Research Plan

The project proposes to design and develop a set of advanced machine learning (ML)-based algorithms and models that link ground-based in situ and space-based remote sensing observations of major air quality components with the aim to (a) identify and classify patterns in urban air quality, (b) enable the deduction and forecast of air pollution events related to PM2.5 and ozone from space-based observations, and ultimately (c) identify similarities in air quality regimes between megacities around the globe for improved air pollution mitigation strategies.

These goals will be achieved over a 2-year timeframe:

Year One

• Identify ground and space-based datasets

• Develop a framework to collect and analyze data, look at historical trends and events

• Select a data architecture and models

• Initialize the computational space and migrate data to it

• Create, run, and validate initial machine learning algorithms against training data

Year Two

• Sister cities will be identified and recruited

• Include possible additional datasets

• Validate the models based on emergent research

• Run and retrain the algorithms against control and expanded data

• Initial open source publication

• Regional and international workshops to socialize the models, promote the open source, and gather requirements

The PWWB team is analyzing publicly available data from various sources:

• NASA Worldview

• NASA Atmospheric Science Data Center (ASDC)

• AQMIS - Air Quality and Meteorological Information System

• European Commission - Copernicus Climate Data Storage (CDS)

• NASA Earth Science - MODIS Data

• Bureau of Street Lighting - sourced by South Coast Air Quality Management District (SCAQMD)

• Port of Los Angeles

• OpenAQ