Six types of dust events in Eastern Mediterranean identified using unsupervised machine-learning classification

Dori Nissenbaum, Ron Sarafian, Yinon Rudich, Shira Raveh-Rubin

Research output: Contribution to journalArticlepeer-review

4 Scopus citations

Abstract

Dust events can be hazardous to society and human health and can cost hundreds of millions of dollars each. Inhalable suspended particulate matter with a diameter of under 10 μm (PM10) is of particular concern since it can cause an array of adverse health effects following short- and long-term exposure. Understanding recurring episodes of dust events will enable accurate and timely forecasts, helping mitigate their impacts and allow for better societal protection. Previous expert-based studies highlighted several dust sources and transport pathways into the Eastern Mediterranean region and further identified several weather systems that sustain these transport routes. But since supervised methods, i.e., manual classifications, may have disadvantages, it is often preferred to use unsupervised methods, or systematic classifications. Here, a novel climatological understanding of the link between weather systems and dust transport is achieved by systematically classifying dust events. Using ground PM10 measurements in Israel to objectively identify a climatological set of extreme dust events between 2003 and 2020, we combine atmospheric data from ERA5 and CAMS reanalysis data sets and apply a new, unsupervised, and unbiased method to classify dust events in the Eastern Mediterranean. Six coherent types emerge, corresponding to events governed by shallow and deep Mediterranean cyclones, Mediterranean dipoles, Sharav thermal lows, Arabian anticyclones, and local factors, respectively. Having different seasonality, these classes are insightful in mapping the meteorological conditions and weather systems governing dust emission and transport towards the Eastern Mediterranean. In this context, slantwise-descending dry intrusions are shown to be a key precursor dynamical feature common to the buildup of elevated dust concentrations in three of the clusters of the highest PM10 concentrations.

Original languageEnglish
Article number119902
JournalAtmospheric Environment
Volume309
DOIs
StatePublished - 15 Sep 2023
Externally publishedYes

Keywords

  • Air quality
  • Atmospheric transport
  • Data science
  • Dry intrusions
  • Extreme weather
  • Machine learning

ASJC Scopus subject areas

  • General Environmental Science
  • Atmospheric Science

Fingerprint

Dive into the research topics of 'Six types of dust events in Eastern Mediterranean identified using unsupervised machine-learning classification'. Together they form a unique fingerprint.

Cite this