain, transition zone radiative effect (given in flux units) is on average equal to 8.0 ± 3.7 W m−2 (heating effect; median: 5.4 W m−2), although cases with radiative effects as large as 50 W m−2 were found.

In this study, the dissipation of fog and low stratus (FLS) over Europe is analyzed based on geostationary satellite data using logistic regression. The dissipation of FLS is a result of the interaction of complex physical processes and its timing has implications for environmental systems, traffic at land, sea and in the air, as well as for the production of solar energy. However, the timing of FLS dissipation, as well as its relationship to meteorological and land surface conditions has not been investigated quantitatively over a large spatial and temporal scale yet. In this study a 10-year FLS dissipation climatology is created using logistic regression. For this, a binary satellite-based FLS mask for each 15-minute interval from 2006-2015 over Europe, by Egli et al. 2017, is used. A logistic regression is applied to identify the dissipation time of each individual fog event from the binary FLS time series. Marked geographic FLS dissipation patterns are apparent, where FLS is found to dissipate earlier in elevated terrains and persist longer in valleys. Furthermore, the influence of different meteorological and land surface conditions on FLS dissipation are investigated.In the future, the presented approach will be extended to analyze FLS formation and its dependency on meteorological and land surface conditions.

Satellite observations are used in regional machine learning models to quantify sensitivities of marine boundary-layer clouds (MBLC) to aerosol changes.MBLCs make up a large part of the global cloud coverage as they are persistently present over more than 20% of the Earth’s oceans in the annual mean.They play an important role in Earth’s energy budget by reflecting solar radiation and interacting with thermal radiation from the surface, leading to a net cooling effect. Cloud properties and their radiative characteristics such as cloud albedo, horizontal and vertical extent, lifetime and precipitation susceptibility are dependent on environmental conditions. Aerosols in their role as condensation nuclei affect these cloud radiative properties through changes in the cloud droplet number concentration and subsequent cloud adjustments to this perturbation. However, the magnitude and sign of these effects remain among the largest uncertainties in future climate predictions.In an effort to help improve these predictions a machine learning approach in combination with observational data is pursued:Satellite observations from the collocated A-Train dataset (C3M) for 2006-2011 are used in combination with ECMWF atmospheric reanalysis data (ERA5) to train regional Gradient Boosting Regression Tree (GBRT) models to predict changes in key physical and radiative properties of MBLCs. The cloud droplet number concentration (Nd) and the liquid water path (LWP) are simulated for the eastern subtropical oceans, which are characterised by a high annual coverage of MBLC due to the occurrence of semi-permanent stratocumulus sheets. Relative humidity above cloud, cloud top height and temperature below the cloud base and at the surface are identified as important predictors for both Nd and LWP. The impact of each predictor variable on the GBRT model's output is analysed using Shapley values as a method of explainable machine learning, providing novel sensitivity estimates that will improve process understanding and help constrain the parameterization of MBLC processes in Global Climate Models.

In this contribution, a significant reduction of low-level marine clouds (LLCs) in the northeastern Pacific is found over a 20-year period in satellite observations and attributed to increasing sea surface temperatures (SSTs).LLCs play a key role for the Earth’s energy balance, however, their response to climatic changes is not clear, yet. Here, 20 years of Clouds and the Earth’s Radiant Energy System (CERES) cloud observations are analyzed together with reanalysis data sets in multivariate-regression and machine-learning frameworks to link an observed decrease of LLCs in the subtropical northern Pacific to changes in environmental factors. In the analyses, the observed LCC trend is explained almost exclusively by an increase in SSTs, but counteracted to some extent by increased low-level moisture availability. The influence of other factors such as estimated inversion strength, local winds and aerosols is investigated in the statistical frameworks but found to be negligible when compared to the effect of SST changes. The results provide observational evidence for the low-cloud feedback that back model findings of reduced LCC due to increased SSTs in a changing climate.

