A5058510800- Meteorological Phenomena and Simulations
- Climate variability and models
- Tropical and Extratropical Cyclones Research
- Atmospheric aerosols and clouds
- Oceanographic and Atmospheric Processes
- Energy Load and Power Forecasting
- Atmospheric chemistry and aerosols
- Hydrological Forecasting Using AI
- Aquatic and Environmental Studies
- Solar Radiation and Photovoltaics
- Brain Tumor Detection and Classification
- Computational Physics and Python Applications
- Model Reduction and Neural Networks
- Big Data Technologies and Applications
- Cryospheric studies and observations
- Plant Water Relations and Carbon Dynamics
- Coastal and Marine Dynamics
- Wind and Air Flow Studies
- Air Quality Monitoring and Forecasting
- Solar and Space Plasma Dynamics
- Fluid Dynamics and Turbulent Flows
- Anomaly Detection Techniques and Applications
- Aeolian processes and effects
- Plant Ecology and Soil Science
- Neural Networks and Applications
Environmental Earth Sciences
2022-2024
Columbia University
2021-2024
Courant Institute of Mathematical Sciences
2024
New York University
2024
École Normale Supérieure
2023-2024
Laboratoire de Météorologie Dynamique
2019-2022
Institute of Science and Technology Austria
2022
Université Paris Sciences et Lettres
2019-2021
Centre National de la Recherche Scientifique
2019-2021
École Normale Supérieure - PSL
2019-2021
Idealized simulations of the tropical atmosphere have predicted that clouds can spontaneously clump together in space, despite perfectly homogeneous settings. This phenomenon has been called self-aggregation, and it results a state where moist cloudy region with intense deep convectivestorms is surrounded by extremely dry subsiding air devoid clouds. We review here main findings from theoretical work idealized models this phenomenon, highlighting physical processes believed to play key role...
Accurate prediction of precipitation intensity is crucial for both human and natural systems, especially in a warming climate more prone to extreme precipitation. Yet, models fail accurately predict intensity, particularly extremes. One missing piece information traditional model parameterizations subgrid-scale cloud structure organization, which affects stochasticity at coarse resolution. Here, using global storm-resolving simulations machine learning, we show that, by implicitly learning...
The added value of machine learning for weather and climate applications is measurable through performance metrics, but explaining it remains challenging, particularly large deep models. Inspired by model hierarchies, we propose that a full hierarchy Pareto-optimal models, defined within an appropriately determined error-complexity plane, can guide development help understand the models' value. We demonstrate use Pareto fronts in atmospheric physics three sample applications, with...
Modern climate projections lack adequate spatial and temporal resolution due to computational constraints. A consequence is inaccurate imprecise predictions of critical processes such as storms. Hybrid methods that combine physics with machine learning (ML) have introduced a new generation higher fidelity simulators can sidestep Moore's Law by outsourcing compute-hungry, short, high-resolution simulations ML emulators. However, this hybrid ML-physics simulation approach requires...
This study investigates the feedbacks between an interactive sea surface temperature (SST) and self-aggregation of deep convective clouds, using a cloud-resolving model in nonrotating radiative-convective equilibrium. The ocean is modeled as one layer slab with temporally fixed mean but spatially varying temperature. We find that SST decelerates aggregation deceleration larger shallower slab, consistent earlier studies. anomaly dry regions positive at first, thus opposing diverging shallow...
Abstract We investigate the role of a warm sea surface temperature (SST) anomaly (hot spot typically 3 to 5 K) on aggregation convection using cloud-resolving simulations in nonrotating framework. It is well known that SST gradients can spatially organize convection. Even with uniform SST, spontaneous self-aggregation possible above critical (here 295 K), arising mainly from radiative feedbacks. how circular hot helps convection, and feedbacks modulate this organization. The significantly...
This work explores the effect of convective self-aggregation on extreme rainfall intensities through an analysis at several stages cloud lifecycle. In addition to increases in 3-hourly extremes consistent with previous studies, we find that instantaneous rainrates increase significantly (+30%). We mainly focus and, using a recent framework, relate their increased precipitation efficiency: local relative humidity drives larger accretion efficiency and lower re-evaporation. An in-depth based...
