On the Prediction of Aerosol‐Cloud Interactions Within a Data‐Driven Framework

Aerosol‐cloud interactions (ACI) pose the largest uncertainty for climate projection. Among many challenges of understanding ACI, the question of whether ACI can be deterministically predicted has not been explicitly answered. Here we attempt to answer this question by predicting cloud droplet numbe...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Geophysical research letters Jg. 51; H. 24
Hauptverfasser: Li, Xiang‐Yu, Wang, Hailong, Chakraborty, TC, Sorooshian, Armin, Ziemba, Luke D., Voigt, Christiane, Thornhill, Kenneth Lee, Yuan, Emma
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Washington John Wiley & Sons, Inc 28.12.2024
American Geophysical Union (AGU)
Wiley
Schlagworte:
ISSN:0094-8276, 1944-8007
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Aerosol‐cloud interactions (ACI) pose the largest uncertainty for climate projection. Among many challenges of understanding ACI, the question of whether ACI can be deterministically predicted has not been explicitly answered. Here we attempt to answer this question by predicting cloud droplet number concentration Nc ${N}_{c}$ from aerosol number concentration Na ${N}_{a}$ and ambient conditions using a data‐driven framework. We use aerosol properties, vertical velocity fluctuations, and meteorological states from the ACTIVATE field observations (2020–2022) as predictors to estimate Nc ${N}_{c}$. We show that the campaign‐wide Nc ${N}_{c}$ can be successfully predicted using machine learning models despite the strongly nonlinear and multi‐scale nature of ACI. However, the observation‐trained machine learning model fails to predict Nc ${N}_{c}$ in individual cases while it successfully predicts Nc ${N}_{c}$ of randomly selected data points that cover a broad spatiotemporal scale. This suggests that, within a data‐driven framework, the Nc ${N}_{c}$ prediction is uncertain at fine spatiotemporal scales. Plain Language Summary Ambient aerosol particles act as seeds for ice crystals and cloud droplets that form clouds. Both aerosols and clouds regulate the energy and water budgets of the Earth via radiative and cloud micro/macro‐processes. This is the so‐called aerosol‐cloud interactions (ACI). ACI remains the source of the largest uncertainty for accurate climate projections, due to incomplete understanding of nonlinear multi‐scale processes, limited observations across various cloud regimes, and insufficient computational power to resolve them in models. Quantifying the relation between the cloud droplet Nc $\left({N}_{c}\right)$ and aerosol Na $\left({N}_{a}\right)$ number concentration has been a central challenge of understanding and representing ACI. In this work, we tackle this challenge by predicting Nc ${N}_{c}$ from observations made during the Aerosol Cloud meTeorology Interactions oVer the western ATlantic Experiment (ACTIVATE) using machine learning models. We show that the climatological Nc ${N}_{c}$ can be successfully predicted despite the strongly nonlinear and multi‐scale nature of ACI. However, the observation‐trained machine learning model fails to predict Nc ${N}_{c}$ at fine spatiotemporal scales. Key Points Three‐year in situ measurements (179 flights) provide adequate data to train and validate a random forest model (RFM) to study aerosol‐cloud interactions The RFM can successfully predict cloud droplet number concentration Nc ${N}_{c}$ and identify importance of key predictors Data‐driven Nc ${N}_{c}$ prediction in individual cases shows strong dependency on sampling strategy
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
USDOE
AC05-76RL01830
PNNL-SA-196006
ISSN:0094-8276
1944-8007
DOI:10.1029/2024GL110757