Marine shallow clouds exhibit a variety of patterns, which have been the subject of extensive research. In this study, we introduce a new approach for their continuous classification using Moderate Resolution Imaging Spectroradiometer (MODIS) true color imagery. This method utilizes a ternary phase space derived from the clouds’ morphological properties, and defined by three end members: closed-cell and open-cell stratocumulus, and Cumulus clouds. We show that a phase space defined by cloud fraction and the number of voids (clear-sky areas surrounded by clouds) successfully captures the inherent mixture of patterns, providing a more nuanced and objective classification scheme. Focusing in two regions located in the North and South Eastern Pacific, we show that closed cells dominate the eastern parts of the study regions, while mixed states, those not purely representing any end member, dominate the western regions. The mixed states exhibit diverse morphologies, showing that this approach anchors classic patterns within a unified space while revealing the richness and variability of intermediate states. To explore and validate this framework, we combine MODIS cloud properties with radiative fluxes from the Clouds and the Earth’s Radiant Energy System (CERES). Examination of cloud microphysical properties shows coherent gradient structures within the phase space, lending support to this classification framework. Importantly, these microphysical distinctions are reflected in corresponding variations in radiative behavior across the spectrum of patterns. Plain Language Summary Marine shallow clouds cover vast parts of the subtropical oceans and help cool the planet by reflecting sunlight, yet their properties remain a major uncertainty in climate projections. These clouds exhibit different organizational patterns that differ in their properties and radiative behavior. In this study, rather than sorting each satellite image into a few fixed categories, we map every scene based on its morphological properties. This creates a simple triangular diagram whose corners represent three familiar patterns: closedcells and open-cells in stratocumulus and shallow cumulus, which we refer to as end members. A scene’s position within the triangle shows the relative contributions (mixture) of the three defining patterns. This diagram emerges directly from the morphological properties of the clouds. We tested this approach over the northeastern and southeastern Pacific using visual classification of satellite images. The results reveal that closed cells and mixed patterns are most abundant, and occur in distinct geographical regions. Examination of the Satellite cloud products distinguishes between areas in the diagram, reproducing known cloud properties and providing new insights into mixed states. Topof-the-atmosphere shortwave flux measurements show that closed cells dominate scene reflectance, while mixed states also contribute significantly to the overall radiative effect.

Modeling emergent, multiscale patterns in complex systems remains a persistent interdisciplinary challenge, particularly when deciphering transient, highly coupled dynamics from severely limited data. To overcome the inherent spectral sparsity of standard finite-dimensional stochastic models, we introduce Multilayered Stochastic Hierarchical Delay Models (MSHDMs). By embedding dynamics within a hierarchical delay structure, MSHDMs leverage an infinite-dimensional phase space to engineer the high spectral density required for generating complex amplitude-frequency modulation (AM-FM) dynamics, avoiding data-heavy neural network parameterizations. To reliably calibrate these sensitive structures from short observational records, we deploy a hybrid offline-online Bayesian optimization framework. Bypassing the failure points of traditional trajectory matching, our algorithm autonomously learns optimal latent coordinates and continuous fractional delays by strictly enforcing spectral consistency against the empirical Global Wavelet Spectrum. Applying this methodology to high-resolution satellite observations of continental cloud fields, the resulting stochastic emulator captures the full spatiotemporal coherence of the turbulent system using just six hyperparameters. The wavelet scalograms demonstrate how the model natively generates emergent wave-packet dynamics and cross-scale energy cascades, recovering semidiurnal, mesoscale, and individual cloud timescales. Supported by rigorous mathematical foundations and robust data-driven calibrations, MSHDMs thus provide a highly compressible, general-purpose tool for resolving latent AM-FM beat patterns across diverse disciplines.

