AOD is a measure of the integrated columnar aerosols load and is an important parameter for evaluating aerosol–radiation interactions. AR4 described early attempts to retrieve AOD from satellites but did not provide estimates of temporal changes in tropospheric aerosol. Little high-accuracy information on AOD changes exists prior to 1995. Better satellite sensors and ground-based sun-photometer networks, along with improved retrieval methods and methodological intercomparisons, allow assessment of regional AOD trends since about 1995.
AOD sun photometer measurements at two stations in northern Germany, with limited regional representativity, suggest a long-term decline of AOD in Europe since 1986 (Ruckstuhl et al., 2008). Groundbased, cloud-screened solar broadband radiometer measurements provide longer time-records than spectrally selective sun-photometer data, but are less specific for aerosol retrieval. Multi-decadal records over Japan (Kudo et al., 2011) indicate an AOD increase until the mid-1980s, followed by an AOD decrease until the late 1990s and almost constant AOD in the 2000s. Similar broad-band solar radiative flux multi-decadal trends have been observed for urban–industrial regions of Europe and North America (Wild et al., 2005), and were linked to successful measures to reduce sulphate (precursor) emissions since the mid-1980s (Section 2.3). An indirect method to estimate AOD is offered by ground-based visibility observations. These data are more ambiguous to interpret, but records go further back in time than broadband, sun photometer and satellite data. A multi-regional analysis for 1973– 2007 (Wang et al., 2009a) shows that prior to the 1990s visibility-derived AOD was relatively constant in most regions analysed (except for positive trends in southern Asia), but after 1990 positive AOD trends were observed over Asia, and parts of South America, Australia and Africa, and mostly negative AOD trends were found over Europe. In North America, a small stepwise decrease of visibility after 1993 was likely related to methodological changes (Wang et al., 2012f).
AOD can be determined most accurately with sun photometers that measure direct solar intensity in the absence of cloud interferences with an absolute uncertainty of single measurements of ± 0.01% (Holben et al., 1998). AERONET (AErosol RObotic NETwork) is a global sun photometer network (Holben et al., 1998), with densest coverage over Europe and North America. AERONET AOD temporal trends were examined in independent studies (de Meij et al., 2012; Hsu et al., 2012; Yoon et al., 2012), using different data selection and statistical methods. Hsu et al. (2012) investigated AOD trends at 12 AERONET sites with data coverage of at least 10 years between 1997 and 2010. Yoon et al. (2012) investigated AOD and size trends at 14 AERONET sites with data coverage varying between 4 and 12 years between 1997 and 2009. DeMeij et al. (2012) investigated AOD trends between 2000 and 2009 (550 nm; monthly data) at 62 AERONET sites mostly located in USA and Europe. Each of these studies noted an increase in AOD over East Asia and reductions in North America and Europe. The only dense sun photometer network over southern Asia, ARFINET (Aerosol Radiative Forcing over India NETwork), shows an increase in AOD of about 2% yr–1 during the last one to two decades (Krishna Moorthy et al., 2013), with an absolute uncertainty of ± 0.02 at 500 nm (Krishna Moorthy et al., 2007). In contrast, negative AOD trends are identified at more than 80% of examined European and North American AERONET sites (de Meij et al., 2012). Decreasing AOD is also observed near the west coast of northern Africa, where aerosol loads are dominated by Saharan dust outflow. Positive AOD trends are found over the Arabian Peninsula, where aerosol is dominated by dust. Inconsistent AOD trends reported for stations in central Africa result from the use of relatively short time series with respect to the large interannual variability caused by wildfires and dust emissions.
Aerosol products from dedicated satellite sensors complement surface-based AOD with better spatial coverage. The quality of the satellite-derived AOD strongly depends on the retrieval’s ability to remove scenes contaminated by clouds and to accurately account for reflectivity at the Earth’s surface. Due to relatively weak reflectance of incoming sunlight by the sea surface, the typical accuracy of retrieved AOD over oceans (uncertainty of 0.03 +0.05*AOD; Kahn et al. (2007)) is usually better than over continents (uncertainty of 0.05 +0.15*AOD, Levy et al. (2010)).
Satellite-based AOD trends at 550 nm over oceans from conservatively cloud-screened MODIS data (Zhang and Reid, 2010) for 2000–2009 are presented in Figure 2.9. Strongly positive AOD trends were observed over the oceans adjacent to southern and eastern Asia. Positive AOD trends are also observed over most tropical oceans. The negative MODIS AOD trends observed over coastal regions of Europe and near the east coast of the USA are in agreement with sun photometer observations and in situ measurements (Section 220.127.116.11) of aerosol mass in these regions. These regional changes over oceans are consistent with analyses of AVHRR (Advanced Very High Resolution Radiometer) trends for 1981–2005 (Mishchenko et al., 2007; Cermak et al., 2010; Zhao et al., 2011), except over the Southern Ocean (45°S to 60°S), where negative AOD trends of AVHRR retrievals are neither confirmed by MODIS after 2001 (Zhang and Reid, 2010) nor by ATSR-2 (Along Track Scanning Radiometer) for 1995–2001 (Thomas et al., 2010). Satellite-based AOD changes for both land and oceans (Figure 2.9b) were examined with re-processed SeaWiFS (Sea-viewing Wide Field-of-view Sensor) AOD data for 1998–2010 (Hsu et al., 2012). A small positive global average AOD trend is reported, which is likely influenced by interannual natural aerosol emissions variability (e.g., related to ENSO or North Atlantic Oscillation (NAO); Box 2.5), and compensating larger positive and negative regional AOD trends. In addition, temporal changes in aerosol composition are ignored in the retrieval algorithms, giving more uncertain trends than suggested by statistical analysis alone (Mishchenko et al., 2012). Thus, confidence is low for global satellite derived AOD trends over these relatively short time periods.
The sign and magnitude of SeaWiFS regional AOD trends over continents are in agreement with most AOD trends by ground-based sun photometer data (see above) and with MODIS trends (Figure 2.9). The strong positive AOD trend over the Arabian Peninsula occurs mainly during spring (MAM) and summer (JJA), during times of dust transport, and is also visible in MODIS data (Figure 2.9). The positive AOD trend over southern and eastern Asia is strongest during the dry seasons (DJF, MAM), when reduced wet deposition allows anthropogenic aerosol to accumulate in the troposphere. AOD over the Saharan outflow region off western Africa displays the strongest seasonal AOD trend differences, with AOD increases only in spring, but strong AOD decreases during the other seasons. SeaWifs AOD decreases over Europe and the USA and increases over southern and eastern Asia (especially during the dry season) are in agreement with reported temporal trends in anthropogenic emissions, and surface observations (Section 18.104.22.168).
In summary, based on satellite- and surface-based remote sensing it is very likely that AOD has decreased over Europe and the eastern USA since the mid 1990s and increased over eastern and southern Asia since 2000. In the 2000s dust-related AOD has been increasing over the Arabian Peninsula and decreasing over the North Atlantic Ocean. Aerosol trends over other regions are less strong or not significant during this period owing to relative strong interannual variability. Overall, confidence in satellite-based global average AOD trends is low.