Abstract

By mass, dust is the largest contributor to global aerosol burden, yet long-term observational records of dust, particularly over the ocean, are limited. Here, two nearly global observational datasets of dust aerosol optical depth τd are created primarily on the basis of optical measurements of the aerosol column from 1) the Moderate Resolution Imaging Spectroradiometer (MODIS) aboard the Terra satellite spanning from 2001 to 2018 and 2) the Advanced Very High Resolution Radiometer (AVHRR) from 1981 to 2018. The quality of the new data is assessed by comparison with existing dust datasets that are spatially more limited. Between 2001 and 2018, τd decreased over Asia and increased significantly over the Sahara, Middle East, and parts of eastern Europe, with the largest increase found over the Aral Sea where emissive playa surfaces have been exposed. These daily, observational, and nearly global records of dust will allow for improvement in understanding the role of dust in climate variability.

1. Introduction

Aeolian dust makes up the largest mass fraction of the global aerosol burden (Textor et al. 2006) and produces profound impacts on natural (Rosenfeld et al. 2001; Swap et al. 1992; Mahowald 2011; Miller et al. 2004; Ault et al. 2011; Evan et al. 2006; Strong et al. 2018) and anthropological systems (Tong et al. 2017; Griffin 2007; Shao 2008; Mani and Pillai 2010; Ai and Polenske 2008; Brown 2002). A major limiting factor in understanding the global distribution of dust, how it will change in the future, and its complex interactions with Earth’s climate, is the limited availability of measurements especially over the ocean (Prospero and Mayol-Bracero 2013). Airborne campaigns have provided measurements for select places and times (Ralph et al. 2016; Formenti et al. 2003; Chen et al. 2011; Formenti et al. 2008; Stith et al. 2009; Ryder et al. 2013; Klaver et al. 2011) and ground-based networks have provided long-term measurements for specific regions (Malm et al. 1994; Prospero and Nees 1986; Prospero 1999; Holben et al. 1998). Most satellite sensors do not isolate the dust contribution to aerosol optical depth (AOD) at the point of retrieval, so several methods have been proposed and utilized for studying spatial and temporal variations of dust using satellite measurements and models. These methods have mostly used optical properties of the aerosol column related to particle size (Ginoux et al. 2012; Kaufman 2005), color, and single scattering albedo (Ginoux et al. 2012), or have used the ultraviolet (UV) absorptive property of dust for identification (Ginoux 2003). Several studies have used satellite-based products to study dust aerosol optical depth τd over either the ocean or land in select locations. Over the Atlantic Ocean, Kaufman (2005) analyzed τd using the Moderate Imaging Spectroradiometer (MODIS) retrievals of AOD and Evan and Mukhopadhyay (2010) used a similar method to create a record of τd based on retrievals by the Advanced Very High Resolution Radiometer (AVHRR). In addition, Ginoux et al. (2012) used τd over bright land surfaces that is based on MODIS Deep Blue AOD to identify dust source regions. Monthly τd has also been used in evaluation of the dust emission schemes of global climate models (Pu and Ginoux 2018b).

More recently, studies have used a combination of satellite observations and models to investigate the distribution of dust globally and trends in dust for select regions. Ridley et al. (2016) produced a long-term mean dataset of τd between 2004 and 2008 using AOD retrievals from multiple satellite platforms and dust estimates from several global models and Chin et al. (2014) used satellite AOD from several sensors alongside the Goddard Chemistry Aerosol Radiation and Transport (GOCART) model to investigate multidecadal regional trends in aerosol species, including dust. These merged approaches have value in allowing for elucidation of aerosol sources and the ability to distinguish anthropogenic from natural dust. However, they are susceptible to model biases in dust emission and transport. Climate models display a large diversity in dust aerosol optical depth and dust emission, mostly underestimating dust emission from major sources, and do not consistently agree with observations (Evan et al. 2014; Huneeus et al. 2011; Pu and Ginoux 2018a; Kok et al. 2017; Kok 2010; Ryder et al. 2018). Therefore, it is valuable to have complimentary extensive observational records of dust for comparison.

Here, we expand on the work of Evan and Mukhopadhyay (2010) by using the extended record of satellite retrievals to produce two nearly global τd datasets developed using MODIS (2001–18) and AVHRR (1981–2017) at daily and monthly resolution, respectively, with the former dataset extended to cover both land and ocean regions. Although dust transport happens on a temporal scale of days to weeks, the previously mentioned efforts have focused on estimating monthly mean τd. This work provides the longest record of observed daily τd to date, at a temporal resolution suitable for analysis of daily transport events (daily) and 1° × 1° spatial resolution.

In the next section we describe the data and methods used in the estimation of global τd and provide an estimation of the associated uncertainty. Our results in section 3 include identification of climatological seasonal patterns in τd, a comparison of our τd estimates as derived from two different sensors, and a comparison of the MODIS-derived dataset with preexisting ground-based measurements of dust aerosol from several locations. We then identify regional trends in τd for the period of each record.

2. Method

Two datasets of τd were created. The first, which hereinafter will be referred to as τdM, is based primarily on AOD (τ) retrievals from the MODIS instrument on the Terra platform, with a companion set of estimates created for the same instrument aboard the Aqua platform, (both accessed at https://ladsweb.modaps.eosdis.nasa.gov) and extends from 2001 to 2018 for the case of Terra and 2003 to 2018 for Aqua. The τdM includes estimates over both land and water-covered surfaces. The τdA is based on aerosol optical thickness (AOT) retrievals from the AVHRR instruments aboard the NOAA series of satellites (Heidinger et al. 2014; Zhao et al. 2008, 2002; Zhao and NOAA CDR Program 2017) and covers the period 1981–2018. AVHRR AOT retrievals are not made over land, and thus τdA is an overwater-only product.