Diurnal variations in sea surface temperature (SST) influence the atmospheric boundary layer and the hydrological cycle. Using a decade of satellite data, we identify global patterns of diurnal SST variability, capturing both warming and cooling phases. We highlight diurnal cooling as a distinct and previously underrecognized phenomenon, accounting for over 38% of observed cases. By focusing on extreme cooling and warming events, defined as diurnal SST changes exceeding +/- 3.16 degrees C $\pm 3.16{}<^>{\circ}\mathrm{C}$ from the mean, we explore the spatial distribution, seasonality, and mechanisms driving transitions between three SST states: diurnal warming, a balanced state, and diurnal cooling. These states correspond to distinct upper-ocean mixing regimes: minimal, neutral, and strong, each shaped by atmospheric forcing, including wind stress, cloud cover, and precipitation. In the tropics and mid-latitudes, extreme warming events dominate, occurring primarily during the transition and summer seasons under calm winds, clear skies, and little to no rain. In contrast, high-latitude regions are characterized by frequent diurnal cooling during winter, with passing storms emerging as key modulators of SST variability. Storms that occur during the day typically trigger strong cooling, while those passing at night can occasionally result in warming. By integrating seasonal context and focusing on extreme events, this study provides new insights into the atmospheric drivers of diurnal SST variability, an important step in constructing and tuning models that capture diurnal layer dynamics. These findings have implications for understanding energy budgets, air-sea interactions, and feedbacks in the coupled climate system.

Shallow, sparse, non-precipitating convective clouds forming over the ocean are considered among the least organized cloud fields. The formation mechanism of these clouds is associated with random, local perturbations that create buoyant parcels. Their sparseness suggests no or very weak interactions between clouds. Here, we show that such clouds form within a well-organized, stable, dense mesh of convective cells that operate continuously, independent of the presence of visible clouds.

Abstract Cloud organization impacts the radiative effects and precipitation patterns of the cloud field. Deviating from randomness, clouds exhibit either clustering or a regular grid structure, characterized by the spacing between clouds and the cloud size distribution. The two measures are coupled but do not fully define each other. Here, we present the deviation from randomness of the cloud‐ and void‐chord length distributions as a measure for both factors. We introduce the LvL representation and an associated 2D score that allow for unambiguously quantifying departure from well‐defined baseline randomness in cloud spacing and sizes. This approach demonstrates sensitivity and robustness in classifying cloud field organization types. Its delicate sensitivity unravels the temporal evolution of a single cloud field, providing novel insights into the underlying governing processes.

&lt;p&gt;Prevalent over the world&amp;#8217;s oceans and continents, shallow clouds still comprise a main aspect of the uncertainty related to cloud feedback and climate sensitivity. Compared to shallow clouds over the ocean, confined to specific marine environments, shallow cumulus (Cu) over land occur in diverse locations throughout the globe.&lt;/p&gt; &lt;div class=&quot;page&quot; title=&quot;Page 1&quot;&gt; &lt;div class=&quot;layoutArea&quot;&gt; &lt;div class=&quot;column&quot;&gt; &lt;p&gt;Motivated by an intriguing observation regarding the universality of continental shallow Cu fields regardless of their geographical location, we explore their similarities. We combine satellite observations, along with machine learning classification and numerical modelling to show that these cloud fields share many important properties, such as the patterns they form and their tendency to form over and near forests and vegetated lands, thus termed &lt;em&gt;greenCu&lt;/em&gt;.&lt;/p&gt; &lt;p&gt;Moreover, we show that in spite of their occurrence in different climatic regions, from the tropics to mid- and high-latitudes, greenCu fields are associated with similar large-scale meteorological conditions.&lt;/p&gt; &lt;/div&gt; &lt;/div&gt; &lt;/div&gt;

Shallow cloud fields exhibit different patterns, such as closed or open hexagonal cells and cloud streets. These patterns play a key role in determining the cloud fields' radiative effects, thereby affecting the climate. Here, we show that a large subset of shallow cloud fields forms organized, mesoscale-sized, regular patterns that persist for extended times. It emanates from the steady state of the underlying rigid configuration of convection cells. From a climate perspective, in a sea of cloud complexity, the convective steady-state provides an "island of simplicity." The convective steady state can be parametrized in climate models to better capture the feedback of such cloud fields in a warming climate.