Aerosols are a critical component of the climate system and a risk to human health. Here, the lockdown response to the coronavirus outbreak is used to analyse effects of dramatic reduction in anthropogenic aerosol sources on satellite-retrieved aerosol optical depth (AOD). A machine learning model is applied to estimate daily AOD during the initial lockdown in China in early 2020. The model uses information on aerosol climatology, geography and meteorological conditions, and explains 69% of the day-to-day AOD variability. A comparison of model-expected and observed AOD shows that no clear, systematic decrease in AOD is apparent during the lockdown in China. During March 2020, regional AOD is observed to be significantly lower than expected by the machine learning model in some coastal regions of the North China Plains and extending to the Korean peninsula. While this may possibly indicate a small lockdown effect on regional AOD, and potentially pointing trans-boundary effects of the lockdown measures, due to uncertainties associated with the method and the limited sample sizes, this AOD decrease cannot be unequivocally attributed to reduced anthropogenic emissions. Climatologically expected AOD is compared to a weather-adjusted expectation of AOD, indicating that meteorological influences have acted to significantly increase AOD during this time, in agreement with recent literature. The findings highlight the complexity of aerosol variability and the challenges of observation-based attribution of columnar aerosol changes.

Air pollution, in particular high concentrations of particulate matter smaller than 1 µm in diameter (PM1), continues to be a major health problem, and meteorology is known to substantially influence atmospheric PM concentrations. However, the scientific understanding of the ways in which complex interactions of meteorological factors lead to high-pollution episodes is inconclusive. In this study, a novel, data-driven approach based on empirical relationships is used to characterize and better understand the meteorology-driven component of PM1 variability. A tree-based machine learning model is set up to reproduce concentrations of speciated PM1 at a suburban site southwest of Paris, France, using meteorological variables as input features. The model is able to capture the majority of occurring variance of mean afternoon total PM1 concentrations (coefficient of determination (R2) of 0.58), with model performance depending on the individual PM1 species predicted. Based on the models, an isolation and quantification of individual, season-specific meteorological influences for process understanding at the measurement site is achieved using SHapley Additive exPlanation (SHAP) regression values. Model results suggest that winter pollution episodes are often driven by a combination of shallow mixed layer heights (MLHs), low temperatures, low wind speeds, or inflow from northeastern wind directions. Contributions of MLHs to the winter pollution episodes are quantified to be on average ∼5 µg/m3 for MLHs below m a.g.l. Temperatures below freezing initiate formation processes and increase local emissions related to residential heating, amounting to a contribution to predicted PM1 concentrations of as much as ∼9 µg/m3. Northeasterly winds are found to contribute ∼5 µg/m3 to predicted PM1 concentrations (combined effects of u- and v-wind components), by advecting particles from source regions, e.g. central Europe or the Paris region. Meteorological drivers of unusually high PM1 concentrations in summer are temperatures above ∼25 ∘C (contributions of up to ∼2.5 µg/m3), dry spells of several days (maximum contributions of ∼1.5 µg/m3), and wind speeds below ∼2 m/s (maximum contributions of ∼3 µg/m3), which cause a lack of dispersion. High-resolution case studies are conducted showing a large variability of processes that can lead to high-pollution episodes. The identification of these meteorological conditions that increase air pollution could help policy makers to adapt policy measures, issue warnings to the public, or assess the effectiveness of air pollution measures.

The Namibian coast line experiences fog when moist air from the southeast Atlantic is advectedover the desert landscape. We run the WRF model with the Thompson (2008) microphysics scheme,with a default CCN number concentration of 100 cm-1, to forecast next day fog over the Namibdesert. Model output of liquid water content at the lowest level in the atmosphere is used torepresent fog and is evaluated against in situ observations of visibility and satellite products offog/low stratus. Preliminary results indicate that the model captures the spatial pattern of fogexcellently, however, the model over predicts fog occurrence. These results serve as the control runfor a future model sensitivity study.

In this study the distribution of fog and low stratus (FLS) relative to land cover and meteorological conditions in continental Europe is investigated using a state-of-the-art machine learning technique and geostationary satellite data. While analyses of the spatial and temporal patterns of FLS exist, the relationships to land cover and meteorological conditions have not been studied explicitly, quantitatively and on a continental scale.The machine learning model is built using daily means of a FLS dataset from Egli et al. (2017) as the predictand, and different land surface and meteorological parameters as predictors over continental Europe from 2006-2015. The application of a machine learning approach provides the ability to explicitly, synchronously and quantitatively link suspected determinants of FLS to its occurrence.The model shows good performance with R² values ranging between 0.5 and 0.9, depending on model grid size, season and further settings such as the exclusion of low pressure values and subtraction of seasonality. It is thus able to adequately represent the dynamics that drive FLS development. Based on a systematic analysis of this model, the most important features for FLS prediction are mean surface pressure, wind speed, and FLS on the previous day. High mean surface pressure, high FLS cover on the previous day, low evapotranspiration, wind speed and land surface temperature lead to higher predicted FLS values.Generally the results show that it is possible to predict FLS occurrence over continental Europe using meteorological as well as land surface parameters with good performance indicating the benefits of using machine learning in the analysis of non-linear, multivariate systems such as the land-atmosphere system. Further studies will integrate the machine learning model into a land surface based model grid and implement fog and low cloud properties as predictands.

