Towards a Flexible Data Fusion Tool for Air Quality Estimation and Forecasting with a Global Scope in Google Earth Engine
Abstract
Researchers and managers of air quality (AQ) around the world have access to a variety of information sources to support their activities. These include in-situ measurements from both regulatory AQ monitoring networks and/or from community-based low-cost sensor networks, remote sensing estimates of various pollutants (e.g., from satellites), and outputs from atmospheric transport and chemistry models. This presentation outlines a system currently under development which will synthesize these diverse data sources into a comprehensive picture of the local AQ situation. The system applies a data fusion approach integrating NASA GEOS-CF composition forecast modeling system outputs, column concentration and aerosol optical depth retrievals from NASA and ESA satellite instruments (VIIRS and TROPOMI, plus TEMPO for the US when available), and surface AQ station measurements, including both regulatory-grade and low-cost sensor technologies. This system will be implemented into a freely available tool on Google Earth Engine to facilitate open science. Key attributes of this tool will be: (1) the ability to perform retrospective analyses and near-term (up to 5 days in advance) forecasts for parameters relevant to AQ (near-surface PM2.5, NO2, and Ozone) at sub-city scale (1-5 km); (2) the flexibility to adapt to varying data availability (e.g., the presence or absence of in-situ AQ monitoring data) and to convey uncertainty based on differing data availability and quality; and (3) the broad applicability of the tool globally and its ease of accessibility by end-users, facilitated by its implementation in Google Earth Engine. Use of the tool will be piloted in collaboration with end-users in Dakar, Senegal (coordinated through UNEP), Rio de Janeiro, Brazil, and multiple US cities (coordinated through US EPA). Ultimately, the tool will be placed under the stewardship of US EPA and UNEP for future applications in the US and globally, with the tool itself remaining freely accessible by all. Here, we present preliminary results related to tool development using offline prototypes for its components, including the forecasting performance of the proposed system and its sensitivity to different data availability (i.e., using model and satellite data only or having access to various numbers of in-situ monitors).
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2022
- Bibcode:
- 2022AGUFM.A53A..08M