Accurately representing vertical turbulent fluxes in the planetary boundary layer is vital for moisture and energy transport. Nonetheless, parameterization of remains a major source inaccuracy climate models. Recently, machine learning techniques have gained popularity oceanic atmospheric processes, yet their high dimensionality limits interpretability. This study introduces new neural network architecture employing non-linear reduction to predict dry convective layer. Our method utilizes...
<p>Idealized simulations of the tropical atmosphere have predicted that clouds can spontaneously clump together in space, despite perfectly homogeneous settings. This phenomenon has been called self-aggregation, and results a state where moist cloudy region with intense deep convective storms is surrounded by extremely dry subsiding air devoid clouds. We review here main findings from theoretical work idealized models, highlighting physical processes believed to play key role...
Accurate prediction of precipitation intensity is crucial importance for both human and natural systems, especially in a warming climate more prone to extreme precipitation. Yet models fail accurately predict intensity, particularly extremes. One missing piece information traditional model parameterizations sub-grid scale cloud structure organization, which affects stochasticity at the grid scale. Here we show, using storm-resolving simulations machine learning, that by implicitly learning...
While the added value of machine learning (ML) for weather and climate applications is measurable, explaining it remains challenging, especially large deep models. Inspired by model hierarchies, we propose that a full hierarchy Pareto-optimal models, defined within an appropriately determined error-complexity plane, can guide development help understand models' value. We demonstrate use Pareto fronts in atmospheric physics through three sample applications, with hierarchies ranging from...
As heatwaves and droughts are becoming more frequent intense, such as in Western Europe, there is a growing interest unraveling the physical mechanisms behind their occurrences changes. Soil desiccation critical for self-intensification self-propagation of heatwaves, but its relative importance compared to other well-known large-scale atmospheric mechanisms, persistent blocking systems horizontal warm advection, remains elusive, particularly context changing climate. Here we utilize machine...
The radiation parameterization is one of the computationally most expensive components Earth system models (ESMs). To reduce computational cost, often calculated on coarser spatial or temporal scales, both, than other physical processes in ESMs, leading to uncertainties cloud-radiation interactions and thereby radiative temperature tendencies. One way around this issue emulation using machine learning which usually faster has good accuracy a high dimensional parameter space. This study...
<p>The spontaneous aggregation of convective clouds over a moist portion the domain is ubiquitous in cloud resolving model simulations. This phenomenon significantly reduces mean total water vapor and enhances outgoing long radiation. In this study we use system atmospheric modeling (SAM) radiative-convective equilibrium (RCE) setup order to investigate impact an interactive sea surface temperature (SST) on progress. We slab ocean (with depth 5, 10 50 m) with constant target...
<p>Convective organization has been associated with extreme precipitation in the tropics. Here we investigate impact of convective self-aggregation on rainfall rates. We find that significantly increases extremes, for 3-hourly accumulations(+70%) consistent earlier studies, but also instantaneous rates (+30%). show this latter enhanced is mainly due to local increase relative humidity which drives larger accretion and lower re-evaporation thus a higher...
<div> </div><div>The representation of turbulent mixing in climate models is challenging due to eddies' wide range scales and modeling limitations thus <div>turbulent flux parameterization remains one major source inaccuracy. Most parameterizations decompose total a diffusion term, which the local small scale non diffusive term  (called non-local) that represents contribution from coherent structure. However,...
<p>This study investigates the impact of diurnal cycle incoming solar radiation on spontaneous organization convective clouds, hereafter self-aggregation. We run 3D cloud-resolving simulations in RCE framework with interactive sea surface temperature (SST). SST is allowed to interact atmosphere using a slab ocean ( H = 1 - 200 meters) fixed mean but locally varying temperature. The self-aggregation deep clouds starts appearance dry patches that grow size while getting drier,...
<p>Convective organisation has been associated with extreme precipitation in the tropics. Here we investigate impact of convective self-aggregation on rainfall rates. We find that significantly increases extremes, for 3-hourly accumulations but also instantaneous rates (+ 30 %). show this latter enhanced is mainly due to local increase relative humidity which drives larger accretion efficiency and lower re-evaporation thus a higher...