Analysis of the cloud fraction adjustment to aerosols and its dependence on meteorological controls using explainable machine learning

Atmospheric chemistry and physics, Journal Year: 2024, Volume and Issue: 24(22), P. 13025 - 13045

Published: Nov. 26, 2024

Abstract. Aerosol–cloud interactions (ACI) have a pronounced influence on the Earth's radiation budget but continue to pose one of most substantial uncertainties in climate system. Marine boundary-layer clouds (MBLCs) are particularly important since they cover large portion surface. One biggest challenges quantifying ACI from observations lies isolating adjustments cloud fraction (CLF) aerosol perturbations covariability and local meteorological conditions. In this study, isolation is attempted using 9 years (2011–2019) near-global daily satellite products combination with reanalysis data parameters. With cloud-droplet number concentration (Nd) as proxy for aerosol, MBLC CLF predicted by region-specific gradient boosting machine learning (ML) models. By means SHapley Additive exPlanation (SHAP) regression values, sensitivity Nd factors well influences Nd–CLF quantified. The regional ML models able capture, average, 45 % variability. Based our statistical approach, global patterns suggest that positively associated Nd, stratocumulus-to-cumulus transition regions Southern Hemispheric midlatitudes. However, retrieval bias may contribute non-causality these positive sensitivities, hence should be considered upper-bound estimates. estimated inversion strength (EIS) ubiquitously strongest tropical subtropical topped stratocumulus within Globally, increased sea-surface temperature (SST) reduces CLF, regions. spatial horizontal wind components free troposphere point impact synoptic-scale weather systems vertical shear MBLCs. relationship found depend more selected thermodynamical variables than dynamical particular EIS SST. midlatitudes, stronger amplify relationship, while not observed regions, amplified higher SSTs, potentially pointing frequently delaying expected climatic changes SST thus future forcings adjustment. novel data-driven framework, whose limitations also discussed, produces quantification response aerosols, taking into account covariations meteorology.

Language: Английский

The Exact Value of the Omitted Variable Bias in Linear Regression

Published: Jan. 1, 2024

Language: Английский

Analyzing Scaling Laws Beyond City Boundaries Reveals Substantial Environmental Benefits of Dense Settlements

Published: Jan. 1, 2024

Language: Английский