This contribution presents a technique for the machine-learning-based retrieval of cloud liquid water path. Cloud effects are among the major uncertainties in climate models for estimating and predicting the Earth’s energy budget. The study of cloud processes requires information on cloud physical properties, such as the liquid water path (LWP), which is commonly retrieved from satellite sensors using look-up table approaches. However, the accuracy of LWP varies temporally and spatially, also due to assumptions inherent in any physical retrieval. The aim of this study is to improve the accuracy of LWP and analyze quantitatively the accuracy and its errors. To this end, a statistical LWP retrieval was developed using spectral information from geostationary satellite channels (Meteosat Spinning-Enhanced Visible and Infrared Imager, SEVIRI), and satellite viewing geometry. The machine-learning method chosen is gradient-boosted regression trees (GBRTs), which is an ensemble of decision trees but more effective than traditional tree-based models. This study reports on first results, as well as a comparison between the GBRT-derived LWP estimates and those from the SEVIRI-based products of the Climate Monitoring Satellite Application Facility (CM-SAF, CLAAS-A2), as well as MODIS products. We use case studies for individual in-situ measurement sites in Europe under varying meteorological conditions to determine the factors influencing LWP retrieval quality.

This contribution presents new findings on synoptic-scale mechanisms that control the day-to-day variability of fog and low clouds (FLCs) in the Namib region.FLCs are a defining element of the Namib-region climate and a crucial source of water for many species and ecosystems. Still, little is known on the processes driving Namib-region FLCs, in large part due to the very sparse observational records. Specifically, there is an ongoing debate in the scientific literature concerning the relevance of different mechanisms responsible for fog formation in the region. In this contribution, a new long-term satellite-based data set of FLC occurrence is used in conjunction with reanalysis data and backtrajectories to systematically analyze dynamical and thermodynamical differences between days with and without FLCs in the central Namib during two different seasons. The main findings are:Central-Namib FLCs are nearly always associated with the advection of marine-boundary-layer air masses. The variability of the overall FLC coverage in the central Namib is largely driven by dynamics at the synoptic scale. Seasonally different synoptic-scale mechanisms determine the probability of the occurrence of FLCs in the central Namib. The findings lead to a better understanding of Namib-region FLCs and help broaden the understanding of low clouds along the southwestern African coastline and the southeast Atlantic.

The quantification of factors leading to harmfully high levels of particulate matter (PM) remains challenging. This study presents a novel approach using a statistical model that is trained to predict hourly concentrations of particles smaller than 10 μm (PM10) by combining satellite-borne aerosol optical depth (AOD) with meteorological and land-use parameters. The model is shown to accurately predict PM10 (overall R 2 = 0.77, RMSE = 7.44 μg/m 3) for measurement sites in Germany. The capability of satellite observations to map and monitor surface air pollution is assessed by investigating the relationship between AOD and PM10 in the same modeling setup. Sensitivity analyses show that important drivers of modeled PM10 include multiday mean wind flow, boundary layer height (BLH), day of year (DOY), and temperature. Different mechanisms associated with elevated PM10 concentrations are identified in winter and summer. In winter, mean predictions of PM10 concentrations >35 μg/m 3 occur when BLH is below ∼500 m. Paired with multiday easterly wind flow, mean model predictions surpass 40 μg/m 3 of PM10. In summer, PM10 concentrations seemingly are less driven by meteorology, but by emission or chemical particle formation processes, which are not included in the model. The relationship between AOD and predicted PM10 concentrations depends to a large extent on ambient meteorological conditions. Results suggest that AOD can be used to assess air quality at ground level in a machine learning approach linking it with meteorological conditions.