a. Datasets used

MODIS level-3 (L3), Collection 6.1 (C061), daily Dark Target AOD is derived from the measured 500-m resolution reflectance from all visible MODIS bands by taking the average of the measured radiance over all scenes that are cloud free and glint free over the ocean within a 10-km grid box and fitting these values to a lookup table. Level-3 MODIS Dark Target AOD is calculated using the quality assurance confidence screened level-2 pixel-level data. MODIS 550-nm fine-mode fraction f is calculated from the ratio of the small-mode AOD to the total AOD. MODIS Collection 6 (C006) AOD over ocean has been shown to have an expected error of +(0.04 + 10%)/−(0.02 + 10%) (Levy et al. 2013). For dust-dominated regions, it has been shown that there is a bias of +5% (Kaufman 2005). There has been no systematic program for evaluation of MODIS f because its definition is somewhat ambiguous, making evaluations against other estimates difficult (Kleidman et al. 2005). At this time, there is no envelope of error for C061 f. However, Collection 5 (C005) f over ocean was found to agree with Aerosol Robotic Network (AERONET) retrieved sky radiance fine-mode fraction within approximately ±0.20 (Kleidman et al. 2005), and Bréon et al. (2011) found agreement between MODIS (τf,MODIS) and AERONET (τf,AERONET) fine-mode AOD, as evidenced by a correlation r value of 0.76, root-mean-square error (RMSE) of 0.08, bias of 0.01, and 53% of retrievals within the C005 envelope of error, δτ = 0.03 + 0.05τAERONET, which describes a confidence envelope containing 1 standard deviation σ (i.e.,68%) of τf,MODIS/τf,AERONET matchups. In our estimate of the known uncertainty in τd, we assume that the C061 f uncertainty has not changed from the C005 values. Although a quality-assured dataset does exist for τf,MODIS, we chose to use the standard dataset so as to retain additional coverage.

The equivalent set of estimates for τdM from the Aqua platform extends from 2003 to 2018. The Aqua platform passes the equator at 1330 local time, whereas Terra passes near 1030 local time. AOD from MODIS Terra and Aqua have been shown to agree very well (Ichoku 2005). However, globally, Terra monthly mean AOD over ocean has been shown to be higher than Aqua by +0.015 (Levy et al. 2018; Remer et al. 2008). This is partially explained by diurnal cycles in cloud fraction (King et al. 2013), which may create differences in sampling between these sensors, and partially explained by issues with the total infrared radiation measured by Terra (Levy et al. 2018). This paper will focus largely on τdM from the Terra platform, because it has a longer record. However, we will discuss decadal trends in τdM derived from Aqua in section 4a.

Because our τd estimate is built upon AOD, it carries with it the known limitations of the sensors from which it is derived. MODIS AOD retrievals are limited to cloud-free pixels, and errors in cloud screening will impact the AOD estimate. The MODIS aerosol cloud mask uses the standard deviation of reflectance in sets of pixels to remove cloudy pixels, which increase spatial variability. Absolute reflectance at 1380 nm and the ratio of reflectances at 1380 and 1240 nm as well as several infrared tests are used for additional screening of thin cirrus. Over ocean, the brightest and darkest 25% of the remaining pixels are arbitrarily removed and the average reflectance in each channel is calculated from the remaining pixels. Over land, the brightest 50% and darkest 30% of pixels are discarded before the averaging step (Remer et al. 2012). While this usually produces a reasonable result, in the case that unscreened clouds remain before the discarding of the brightest and darkest retrievals, this can produce AOD estimates that are biased high (Kahn et al. 2007). It is known that MODIS AOD is biased very positively (high) in the Southern Hemisphere mid–high latitudes over the Southern Ocean because of extensive broken stratocumulus and cirrus cloud contamination (Toth et al. 2013). We have thus limited these datasets to extend from 60°N to 50°S.

Level-2, version 3, AOD and fine-mode fraction from select sites, as will be discussed in section 2b, retrieved from AERONET (Holben et al. 1998, 2001) sunphotometers using the Spectral Deconvolution Algorithm (SDA), version 4.1 (O’Neill 2003; https://aeronet.gsfc.nasa.gov) are used to estimate τdM and τdA over the ocean. In the determination of AERONET fine-mode fraction, two spectral modes of AOD are defined on the basis of the premise that the coarse-mode spectral variation is approximately neutral while the derivative of the fine-mode spectral variation is an approximate function of the fine-mode Ångstrom exponent (O’Neill 2003). From this, the ratio of fine-mode to total AOD is determined at a reference wavelength of 500 nm from a second-order polynomial fit of the natural logarithm of AOD and wavelength applied to each AOD spectrum across six bands. This definition of fine-mode fraction is different from that used in the MODIS dataset and may result in some disagreement. MODIS f has been shown to slightly overestimate the fine-mode fraction in dust or salt dominated regions by 0.1–0.2 relative to AERONET fine-mode fraction as generated from the O’Neill algorithm (Kleidman et al. 2005), which has the potential to introduce a small positive bias in our estimated τd.

MODIS C061 L3 Deep Blue products [AOD, single scattering albedo (SSA), and Ångstrom exponent (AE); https://ladsweb.modaps.eosdis.nasa.gov] were used in the estimation of τdM over land (Hsu et al. 2013). These products use the blue channels of MODIS where surface reflectance is very low. Pixel-level retrievals are averaged over 10 km × 10 km grids, and data are then aggregated into granules. The L3 products are provided at 1° × 1° resolution. SSA at 412 nm and at 660 nm were used in addition to AE, which is inversely proportional to particle size (Angstrom 1929). Deep Blue AOD was used as the basis for estimation of τdM over land and is reported to have an expected error of +(0.04 + 10%)/−(0.02 + 10%) (Hsu et al. 2014). The C061 Deep Blue AOD release extended coverage to all nonsnow land surfaces, whereas C005 only included bright land surfaces.

The 630-nm AVHRR AOT climate data record (CDR) (Zhao and NOAA CDR Program 2017; accessed at https://www.ncei.noaa.gov/data/) was used in the estimation of τdA (Zhao and NOAA CDR Program 2017; Chan et al. 2013). The CDR was created using the NOAA AVHRR PATMOS-x level-2B product (Zhao et al. 2008, 2002; Zhao and NOAA CDR Program 2017). Although there is no difference between AOD and AOT, we refer to the MODIS data as AOD and the AVHRR data as AOT because this is how each is referenced in their documentation. It combines retrievals from 17 different sensors, and the overpass time and number of observations per day (4–8) varied between sensors. Large data gaps occur for years before 1985. AVHRR AOT was regridded from its native resolution of 0.1° × 0.1° to a resolution of 1° × 1° using inverse distance weighted interpolation. Aerosol retrievals are only available over ocean and require clear-sky conditions (Heidinger et al. 2014). The CDR AVHRR AOT dataset has a systematic error (positive bias) of 0.03 ± 0.006 and a random error of ±0.113 (Zhao and NOAA CDR Program 2017). The error budget of AVHRR AOT CDR is based on AERONET validation. Another product of AVHRR aerosol loading is available, derived using algorithms from modern sensors (Hsu et al. 2017; Sayer et al. 2017), but it is limited to several years. Therefore, we did not use this product.