The emergence of organized multiscale patterns resulting from convection is ubiquitous, observed throughout different cloud types. The reproduction of such patterns by general circulation models remains a challenge due to the complex nature of clouds, characterized by processes interacting over a wide range of spatio-temporal scales. The new advances in data-driven modeling techniques have raised a lot of promises to discover dynamical equations from partial observations of complex systems. This study presents such a discovery from high-resolution satellite datasets of continental cloud fields. The model is made of stochastic differential equations able to simulate with high fidelity the spatio-temporal coherence and variability of the cloud patterns such as the characteristic lifetime of individual clouds or global organizational features governed by convective inertia gravity waves. This feat is achieved through the model's lagged effects associated with convection recirculation times, and hidden variables parameterizing the unobserved processes and variables.

One of the major sources of uncertainty in climate prediction results from the limitations in representing shallow cumulus (Cu) in models. Recently, a class of continental shallow convective Cu was shown to share distinct morphological properties and to emerge globally mostly over forests and vegetated areas, thus named greenCu. Using machine‐learning supervised classification, we identify greenCu fields over three regions, from the tropics to mid‐ and higher‐latitudes, and establish a novel satellite‐based data set called greenCuDb, consisting of 1° × 1° sized, high‐resolution MODIS images. Using greenCuDb in conjunction with ERA5 reanalysis data, we create greenCu composites for different regions and reveal that greenCu are driven by similar large‐scale meteorological conditions, regardless of their geographical locations throughout the world's continents. These conditions include distinct profiles of temperature, humidity and large‐scale vertical velocity. The boundary layer is anomalously warm and moderately humid, and is accompanied by a strong large‐scale subsidence in the free troposphere.

Warm convective clouds play a key role in the Earth’s radiative and water budgets. Nonetheless, they still comprise the largest source of uncertainty in climate model’s prediction of cloud feedback and climate sensitivity. The latter might be affected by the variety of patterns that warm convective clouds form on the mesoscale, an effect which is largely uninvestigated, and even more so over land. A large subset of continental shallow convective cumulus (Cu) fields was shown to have unique spatial properties and to form mostly over forests and vegetated areas thus referred to as green Cu. Green Cu fields form organized mesoscale patterns, yet the underlying mechanisms, as well as the time variability of these patterns, are still lacking understanding. In this work, we characterize the organization of green Cu in space and time, by using data-driven organization metrics, and by decomposing the high-resolution GOES–16 data using an Empirical Orthogonal Function (EOF) analysis. We extract and quantify modes of organization present in a green Cu field, during the course of a day. The EOF decomposition shows the field's key organization features such as cloud streets, and it also reveals hidden ones, as the propagation of gravity waves (GW), and the development of a highly ordered grid of clouds that extends over hundreds of kilometers, over a time span that scales as the field's lifetime. We then use cloud fields that were reconstructed from different subgroups of modes to quantify the cloud street's wavelength and aspect ratio, as well as the GW dominant period.

A subset of continental shallow convective cumulus (Cu) cloud fields has been shown to have distinct spatial properties and to form mostly over forests and vegetated areas, thus referred to as "green Cu" (Dror et al., 2020). Green Cu fields are known to form organized mesoscale patterns, yet the underlying mechanisms, as well as the time variability of these patterns, are still lacking understanding. Here, we characterize the organization of green Cu in space and time, by using data-driven organization metrics and by applying an empirical orthogonal function (EOF) analysis to a high-resolution GOES-16 dataset. We extract, quantify, and reveal modes of organization present in a green Cu field, during the course of a day. The EOF decomposition is able to show the field's key organization features such as cloud streets, and it also delineates the less visible ones, as the propagation of gravity waves (GWs) and the emergence of a highly organized grid on a spatial scale of hundreds of kilometers, over a time period that scales with the field's lifetime. Using cloud fields that were reconstructed from different subgroups of modes, we quantify the cloud street's wavelength and aspect ratio, as well as the GW-dominant period.