Fog is a defining characteristic of the climate of the Namib Desert and its water and nutrient input are important for local ecosystems. In part due to sparse observation data, the local mechanisms that lead to fog occurrence in the Namib are not yet fully understood, and to date, potential synoptic-scale controls have not been investigated. In this study, a recently established 14-year data set of satellite observations of fog and low clouds in the central Namib is analyzed in conjunction with reanalysis data to identify typical synoptic-scale conditions associated with fog and low-cloud occurrence in the central Namib during two seasons that characterize seasonal fog variability. It is found that during both seasons, mean sea level pressure and geopotential height at 500 hPa differ significantly between fog/low-cloud and clear days, with patterns indicating seasonally different synoptic-scale disturbances on fog and low-cloud days: cut-off lows during September, October, and November, and breaking Rossby waves during April, May, and June. These regularly occurring disturbances increase the probability of fog and low-cloud occurrence in the central Namib in two main ways: 1) an anomalously dry free troposphere in the coastal region of the Namib leads to stronger longwave cooling, especially over the ocean, facilitating low-cloud formation, and 2) local wind systems are modulated, leading to an onshore anomaly of marine boundary-layer air masses. This is consistent with air mass backtrajectories and a principal component analysis of spatial wind patterns that point to advected marine boundary- layer air masses on fog and low-cloud days, whereas subsiding continental air masses dominate on clear days. Large-scale free-tropospheric moisture transport into southern Africa seems to be a key factor modulating the onshore advection of marine boundary-layer air masses during April, May, and June, as the associated increase in greenhouse gas warming and thus surface heating is observed to contribute to a continental heat low anomaly. A statistical model is trained to discriminate between fog/low-cloud and clear days based on large-scale mean sea level pressure fields. The model accurately predicts fog and low-cloud days, illustrating the importance of large-scale pressure modulation and advective processes. It can be concluded that Namib-region fog is predominantly of advective nature, but also facilitated by increased radiative cooling. Seasonally different manifestations of synoptic-scale disturbances act to modify its day-to-day variability and the balance of mechanisms leading to its formation. The results are the basis for a new conceptual model on the synoptic-scale mechanisms that control fog and low clouds in the Namib Desert, and will guide future studies of coastal fog regimes.

The formation and development of fog and low stratus clouds (FLS) depend on meteorological and land surface conditions and their interactions with each other. While analyses of temporal and spatial patterns of FLS in Europe exist, the interactions between FLS determinants underlying them have not been studied explicitly and quantitatively at a continental scale yet. In this study, a state-of-the-art machine learning technique is applied to model FLS occurrence over continental Europe, using meteorological and land surface parameters from geostationary satellite and reanalysis data. Spatially explicit model units are created to test for spatial and seasonal differences in model performance and FLS sensitivities to changes in predictors, and effects of different data preprocessing procedures are evaluated. The statistical models show good performance in predicting FLS occurrence during validation, with R2>0.9 especially in winter high pressure situations.The predictive skill of the models seems to be dependent on data availability, data preprocessing, time period, and geographic characteristics. It is shown that atmospheric proxies are more important determinants of FLS presence than surface characteristics, in particular mean sea level pressure, near-surface wind speed and evapotranspiration are crucial, together with FLS occurrence on the previous day. Higher wind speeds, higher land surface temperatures and higher evapotranspiration tend to be negatively related to FLS. Spatial patterns of feature importance show the dominant influence of mean sea level pressure on FLS occurrence throughout the central European domain. When only high pressure situations are considered, wind speed (in the western study region) and evapotranspiration (in the eastern study region) gain importance, highlighting the influence of moisture advection on FLS occurrence in the western parts of the central European domain. This study shows that FLS occurrence can be accurately modeled using machine learning techniques in large spatial domains based on meteorological and land surface predictors. The statistical models used in this study provide a novel analysis tool for investigating empirical relationships in the FLS – land surface system and possibly infer processes.