Daily mean surface wind speed from the second Modern-Era Retrospective Analysis for Research and Applications (MERRA-2) (Gelaro et al. 2017) 1-h time-averaged surface flux assimilation product (accessed at https://disc.gsfc.nasa.gov/) was used in the estimation of marine AOD. This was regridded from its native resolution of 0.625° longitude × 0.5° latitude to a resolution of 1° × 1° using bilinear interpolation. MERRA-2 surface wind patterns have been shown to be similar to other datasets, but greater in magnitude in most regions than MERRA and ERA-Interim and weaker in magnitude than NCEP-R2 (Bosilovich et al. 2015). Similar to most reanalysis products, MERRA-2 winds are weaker than 93% of observations considered in Bosilovich et al. (2015).

b. MODIS dust optical depth over ocean

Our method for estimating τdM over ocean surfaces is based on Kaufman (2005). Underlying this method is the assumption that the major contributions to total AOD are anthropogenic and biomass-burning aerosol, marine aerosol, and dust aerosol, and thus one can deduce τd by estimating the contributions from the other aerosol types. As anthropogenic and biomass-burning aerosols are generally dominated by submicrometer particles with AE greater than 1 (Dubovik et al. 2002; Schmeisser et al. 2017), f can be used to discriminate these types from the contributions by coarse-mode-dominant aerosol types.

The total aerosol optical depth τ is

 
τ=τd+τm+τa,
(1)

where τm and τa are the marine and anthropogenic or biomass-burning contributions to τ, respectively. Here we have combined the biomass-burning and anthropogenic aerosols for convenience, because they both have a dominant fine mode and will thus be isolated and removed from τ simultaneously. The fine-mode aerosol optical depth is given by and can be expressed as a function of the fine-mode optical depth contributions from the individual aerosol species,

 
fτ=fdτd+fmτm+faτa,
(2)

where f, fd, fm, and fa are the total, dust, marine, and anthropogenic or biomass-burning contributions to f, respectively. Equations (1) and (2) can be combined and rearranged to yield τd,

 
τd=τ(faf)τm(fafm)fafd.
(3)

Data sources for each of these variables can be found in Table 1. Characteristic values of fm, fa, and fd were determined using the average of fine-mode fraction from AERONET stations dominated by each aerosol type, weighted by the total τ (Table 2). This weighting was performed for each station according to

 
f=i=1Nfi*AODi/i=1NAODi,
(4)

where f is fa, fd, or fm. For fa and fd, AERONET stations were chosen on the basis of Cazorla et al. (2013). For fm, marine aerosol-dominated AERONET island stations were selected based on Lehahn et al. (2010) and Chaubey et al. (2011) and through visual selection of additional remote island sites. MODIS retrievals were not used for determining these coefficients since f is only distinguishable over the ocean with this instrument. This makes it likely that all MODIS f estimates contain some significant contribution from sea spray and as such represents a deviation from the methods of Kaufman (2005).

Table 1.

Summary of variables and data types used in Eqs. (1)(3). The fd, fm, and fa represent characteristic AERONET fine-mode fractions for dust, marine, and anthropogenic or biomass-burning-dominated AERONET stations, respectively; τ and f represent the total AOD and fine-mode fraction; and τm represents the marine aerosol contribution to the total AOD.

Summary of variables and data types used in Eqs. (1)–(3). The fd, fm, and fa represent characteristic AERONET fine-mode fractions for dust, marine, and anthropogenic or biomass-burning-dominated AERONET stations, respectively; τ and f represent the total AOD and fine-mode fraction; and τm represents the marine aerosol contribution to the total AOD.
Summary of variables and data types used in Eqs. (1)–(3). The fd, fm, and fa represent characteristic AERONET fine-mode fractions for dust, marine, and anthropogenic or biomass-burning-dominated AERONET stations, respectively; τ and f represent the total AOD and fine-mode fraction; and τm represents the marine aerosol contribution to the total AOD.
Table 2.

AERONET stations used for calculation of fa, fd, and fm, with the years of data used from each station in this calculation.

AERONET stations used for calculation of fa, fd, and fm, with the years of data used from each station in this calculation.
AERONET stations used for calculation of fa, fd, and fm, with the years of data used from each station in this calculation.

Marine aerosol contribution

Kaufman (2005) parameterized marine aerosol based on NCEP surface (1000 hPa) wind speed at 2.5° × 2.5° resolution. This parameterization was based on the relationship between surface wind speed from a National Climatic Data Center meteorological station and an AERONET station on Midway Island from 14 months of data, excluding the dust season (February–May) (Smirnov et al. 2003).

Here we consider that the total marine aerosol optical depth is given by

 
τm=τm,c+τm,f,
(5)

where τm,c and τm,f are the coarse- and fine-mode contributions to marine aerosol. We make the assumption that τm,f is weakly dependent on wind speed (Satheesh et al. 2006), and that the dependence of τm on surface wind speed is primarily from τm,c (Lehahn et al. 2010). We identified a region dominated by marine aerosols (0°–25°S, 178°–130°W), and set τm,f equal to the long-term MODIS fine-mode aerosol optical depth,

 
τm,f=0.05(±0.03),
(6)

where ±0.03 is the standard deviation of the daily MODIS fine-mode aerosol optical depth at 0°–25°S, 178°–130°W from 2001 to 2017 (Fig. 1c). Assuming the linear relationship,

 
τm,c=αw,
(7)

where w denotes surface wind speed and it is implied that τm,c = 0 in the absence of surface winds, the proportionality term α was calculated via regression of MODIS Terra coarse-mode AOD onto MERRA-2 surface wind speeds, both interpolated to a 1° × 1° horizontal resolution, and over the same region used to calculate τm,f. Via the linear regression (Fig. 1b),

 
τm,c=0.008(±0.0001)w,
(8)

which is consistent with Lehahn et al. (2010), who found τm,c to be linearly related to surface wind speed with a slope of 0.009 ± 0.002 m s−1. We tested three other parameterizations for marine aerosol and found that this parameterization resulted in the lowest bias (+0.02) and lowest RMSE (0.04) when compared with AERONET AOD at Nauru, a remote island station, from 2001 to 2014 for days on which there were clear-sky measurements from both instruments. These other parameterizations are described in the online supplemental material.

Fig. 1.

Joint probability distribution of MODIS Terra (a) AOD, (b) coarse-mode AOD, and (c) fine-mode AOD binned by daily mean MERRA-2 wind speed in the southern equatorial Pacific (0°–25°S, 178°–130°W) from 2001 to 2017. The color contours indicate the percent of total pixels falling within each wind speed and AOD bin. The linear parameterization used to estimate coarse-mode AOD from wind speed (τm,c = 0.008) is shown by the white dashed line in (b). The constant fine-mode AOD parameterization (τm,f = 0.05) is shown by the white dashed line in (c).

Fig. 1.

Joint probability distribution of MODIS Terra (a) AOD, (b) coarse-mode AOD, and (c) fine-mode AOD binned by daily mean MERRA-2 wind speed in the southern equatorial Pacific (0°–25°S, 178°–130°W) from 2001 to 2017. The color contours indicate the percent of total pixels falling within each wind speed and AOD bin. The linear parameterization used to estimate coarse-mode AOD from wind speed (τm,c = 0.008) is shown by the white dashed line in (b). The constant fine-mode AOD parameterization (τm,f = 0.05) is shown by the white dashed line in (c).

Sea salt aerosol has been parameterized previously using linear (Lehahn et al. 2010; Smirnov et al. 2003; Mulcahy et al. 2008; Glantz et al. 2009; Huang et al. 2010), power-law (Mulcahy et al. 2008; Glantz et al. 2009), and exponential (Moorthy and Satheesh 2000) relationships to 10-m or surface wind speeds from in situ and remote sensing measurements. It is well known that, to first order, sea salt emission is primarily dependent on wind speed (Jaeglé et al. 2011). However, other factors including sea surface temperature (SST), organic material, vertical mixing, advection, wet and dry deposition, and relative humidity could impact sea salt emission to a lesser extent. The second most commonly used predictor, after wind speed, is SST (Jaeglé et al. 2011; Gong 2003). However, laboratory and in situ investigations of the SST dependence of marine aerosol have had contradictory results, indicating that we do not yet fully understand how temperature impacts SSA (Grythe et al. 2014; Mårtensson et al. 2003; Zábori et al. 2012). Therefore, we have chosen to retain a parameterization that is only based upon surface winds.

c. τdM over land

The τdM over land for the time period between 2001 and 2018 was estimated on the basis of the methods of Ginoux et al. (2012) with minor modifications. Ginoux et al. (2012) used C005 MODIS Deep Blue AOD to identify pixels over land that contained dust by their column optical characteristics including AOD of greater than 0.1, AE of less than 0, SSA at 412 nm of less than 0.95, and SSA at 412 nm that is less than or equal to SSA at 660 nm. The AE criterion selects for pixels with a dominant coarse mode to the aerosol size distribution (Ginoux et al. 2012), which should primarily be dust and marine aerosol (Dubovik et al. 2002). The criterion for SSA at 412 nm is related to absorption of solar radiation in the green channel, characteristic of dust, and is effective at removing influences of marine aerosol. The difference in SSA between 660 and 412 nm is used because an optical property of dust is a positive spectra variation of SSA with wavelength, which manifests as a sharp decrease in absorption from the red (660 nm) to blue (412 nm) channels. This criterion is effective at removing any additional influence of biomass burning, because biomass-burning particles have spectral variation with the opposite slope. If a retrieval met all of these criteria, its Deep Blue AOD contributed to the long-term mean of τd at that grid cell. To test the thresholds used in Ginoux et al. (2012), we collected these same data within three overland regions where the aerosol signal is dominated by dust, and one overland region where mineral aerosols are unlikely to be present, and binned the pixels of this data for each location by their optical properties using the time period 2006 to 2010. The boundaries of these regions were chosen visually, selecting for either desert surface boundaries or for vegetation cover (Fig. S8 in the online supplemental material). The dust-dominated locations included a region over the Sahara Desert (10°–30°E, 19°–30°N) (Fig. 2a), a region over the Arabian Desert (40°–50°E, 23°–35°N) (Fig. 2b), and a region over the Taklamakan Desert (77°–88°E, 36°–40°N) (Fig. 2c). The non-dust-dominated location chosen was over equatorial Africa (10°–32°E, 10°S–5°N) (Fig. 2d).

Fig. 2.

Histograms of (top) Ångstrom exponent, (middle) single scattering albedo, and (bottom) ratio of SSA at 412 nm divided by SSA at 660 nm for three characteristically dusty regions [(a) the Sahara Desert (19°–30°N, 10°–30°E), (b) the Arabian Desert (23°–35°N, 40°–50°E), and (c) the Taklamakan Desert (36°–40°N, 77°–88°E)] and one anthropogenic aerosol–dominated region [(d) equatorial Africa (10°S–5°N, 10°–32°E)] for the period from 2006 to 2010 for daily C061 L3 MODIS Deep Blue products. Colored brown [in (a)–(c)] or green [in (d)] bars indicate bins that match the criterion used to identify dust for the respective optical property.

Fig. 2.

Histograms of (top) Ångstrom exponent, (middle) single scattering albedo, and (bottom) ratio of SSA at 412 nm divided by SSA at 660 nm for three characteristically dusty regions [(a) the Sahara Desert (19°–30°N, 10°–30°E), (b) the Arabian Desert (23°–35°N, 40°–50°E), and (c) the Taklamakan Desert (36°–40°N, 77°–88°E)] and one anthropogenic aerosol–dominated region [(d) equatorial Africa (10°S–5°N, 10°–32°E)] for the period from 2006 to 2010 for daily C061 L3 MODIS Deep Blue products. Colored brown [in (a)–(c)] or green [in (d)] bars indicate bins that match the criterion used to identify dust for the respective optical property.

In all three dust-dominated regions, 97%–100% of pixels meet the criterion for SSA at 412 nm < 0.95; in the region dominated by other aerosol types, fewer than 26% of pixels meet this criterion. Similarly, in each dust-dominated region, more than 99% of pixels meet the criterion of SSA at 412 nm ≥ SSA at 660 nm, and only 40% meet this criterion for the nondust region. The criterion used in Ginoux et al. (2012) of AOD greater than 0.1 was not used in the creation of τdM here, because we were concerned that this threshold would systematically bias our results.

Ginoux et al. (2012) used a conservative requirement of AE less than 0 to identify dust. However, the updated algorithm used to create the C006 Deep Blue AE product now limits the valid range of values to 0 ≤ AE ≤ 1.8 (Sayer et al. 2013; Hsu et al. 2013). Therefore, we chose a new threshold that is based on the distribution of AE in our three desert regions. When this threshold was replaced with a criterion of AE less than 1, 90% of pixels in the Sahara Desert region, 91% of pixels in the Arabian Desert region, and 73% of pixels in the Taklamakan Desert region met the new threshold and only 5% met it in the nondust region. The new threshold of 1 we have used here is within the values of AE reported for dust regimes in Dubovik et al. (2002). In summary, daily estimates of τdM over land grid cells are composed of the Deep Blue AOD if that pixel met the following criteria:

  1. MODIS C061 AE 470–670 nm, which is inversely proportional to particle size, of less than 1,

  2. MODIS C061 SSA at 412 nm, the ratio of aerosol scattering to extinction coefficients at that wavelength, of less than 0.95, and

  3. MODIS C061 SSA at 412 nm is less than or equal to SSA at 660 nm.

If a pixel did not meet these criteria, its τd was set equal to zero. An analysis of the accuracy of this method, through comparison with AERONET, is presented in the online supplemental materials.

d. τdA (1981–2018)

The τdA over the ocean from 1981 to 2018 was derived similarly to what was previously described for MODIS. However, AOT from AVHRR was used rather than MODIS AOD, and the climatological seasonal mean f from the MODIS period, calculated from the MODIS small-mode 660-nm AOD divided by the total 660-nm AOD, was used in Eq. (3) because there is a spectral dependence of f. Evan and Mukhopadhyay (2010) showed that over the tropical North Atlantic, a dusty region, the seasonal cycle in f is far larger than the interannual variability, such that monthly climatological mean f can be used in place of monthly-mean f without introducing significant biases. Here we also assume that the use of monthly climatological f can be used in place of daily f. We tested this assumption by calculating the τdM using climatological f. The results were similar to τdM calculated using daily f with a bias of 0.00 in the long-term global mean over the ocean for the period 2001–17 (RMSE: 0.03). The characteristic fine-mode fractions, fm, fd, and fan, were the same as those used in the calculation of τdM.

Following Evan and Mukhopadhyay (2010), a time series of zonal mean stratospheric aerosol optical thickness τs as estimated in Sato et al. (1993) was removed from AVHRR AOT before calculation of τdA for years prior to 1999 and was assumed to be zero in subsequent years. This was done to remove the impact of large volcanic eruptions, including Pinatubo (June 1991) and El Chichón (March and April 1982), on τdA. τs was not removed in years following 1999 because there have been no volcanic eruptions of the scale of Pinatubo or El Chichón since 1999. The total column AOT peaks immediately following these major eruptions, but τs does not peak until 1–3 months later (Fig. 3). This offset results in a remaining peak in the τs-corrected AOT, which leads to a coincident peak in τdA. In our analysis of trends in τdA, we exclude 1982 and 1991 to avoid the impact of these events.

Fig. 3.

Uncorrected global mean time series of AVHRR AOT (τ; red), AVHRR retrieval solar zenith angle (black), and stratospheric AOD (τs; cyan) from Sato et al. (1993), each from monthly mean with seasonal cycle removed.

Fig. 3.

Uncorrected global mean time series of AVHRR AOT (τ; red), AVHRR retrieval solar zenith angle (black), and stratospheric AOD (τs; cyan) from Sato et al. (1993), each from monthly mean with seasonal cycle removed.

Our resultant global mean time series of τdA exhibits fluctuations in years prior to 1995 that are associated with satellite drift, a known issue with the pre–K, L, and M series (KLM) NOAA satellites, which has been investigated with respect to long-term variability in cloud cover (Foster and Heidinger 2013; Norris and Evan 2015). Satellite drift, or orbital decay, results in a progressively later equatorial crossing time. Thus, biases in optical depth retrievals may be due to retrieval errors associated with larger solar zenith angles (Fig. 3), or sampling of the diurnal cycle (Norris and Evan 2015). We remove the signal associated with satellite drift via linear regression, following the methods of Norris and Evan (2015). Consequently, at each location, the resultant τdA values are anomalies relative to the global mean.

3. Results

The primary results of this work are two nearly global, daily, observational datasets of τd at a 1° horizontal resolution that can be used for studies of the global dust cycle. The long-term mean global τdM over the ocean was found to be 0.03 ± 0.06 where ±0.06 is the ±1σ error estimate. For comparison, Ridley et al. (2016) found global dust AOD at 550 nm to be 0.03 ± 0.005 using in situ and satellite observations combined with global models. The long-term mean global τdM over land was found to be 0.1. The remainder of this section will be used to present a comparison of τdA and τdM (section 3a), verification of τdM against ground-based measurements of dust (section 3b), and a description of seasonal climatological mean τd (section 3c).

a. Comparison of τdM and τdA

Globally, monthly mean estimates of τdM and τdA agree well, as evidenced by a high correlation coefficient between the two data (r2 = 0.41), a linear least squares regression slope near unity [y = 0.90(±0.001)x], a low RMSE (0.07), and a low offset (−0.01) (Fig. 4a). Much of the disagreement between the two occurs in locations where there are few days on which both AVHRR and MODIS have retrievals (Fig. 5). This is often the case in latitudes above 30°N and below 30°S as well as in characteristically cloudy regions such as the ITCZ (0°–10°N), the southeastern Atlantic Ocean (10°–20°S, 10°W–10°E), and the southeastern boundary of the Pacific Ocean. In dust-dominated regions,τdM and τdA are very well correlated, but with a slope that is greater than 1. For example, in the tropical North Atlantic, the only ocean region where mean τdM is greater than 0.3 in all seasons (Fig. 10, described in more detail below), they are linearly related with a form y = 1.21(±0.005)x − 0.03 (r2 = 0.76, RMSE: 0.08, and offset: −0.00) (Fig. 4b). Monthly mean τdA is offset low (bias: −0.18) relative to τdM for τdM greater than 1. We hypothesize that the low offset in τdA is an artifact of misclassification of optically thick dust layers as clouds in the AVHRR AOT cloud clearing algorithm (X. Zhao 2018, personal communication), although it is also contributed to by the difference wavelengths between MODIS AOD retrievals and AVHRR AOT retrievals (Zhao et al. 2008). To determine boundaries around the potential contribution to the offset by the difference in wavelength of AOD retrieval, we perform a back-of-the-envelope calculation. The largest percentage difference in AOD at each wavelength should occur when the AE is large. We use AERONET 500- and 675-nm AOD at the (dust dominated) Tamanrassett station to calculate the mean and upper and lower quartiles of AE for these wavelengths, using only days that met the criteria used for ground-truth validation of dust (see the online supplemental material). In the absence of a 550–630-nm AE from MODIS, we assume that this 500–675-nm AE from AERONET is similar enough to approximate the influence of wavelength differences on the offset between MODIS and AVHRR. Using the lower-quartile AE at Tamaransett, 0.12, we find that the wavelength difference would contribute a 1.6% offset with τdM greater than τdA. Using the mean AE at Tamaransett, we find a potential offset of 3.5%; using the upper quartile, it is 5.8%.

Fig. 4.

Scatterplots of (a),(b) monthly mean and (c),(d) daily τdM and τdA for (top) 60°S–60°N and (bottom) a region over the tropical Atlantic where dust storms are frequent (5°S–20°N, 20°–30°W) for 2001–17. The solid red lines represent the linear least squares regression, and the corresponding slopes and offsets are indicated in each panel, along with RMSE.

Fig. 4.

Scatterplots of (a),(b) monthly mean and (c),(d) daily τdM and τdA for (top) 60°S–60°N and (bottom) a region over the tropical Atlantic where dust storms are frequent (5°S–20°N, 20°–30°W) for 2001–17. The solid red lines represent the linear least squares regression, and the corresponding slopes and offsets are indicated in each panel, along with RMSE.

Fig. 5.

Pearson correlation coefficient between monthly τdM and monthly τdA from 2001 to 2017 with the seasonal cycle removed. Regions where the correlation is not significant at the 95% confidence level are masked. Hatched areas indicate locations where daily matched observations from AVHRR and MODIS Terra are present for fewer than 50% of days in the period 2001–17.

Fig. 5.

Pearson correlation coefficient between monthly τdM and monthly τdA from 2001 to 2017 with the seasonal cycle removed. Regions where the correlation is not significant at the 95% confidence level are masked. Hatched areas indicate locations where daily matched observations from AVHRR and MODIS Terra are present for fewer than 50% of days in the period 2001–17.

We also calculated the linear least squares best fit between the daily τdM and daily τdA, including only dates and pixels where both sensors retrieved AOD (Fig. 4d). We expect that if the low offset of τdA relative to τdM is contributed to by missing data in regions of optically thick dust, the slope of the regression between these datasets will be significantly less steep for the daily data than for the monthly data in the tropical North Atlantic (Fig. 4d). This is because for the monthly mean, MODIS would incorporate high τd days that AVHRR “missed.” Indeed, we found that the daily τdM and daily τdA are related by the relationship y = 0.95(±0.0002)x − 0.01 (r2 = 0.54, RMSE: 0.13, and offset: −0.01) in the tropical North Atlantic. Globally, daily τdM and daily τdA are related by the relationship y = 0.70(±0.0003)x + 0.01 (r2 = 0.26, RMSE: 0.09, and offset: −0.01) (Fig. 4c).

The discrepancy in retrievals over large dust plumes can be easily visualized in the case of 27 June 2014, over the tropical Atlantic (Fig. 6). In this case, τdM indicates an optically thick dust plume, centered near 15°N, 30°W, that has been advected over the tropical Atlantic from the Sahara (Fig. 6a). This particularly strong event allowed for dust to be measured in situ as far as Colombia (Bedoya et al. 2016). In Fig. 6, grid cells that are missing estimates of τd or AOD, or grid cells over land, are masked in gray. For the same day, the τdA (Fig. 6b) is largely missing estimates in the region of optically thick dust. This is a result of missing retrievals of AVHRR AOT (Fig. 6d), which occur due to a combination of sun glint and because of the cloud-screening algorithm classifying optically thick dust as cloud.

Fig. 6.

The (a) τdM and (b) MODIS AOD from the Terra platform and (c) τdA and (d) AVHRR AOT for 27 Jun 2014. Retrievals and estimates over land and locations where there are missing data are shown in gray.

Fig. 6.

The (a) τdM and (b) MODIS AOD from the Terra platform and (c) τdA and (d) AVHRR AOT for 27 Jun 2014. Retrievals and estimates over land and locations where there are missing data are shown in gray.

b. Comparison with existing datasets

It is informative to compare our τd with previously published estimates of atmospheric dust. Here, we compare τdM with two recently published time series of dust concentrations over the western United States (Tong et al. 2017; Hand et al. 2016) and one over the Caribbean (Prospero and Nees 1986; Prospero et al. 1996). The measurements of dust over the western United States are based on surface concentrations of aerosol species. In Hand et al. (2016), iron from IMPROVE stations was used as a proxy for fine dust concentrations and τdM compares well to these annual mean fine dust concentrations (r value = 0.61; p value < 0.01) (Fig. 7). In Tong et al. (2017) several criteria from ground-based IMPROVE data were used to confirm dust events that had been detected in satellite imagery. The criteria included high PM10 and PM2.5 concentrations, a low ratio of PM2.5 and PM10, high concentrations of crustal elements (SI, CA, K, Fe, and Ti), low concentration of anthropogenic components, and low enrichment factors of pollution elements. Annual mean τdM is strongly correlated with estimates of annual number of dust events (r value = 0.73; p value < 0.05) from this work. Toth et al. (2014) found that globally, the average correlation between total AOD and AOD integrated from 500 m AGL to the surface was 0.61 over land. For the contiguous United States, this correlation was found to be 0.62, with 0.57 over the western time zone. Therefore, although τd is a measure of column integrated aerosol, we still expect surface concentrations to be correlated with column integrated values, especially when averaged over a large region. Interestingly, while an increasing trend in dust was noted for these studies for the period between 2001 and 2015, from these results, the extension of the time series through 2017 indicates that dust has been decreasing since 2011.

Fig. 7.

Annual mean τdM from land grid points in the southwestern United States within the region shown in the red-outlined box (30°–42°N, 125°–95°W) in comparison with number of dust events annually in the southwestern United States as reported in Tong et al. (2017) (green) and dust concentration from IMPROVE sites in the southwestern United States as reported in Hand et al. (2016) (blue). IMPROVE stations used in the Tong et al. (2017) analysis are marked with green stars on the inset map, and stations used in the Hand et al. (2016) analysis are marked with blue stars. There are several overlapping stations. The r value and p value of the correlation between tdM from Terra and each other time series are shown in the figure legend.

Fig. 7.

Annual mean τdM from land grid points in the southwestern United States within the region shown in the red-outlined box (30°–42°N, 125°–95°W) in comparison with number of dust events annually in the southwestern United States as reported in Tong et al. (2017) (green) and dust concentration from IMPROVE sites in the southwestern United States as reported in Hand et al. (2016) (blue). IMPROVE stations used in the Tong et al. (2017) analysis are marked with green stars on the inset map, and stations used in the Hand et al. (2016) analysis are marked with blue stars. There are several overlapping stations. The r value and p value of the correlation between tdM from Terra and each other time series are shown in the figure legend.

We also compared our new climatology of τdM with long-term, independent, ground-based dust measurements of dust taken at Ragged Point, Barbados (Prospero and Nees 1986; Prospero et al. 1996; Prospero 2014). These measurements are based on ash residue from extracted filters with an adjustment factor based on average crustal abundance of Aluminum in soil dust. These data have a standard error that is approximately constant at ±0.1 μ m−3 for concentrations less than 1 μg m−3 and about ±10% for higher concentrations. We found that in addition to matching the seasonal cycle of dust at this location (Fig. 8b), monthly mean τdM over Barbados is positively and statistically significantly correlated (r value = 0.50; p value < 0.01) with monthly mean ground-based dust aerosol measurements when the seasonal cycle is removed from both time series (Fig. 8a). Monthly mean τdA from 1982 to 2013 is also statistically correlated (r value = 0.33) with dust concentrations at Barbados, and shows the same seasonality there as τdM (Fig. S3 in the online supplemental material).

Fig. 8.

(a) Time series of monthly mean ground-based dust aerosol measurements on Barbados (red) in comparison with τdM over Barbados (blue), each with the seasonal cycle removed. (b) Climatological monthly mean, with long-term mean removed, of measurements on Barbados (red) and of τdM over Barbados (blue). The r value and p value for the correlation between these time series are shown in the bottom left-hand corner of (a).

Fig. 8.

(a) Time series of monthly mean ground-based dust aerosol measurements on Barbados (red) in comparison with τdM over Barbados (blue), each with the seasonal cycle removed. (b) Climatological monthly mean, with long-term mean removed, of measurements on Barbados (red) and of τdM over Barbados (blue). The r value and p value for the correlation between these time series are shown in the bottom left-hand corner of (a).

We also compared our dataset with a monthly time series of dust optical depth (DOD) over Syria as published by Pu and Ginoux (2016) (Fig. 9). Similar to Ginoux et al. (2012), Pu and Ginoux (2016) used MODIS Deep Blue C006 AOD, SSA, and AE to construct DOD, and thus the datasets are not independent. However, Pu and Ginoux (2016) updated their method by interpolating optical properties to a 0.1° × 0.1° grid and using a threshold of SSA at 470 nm along with an empirical relationship between AE and f to derive coarse DOD. The time series are related with a correlation coefficient of 0.87 (p value < 0.01) and show the same trend of increasing dust over Syria (Fig. 9). We are not aware of another independent estimate of τdA against which we can make a comparison.

Fig. 9.

Time series of monthly mean dust optical depth (brown) in comparison with τdM (blue) over Syria (34°–36.5°N, 36.5°–41°E), each with the seasonal cycle removed. The r value and p value for the correlation between these time series are shown in the bottom left-hand corner.

Fig. 9.

Time series of monthly mean dust optical depth (brown) in comparison with τdM (blue) over Syria (34°–36.5°N, 36.5°–41°E), each with the seasonal cycle removed. The r value and p value for the correlation between these time series are shown in the bottom left-hand corner.

c. Seasonal dust AOD

Having shown that this satellite-based climatology of dust appears to reflect changes in the atmospheric concentration of dust, we next summarize some of the broader characteristics of the global dust cycle, starting with seasonality. Maxima in seasonal mean τdM are found over the Sahara and the Arabian Peninsula in austral summer (July–August) (Fig. 10d) and over the Taklamakan and Gobi Deserts during spring (March–May) (Fig. 10c). Aeolian transport from the Sahara toward the Americas is visible during all seasons with a maximum extension westward during the summer (Figs. 10d,h), which is well known (Prospero and Mayol-Bracero 2013; Prospero et al. 1996). Seasonal mean τdA, appears far lower than τdM near to major dust sources in all seasons, due to limited retrievals in regions of optically thick dust as described in the previous section. The spring maximum in dust over Asia can be attributed to the seasonal cycle of the Mongolian cyclonic depression and the related high frequency of springtime cold air outbreaks (Sun et al. 2001; Ge et al. 2014). During this season, dust is transported from the Asian continent over the North Pacific Ocean toward North America at latitudes between 30° and 50°N. It is presumed that this dust is primarily from the Taklimakan Desert as the local topographical and meteorological conditions allow for it to be entrained to elevations exceeding 5 km while dust from the Gobi Desert is, for the most part, confined closer to the surface (<3 km) (Sun et al. 2001). It has also been shown that some portion of this trans-Pacific dust originates from Africa (Creamean et al. 2013).