Warm convective clouds play a significant role in the earth's energy and water budgets. However, they still pose a challenge in climate research as their feedback to predicted thermodynamic changes is highly uncertain and considered critical to the overall climate system's response. The focus of this study is continental, organized shallow convective clouds that, although they are spread globally and form in a variety of environments, seem to have common properties. One of these properties seems to be their preferred formation over vegetated areas, thus referred hereafter as green Cu. In this article, we present new observations of emerging universality and explore them using a method that combines fine- and coarse-resolution remote-sensing data sets. First, we use Moderate Resolution Imaging Spectroradiometer (MODIS) true-color images to visually classify cloud fields into different classes and identify green Cu fields. We show that the level and type of organization and the properties of these fields (e.g., cloud size distribution and cloud fraction) are similar throughout the world, regardless of their location. Second, we match the corresponding MODIS level-3 cloud properties to the identified cloud classes, and based on this data sets statistics, we develop a detection method for green Cu along ten years of measurements (2003-2012). We examine the geographical distribution and seasonality of this class and show that these fields are highly abundant over many continental areas and indeed mostly in the vicinity of vegetated regions.

The discovery of dynamical equations governing time-evolving observations issued from a complex dynamical system requires a statistical formulation, since information concerning neglected variables or unobserved degrees of freedom is necessarily incomplete. At the same time, an equation that is closed within a small number of observables is often obtained only by approximations. Thus, the relevance of approximations must be understood before any attempt to derive a closed set of equations. This is where closure formalisms are of usefulness and the corresponding mathematical structures serve as a guide for knowing what to approximate. Many such formalisms are available from turbulence theory, quantum field theories, to statistical physics. Observables of interest often include response functions, spectra of fluctuations, or low-order moments, etc. These quantities correspond to moments of the full probability density function (PDF), the mother of all system's statistics but itself beyond the reach of standard closure theories, except in special cases. Yet, to have, for a given choice of observables, a (good) class of closure models able to produce out-of-sample reliable occurrences, is of prime importance. When derived on a firm basis, such closure models may indeed allow for analyzing in greater details certain features of a given phenomenon for which available data are limited, by e.g. drawing a large ensemble of statistical emulations of this phenomenon, from the closure model. This is the goal that will be pursued here for a special but common class of clouds, namely continental shallow cumulus (Cu) that can be found from low to mid/high latitudes, across a wide range of scales, and that play a growing role in the Earth's radiative budget. These clouds typically organize through a variety of patterns such as cloud streets, clusters, or mesoscale arcs. Based on observables suitably extracted from high-resolution satellite observations, it will be shown that the efficient learning of hidden, stochastic, variables along with their interaction laws with the observed variables is key for the derivation of relevant stochastic data-driven models. To do so, our approach will rely on the Mori-Zwanzig closure theory to guide the search of the constitutive elements, on one hand, while their learning will exploit recent advances in data-driven stochastic modeling techniques, on the other. As a byproduct, dynamical equations involving a few variables are learned from high-resolution satellite observations of continental shallow Cu. These equations will be shown to take the form of differential equations that include lagged effects, and are driven by a spatially correlated white noise. It will be finally shown that the combined effects of these terms allow to generate easily statistical ensembles of shallow Cu that exhibit a wide range of spatio-temporal variability while displaying consistency with the shallow Cu's organizational and multiscale features, from observations. Based on such large ensembles, new physical insights are attainable and their interpretation will be discussed. This work is supported by the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation program [Grant Agreement No. 810370].