During the West African summer monsoon season, extended nocturnal stratiform low-level clouds (LLCs) frequently form in the atmospheric boundary layer over southern West Africa and persist long into the following day affecting the regional climate. A unique data set was gathered within the framework of the Dynamics–Aerosol–Chemistry–Cloud Interactions in West Africa (DACCIWA) project, which allows, for the first time, for an observational analysis of the processes and parameters crucial for LLC formation. In this study, in situ and remote sensing measurements from radiosondes, ceilometer, cloud radar and energy balance stations from a measurement site near Savè in Benin are analyzed amongst others for 11 nights. The aim is to study LLC characteristics, the intranight variability of boundary layer conditions and physical processes relevant for LLC formation, as well as to assess the importance of these processes. Based on the dynamic and thermodynamic conditions in the atmospheric boundary layer we distinguish typical nocturnal phases and calculate mean profiles for the individual phases. A stable surface inversion, which forms after sunset, is eroded by differential horizontal cold air advection with the Gulf of Guinea maritime inflow, a cool air mass propagating northwards from the coast in the late afternoon and the evening, and shear-generated turbulence related to a nocturnal low-level jet. The analysis of the contributions to the relative humidity changes before the LLC formation reveals that cooling in the atmospheric boundary layer is crucial to reach saturation, while specific humidity changes play a minor role. We quantify the heat budget terms and find that about 50 % of the cooling prior to LLC formation is caused by horizontal cold air advection, roughly 20 % by radiative flux divergence and about 22 % by sensible heat flux divergence in the presence of a low-level jet. The outcomes of this study contribute to the development of a conceptual model on LLC formation, maintenance and dissolution over southern West Africa.

Fog is an essential component of Namib-region ecosystems. Current knowledge on Namib-region fog patterns and processes is limited by a lack of coherent observations in space and time. In this study, data from multiple satellite platforms and station measurements paint a coherent picture of the spatiotemporal dynamics of fog and low cloud (FLC) distribution. It is found that observed seasonal FLC patterns derived from satellite observations differ from fog measurements at coastal station locations, whereas they agree further inland. This is linked to an observed seasonal cycle in the vertical structure of FLCs that determines the probability of low-level clouds touching the ground. For the first time, these observations are complemented by spatially coherent statistics concerning the diurnal cycle of FLCs using geostationary satellite data. The average timing of the start of the diurnal FLC cycle is found to strongly depend on the distance to the coastline (correlation ≈0.85 north of 25∘ S), a clear indication of dominant advective processes. In the central Namib, FLCs typically occur 2–4 h later than in other coastal regions, possibly due to local advection patterns. The findings lead to a new conceptual model of the spatiotemporal dynamics of fog and low clouds in the Namib.

Understanding the processes that determine low-cloud properties and aerosol–cloud interactions (ACIs) is crucial for the estimation of their radiative effects. However, the covariation of meteorology and aerosols complicates the determination of cloud-relevant influences and the quantification of the aerosol–cloud relation. This study identifies and analyzes sensitivities of cloud fraction and cloud droplet effective radius to their meteorological and aerosol environment in the atmospherically stable southeast Atlantic during the biomass-burning season based on an 8-day-averaged data set. The effect of geophysical parameters on clouds is investigated based on a machine learning technique, gradient boosting regression trees (GBRTs), using a combination of satellite and reanalysis data as well as trajectory modeling of air-mass origins. A comprehensive, multivariate analysis of important drivers of cloud occurrence and properties is performed and evaluated. The statistical model reveals marked subregional differences of relevant drivers and processes determining low clouds in the southeast Atlantic. Cloud fraction is sensitive to changes of lower tropospheric stability in the oceanic, southwestern subregion, while in the northeastern subregion it is governed mostly by surface winds. In the pristine, oceanic subregion large-scale dynamics and aerosols seem to be more important for changes of cloud droplet effective radius than in the polluted, near-shore subregion, where free tropospheric temperature is more relevant. This study suggests the necessity to consider distinct ACI regimes in cloud studies in the southeast Atlantic.

