Assessing satellite AOD based and WRF/CMAQ output PM2.5 estimators

摘要

Fine particulate matter measurements (PM_2.5) are essential for air quality monitoring and related public health; however, the shortage of reliable measurmennts constrains researchers to use other means for obtaining reliable estimates over large scales. In particular, model forecasters and satellite community use their respective products to develop ground particulate matter estimations but few experiments have explored how the remote sensing approaches compare to the high resolution models. . In this paper we focus on studying the performance of the Moderate Resolution Imaging Spectroradiometer (MODIS) and the Geostationary Operational Environmental Satellites (GOES) regression based estimates in comparison to more direct bias corrected outputs from the Community Multiscale Air Quality (CMAQ) model, We use a two-year dataset (2005-2006) and apply urban, season and hour filters to illustrate the agreement between estimated and in-situ measured fine particulate matter from the New York State Department of Environmental Conservation (NYSDEC). We first begin by analyzing the correspondence between ground aerosol optical depth (AOD) measurements from an AERONET (AErosol RObotic NETwork) Cimel sun/sky radiometer with both satellite and model products in one urban location; we show that satellite readings perform better than model outputs, especially during the summer (R_MODIS&gt;=0.65, R_CMAQ&gt;=0.37). This is a clear symptom of the difficulty in the models to properly model realistic optical properties. We then turn to a direct assessment of PM2.5 presenting individual comparisons between ground PM_2.5 measurements with satellite/model predictions and demonstrate the higher accuracy from model estimations (<i>R^urban_MODIS</i> &ge; 0.74, <i>R^urban_CMAQ</i> &ge; 0.77; <i>R^non-urban_MODIS</i> &ge; 0.48, <i>R^non-urban_CMAQ</i> &ge; 0.78). In general, we find that the bias corrected CMAQ estimates are superior to satellite based estimators except at very high resolution. Finally, we show that when using both model and satellite approximations as separate estimators merged optimally, our product (PM_2.5 average) becomes closer to real measurements with improved correlations (R_AVE ~ 0.86) in urban areas during the summer.