General comments This Supporting Information (SI) contains detailed explanations regarding (i) the gamma correction applied to the reflectance images prior to the analyses, and (ii) the calculation and derivation of spatio-temporal characteristics of the gravity waves (GW) and the cloud streets (Text S1). An explanation regarding the identification of the different types of clouds exhibited in 10 the region of interest (ROI) (Text S2). It also presents the caption for the video (Movie S1; uploaded separately) showing the diurnal evolution of the cloud field. Additionally, it includes Figs. S1,S2, showing the diurnal evolution of cloud properties, the topography of the examined ROI (Fig. S3), and a figure showing the early–morning clouds (Fig. S4). Finally, figures showing EOF3–EOF8 (Fig. S5), EOF9–EOF14 (Fig. S6) and EOF15–EOF20 (Fig. S7) are also included. 15

Aerosol size distribution has major effects on warm cloud processes. Here, we use newly acquired marine aerosol size distributions (MSDs), measured in situ over the open ocean during the Tara Pacific expedition (2016–2018), to examine how the total aerosol concentration (Ntot) and the shape of the MSDs change warm clouds' properties. For this, we used a toy model with detailed bin microphysics initialized using three different atmospheric profiles, supporting the formation of shallow to intermediate and deeper warm clouds. The changes in the MSDs affected the clouds' total mass and surface precipitation. In general, the clouds showed higher sensitivity to changes in Ntot than to changes in the MSD's shape, except for the case where the MSD contained giant and ultragiant cloud condensation nuclei (GCCN, UGCCN). For increased Ntot (for the deep and intermediate profiles), most of the MSDs drove an expected non-monotonic trend of mass and precipitation (the shallow clouds showed only the decreasing part of the curves with mass and precipitation monotonically decreasing). The addition of GCCN and UGCCN drastically changed the non-monotonic trend, such that surface rain saturated and the mass monotonically increased with Ntot. GCCN and UGCCN changed the interplay between the microphysical processes by triggering an early initiation of collision–coalescence. The early fallout of drizzle in those cases enhanced the evaporation below the cloud base. Testing the sensitivity of rain yield to GCCN and UGCCN revealed an enhancement of surface rain upon the addition of larger particles to the MSD, up to a certain particle size, when the addition of larger particles resulted in rain suppression. This finding suggests a physical lower bound can be defined for the size ranges of GCCN and UGCCN.

AbstractAn understanding of sea spray aerosol (SSA) production is needed to better assess its influence on climate. Using satellite data, we investigated the production of the coarse mode of aerosol optical depth (AODc), a proxy for SSA, over the pristine South Pacific Gyre. The analysis was done on three time scales: daily, seasonal, and interannual. Scale‐dependent links were shown between the AODc and wind speed (W). AODc and W were positively correlated on both daily and interannual time scales but were significantly anticorrelated on the seasonal time scale. Seasonality of the AODc − W link suggests contribution of other environmental factors. The main variable that could statistically explain trends in AODc on the seasonal time scale was chlorophyll a concentration, which showed a clear negative correlation with AODc. The AODc yield per W unit was clearly reduced when chlorophyll a concentration was high, suggesting a secondary, but important influence of marine biological activity on SSA production.

Abstract The midlatitude atmosphere is characterized by turbulent eddies that act to produce a depth‐independent (barotropic) mean flow. Using the NCEP (National Centers for Environmental Prediction) Reanalysis 2 data, the latitudinal dependence of barotropic kinetic energy and enstrophy are investigated. Most of the barotropization takes place in the extratropics with a maximum value at midlatitudes, due to the latitudinal variations of the static stability, tropopause height, and sphericity of the planet. Barotropic advection transfers the eddy kinetic energy to the zonal mean flow and thus maintains the barotropic component of the eddy‐driven jet. The classic description of geostrophic turbulence exists only at high latitudes, where the quasi‐geostrophic flow is supercritical to baroclinic instability; the eddy‐eddy interactions carry both the barotropization of eddy kinetic energy upscale to the Rhines scale and the barotropization of eddy potential enstrophy downscale.