This study presents the first detailed observational analysis of the complete diurnal cycle of stratiform low-level clouds (LLC) and involved atmospheric processes over southern West Africa. The data used here were collected during the comprehensive DACCIWA (Dynamics-Aerosol-Chemistry-Cloud-Interactions in West Africa) ground-based campaign, which aimed at monitoring LLC characteristics and capturing the wide range of atmospheric conditions related to the West African monsoon flow. In this study, in-situ and remote sensing measurements from the supersite near Savè (Benin) collected during a typical day, which is characterized by the onset of a nocturnal low-level jet (NLLJ) and the formation of LLC, are analyzed. The associated dynamic and thermodynamic conditions allow the identification of five different phases of the LLC diurnal cycle: the Stable, Jet, Stratus I, Stratus II and Convective phase. The analysis of relative humidity tendency shows that cooling is a dominant process for LLC formation, which leads to a continuous increase of relative humidity at a maximum rate of 6 % per hour, until finally saturation is reached and LLC form with a cloud-base height near the height of NLLJ maximum. Results of heat budget analysis illustrate that horizontal cold air advection, related to the maritime inflow, which brings the cool maritime air mass and a prominent NLLJ wind profile, has the dominant role on the observed strong cooling of −1.2 K per hour during the Jet phase. The contribution from horizontal cold advection is quantified to be up to 72 %, while radiative cooling and sensible heat flux divergence contribute with 16 and 12 %, respectively, to the observed heat budget below the NLLJ maximum. After the LLC form (Stratus phase I and II), turbulent mixing is an important factor leading to the cooling below the cloud base, while strong radiative cooling at the cloud top helps to maintain thick stratus.

Fog and low clouds (FLC) are a typical feature along the southwestern African coast, especially in the central Namib, where fog constitutes a valuable resource of water for many ecosystems. In this study, a novel algorithm to detect FLC over land from geostationary satellite data using only infrared observations is presented. The algorithm is the first of its kind as it is stationary in time and thus able to reveal a detailed view into the diurnal and spatial patterns of FLC in the Namib region. A validation against net radiation measurements from a station network in the central Namib reveals a high overall accuracy with a probability of detection of 94 %, a false alarm rate of 12 % and an overall correctness of classification of 97 %. The average timing and persistence of FLC seem to depend on the distance to the coast, suggesting that the region is dominated by advection-driven FLC. While the algorithm is applied to study Namib-region fog and low clouds, it is designed to be transferable to other regions and can be used to retrieve long-term data sets.

This study investigates the impact of air-mass origin and dynamics on cloud property changes in the South-East Atlantic (SEA) during the biomass-burning season. The understanding of clouds and their determinants at different scales is important for constraining the Earth's radiative budget, and thus prominent in climate-system research. In this study, the thermodynamically stable SEA stratocumulus cover is observed not only as the result of local environmental conditions but also as connected to large-scale meteorology by the often neglected but important role of spatial origins of air masses entering this region. In order to assess to what extent cloud properties are impacted by aerosol concentration, air mass history, and meteorology, a HYSPLIT cluster analysis is conducted linking satellite observations of cloud properties (SEVIRI), information on aerosol species (MACC) and meteorological context (ERA-Interim reanalysis) to air-mass clusters. It is found that a characteristic pattern of air-mass origins connected to distinct synoptical conditions leads to marked cloud property changes in the southern part of the study area. Long-distance air masses are related to midlatitude weather disturbances that affect the cloud microphysics, especially in the southwestern subdomain of the study area. Changes in cloud effective radius are consistent with a boundary layer deepening and changes in LTS. In the southeastern subdomain cloud cover is controlled by a generally higher LTS, while air-mass origin plays a minor role. This study leads to a better understanding of the dynamical drivers behind observed stratocumulus cloud properties in the SEA and frames potentially interesting conditions for aerosol-cloud interactions.

Aerosol-cloud interaction is a key component of the Earth's radiative budget and hydrological cycle, but many facets of its mechanisms are not yet fully understood. In this study, global satellite-derived aerosol and cloud products are used to identify at what aerosol loadings cloud droplet size shows the greatest sensitivity to changes in aerosol loading (ACSmax). While on average, cloud droplet size is most sensitive at relatively low aerosol loadings, distinct spatial and temporal patterns exist. Possible determinants for these are identified with reanalysis data. The magnitude of ACSmax is found to be constrained by the total columnar water vapor. Seasonal patterns of water vapor are reflected in the seasonal patterns of ACSmax. Also, situations with enhanced turbulent mixing are connected to higher ACSmax, possibly due to intensified aerosol activation. Of the analyzed aerosol species, dust seems to impact ACSmax the most, as dust particles increase the retrieved aerosol loading without substantially increasing the concentration of cloud condensation nuclei.