An official website of the United States government.

We've made some changes to EPA.gov. If the information you are looking for is not here, you may be able to find it on the EPA Web Archive or the January 19, 2017 Web Snapshot.

Downscaler Model for predicting daily air pollution

What is the Downscaler Model?

The Downscaler Model (DS) fuses output from a gridded atmospheric model known as the Community Multi-Scale Air Quality Model (CMAQ) with point air pollution measurements. The two sources of information are valuable in different ways. The monitoring data are sparsely collected with some missing data, but provide direct measurement of the true pollutant value up to measurement error. CMAQ estimates gridded averages with no missing values, but is subject to calibration error. However, CMAQ calibration bias can be accounted for through fusion modeling with accurate monitoring data. The DS model used here is a spatially-varying weighted model that regresses monitoring data on a derived regressor obtained by smoothing the entire CMAQ output with weights that vary both spatially and temporally. This adaptive smoothing of CMAQ was used to achieve stronger association with the monitoring data by taking advantage of useful spatial information in neighboring CMAQ cells surrounding the cell where the monitoring data occurs.

What are the benefits of using the Downscaler Model?

The DS model combines air quality monitoring and modeling data to provide better fine-scale predictions of air pollutant levels at local and community scales. The DS model used here provides the best out-of-sample validation predictions relative to earlier DS models and other traditional geostatistical methods. It allows users to zoom in to specific locations to obtain information about daily pollutant patterns.

Who should use this predictive database?

This model provides spatial air pollution information for statisticians, environmental scientists, state and local agencies, and anyone interested in establishing and quantifying air pollution levels at local and national scales.

What Predictive Surfaces are available?

Daily spatial surfaces of O3 (ppb) and PM2.5 (ug/m3) are archived in the link below. For 2002-2012, daily predictions are provided at the 2010 U.S. Census Tract centroid locations. DS references and other modeling details can be found in the Air MetaData file on the Remote Sensing Information Gateway (RSIG) website below.

Related Link:

Technical Contact: