PDF Version

How Well Do Satellites Measure Aerosol Optical Depth over Mountainous U.S. Sites Co-located with Sunphotometers?

J.P. Sherman1, I. Krintz1, A.G. Hallar2,3 and W.P. Arnott4

1Appalachian State University, Department of Physics and Astronomy, Boone, NC 28608; 724-664-9077, E-mail: shermanjp@appstate.edu
2University of Utah, Salt Lake City, UT 84112
3Storm Peak Laboratory, Desert Research Institute, Steamboat Springs, CO 80488
4University of Nevada, Reno, NV 89557

Aerosol optical depth (AOD) is a critical parameter for quantifying aerosol radiative effects and for mapping surface air quality from space. Polar orbiting satellite-based platforms can provide near-global coverage on timescales ranging from daily (MODIS) to weekly (MISR). Satellite AOD retrievals are limited by the need for cloud screening and the difficulty of separating the relative contributions from aerosols, trace gases, and surface reflectance to the observed top-of-atmosphere (TOA) radiances. Low aerosol loading, mountainous terrain, and snow or ice-covered surfaces exacerbate these retrieval challenges. Validation of satellite-based AOD retrievals against surface-based “ground truth” measurements is performed by collocating the results within a spatio-temporal window. This window is typically selected based on the estimated spatial and temporal extent over which a given air mass is measured by both the satellite and ground-based sensors. Often, such a window may not be optimized for the unique topography and surface properties at the ground site.

In this work, we evaluate AOD (at 550 nm) measured by NASA’s Moderate Resolution Imaging Spectrometer (MODIS) and NASA’s Multi-angle Imaging Spectrometer (MISR) above four mountainous U.S. sites: (1) Appalachian State University (APP-Boone, NC); (2) Walker Branch, TN (WB); (3) Storm Peak Laboratory (SPL-Steamboat Springs, CO); and (4) University of Nevada (Reno, NV). Satellite AOD products evaluated include MODIS Dark Target (DT) at 3 km and 10 km resolutions, MODIS Deep Blue (DB) 10 km, MODIS combined DT / DB (COMBINED) 10 km, and the MISR 4.4 km product. Multiyear ground-based measurements are made at each site by either a Cimel sunphotometer (as part of NASA AERONET) or Multifilter Rotating Shadowband Radiometer (MFRSR). The relatively low AOD conditions present for most of the sites/seasons provides opportunity to examine the influence of vegetation type and surface topography on satellite AOD performance. The eastern U.S. sites (APP and WB) have comparable numbers of collocations during summer and winter while the number of winter collocations at western U.S. sites (SPL and Reno) is minimal, especially at SPL.

All satellite products have small mean AOD biases (relative to ground-truth measurements) above the heavily vegetated APP and WB sites, with magnitudes less than ~0.03-0.04 over the 10-15 km radius circle centered at each ground site. The small mean biases indicate that the surface models and aerosol models lead to AOD that agrees well with AERONET for these sites. While the mean satellite-sunphotometer AOD differences at / near the western U.S. sites exhibit less spatial variability for DB, those for DT (and to lesser extent, MISR) follow the spatial variability in landcover type and Normalized Differential Vegetation Index (NDVI). Mean MODIS DT-sunphotometer AOD differences are most positive over densely vegetated evergreen forests near SPL (NDVI > 0.60) and over sparsely vegetated desert terrain (NDVI less than ~0.20-0.30) near Reno, with the smallest differences for moderately vegetated terrain (~0.40-0.60) near both sites. The positive bias for Terra DT is ~0.03 higher than for Aqua DT for most landcover types in the regions. The mean MISR-sunphotometer AOD differences are close to zero for NDVI greater than ~0.40 and the differences become increasingly positive over lower NDVI areas, reaching values close to ~0.06 over desert terrain at / near Reno. Evaluations of satellite-based AOD must consider the effects of a collocation radius that includes vegetation types not representative of the ground-truth site.

Figure 1

Figure 1. Maps of mean satellite-sunphotometer AOD difference for satellite AOD product collocations in 50 km radius circle centered at the Reno site.