Fig. 10.

Seasonal mean (a)–(d) τdM averaged over the period from 2001 to 2018 and (e)–(h) τdA over the period from 1981 to 2018. The long-term global mean τd ± 1σ uncertainty over the ocean for each season and dataset is featured in the bottom left-hand corner of each panel.

Fig. 10.

Seasonal mean (a)–(d) τdM averaged over the period from 2001 to 2018 and (e)–(h) τdA over the period from 1981 to 2018. The long-term global mean τd ± 1σ uncertainty over the ocean for each season and dataset is featured in the bottom left-hand corner of each panel.

Discrepancies between τdM over land and τdM over adjacent ocean are visible in some areas with significant anthropogenic aerosol sources, including the Sahel region of Africa and India during austral autumn and spring (Figs. 10a,c). This may be due to different methods and data sources used for estimating τdM over land and ocean. The method used for τdM over the ocean leads to a continuous range of τdM, whereas the threshold approach used for estimation of τdM over land may prevent these gradients when there is mixing between anthropogenic aerosols and dust aerosol. However, this may also partially be the result of the different algorithms used in the retrieval of AOD over land and ocean (Sayer et al. 2013; Hsu et al. 2013) (Fig. S4 in the online supplemental material). While these unphysical discrepancies could be avoided through spatial interpolation across these boundaries, we have chosen not to do this in order to minimize our assumptions and retain a daily dataset that is directly representative of the satellite observations. At the very least, these results suggest that direct comparison of overland and overwater τd may be problematic.

A region of high summertime mean τdM is visible over the North Pacific Ocean near 45°N (Fig. 10d). Summertime trans-Pacific transport of dust has been observed (Yumimoto et al. 2010). However, visual inspection of MODIS Terra corrected reflectance on days when τdM over the western North Pacific was high during summer months indicated the possibility of contamination from biomass burning over Europe and Russia. We compared summer seasonal mean τdM and MERRA-2 organic carbon extinction optical thickness at 550 nm (online supplemental Fig. S5; https://giovanni.gsfc.nasa.gov/) and confirmed that summer seasonal mean organic carbon was highest over the western North Pacific in years when summer seasonal mean τdM was highest. This supports the hypothesis that there may be biomass-burning contamination over the western North Pacific during these months. The fine-mode fraction f during these summer high τdM events is approximately 0.57–0.63. In the springtime, most of the high τdM events are associated with fine-mode fractions near 0.50, which is closer to fd.

In addition, because of extensive low cloud cover this region in the western North Pacific has few retrievals during the summer months. The limited number of retrievals is evident in the number of pixels used in the estimate of MODIS Dark Target AOD during the summer in that region. While greater than 36 pixels are incorporated on average in the equatorial Pacific estimations of MODIS Dark Target AOD, only approximately 26 on average are used in this estimation in parts of the western North Pacific. Long-term seasonal mean τdA appears low in the western North Pacific in all seasons (Figs. 10c–f).

4. Discussion

Having presented two datasets of τd derived from satellite observations, we next present regional trends in τd (section 4a) and a comparison of τdM with a modern reanalysis dust extinction aerosol optical depth product (section 4b).

a. Trends in τdM

In Fig. 11a we show regional trends in monthly mean τdM from Terra between 2001 and 2018, where the seasonal cycle has been removed. Only trends that are significant at the 95% confidence level are shown. One region that stands out with a large upward trend in dust is the Arabian Peninsula, where values exceed 0.3 (10 yr)−1 for the MODIS period. This is consistent with previous studies that found increasing trends, of smaller magnitude, in τ in this region (Hsu et al. 2012; Alfaro-Contreras et al. 2017; Toth et al. 2016). Parts of the Saharan Desert appear to be getting dustier in τdM from Aqua. However, coastal Northern Africa and the edges of the Sahara display decreasing trends in τdM from Terra (Fig. 11). Our results show a statistically significant decrease in τdM over northern China (approximately 100°E, 40°N), and over southwest Asia (approximately 180°E, 30°N). This is consistent with recent findings by Pandey et al. (2017), who found that premonsoon dust loading decreased between 2000 and 2015. This decrease has been linked to increased rainfall, and resultant wet scavenging and increased soil moisture, and, to a lesser extent, decreased wind strength. The decrease in τdM over northeast Asia may also be the continuation of a documented decreasing trend from the late 1950s through the 1980s in the region due to reduced cyclone frequency (Qian et al. 2002).

Fig. 11.

Trend in (a) τdM Terra (2001–18), (b) τdM Aqua (2003–18), and (c) τdM (1981–2018) with the seasonal cycle removed. Only trends that are significant at the 95% confidence level are shown.

Fig. 11.

Trend in (a) τdM Terra (2001–18), (b) τdM Aqua (2003–18), and (c) τdM (1981–2018) with the seasonal cycle removed. Only trends that are significant at the 95% confidence level are shown.

Interestingly, the greatest increase in τdM (1.04 units per 10 yr) is seen over the Aral Sea at the intersection of Kazakhstan and Uzbekistan (44.5°N, 59.5°E). This was once the location of one of the world’s largest lakes, spanning over 66 000 km2 (Izhitskiy et al. 2016). However, the lake has been drying up since the early 1960s, when irrigation projects diverted the rivers that fed it leaving behind a highly emissive exposed playa (Micklin 1988; Indoitu et al. 2015). The eastern basin of the lake dried up completely in 2014, but the northern parts of the sea remain filled and have largely stabilized. A dramatic increase in τdM is also seen over Razzaza Lake in Iraq, where the shoreline is receding due to diverted irrigation water. The northern-most region of the Caspian Sea, which is also the shallowest portion of the sea, shows a smaller but significant increase in τdM potentially due to the recession of this body of water over the past two decades (Chen et al. 2017).

τd derived from MODIS Aqua (2003–18) displays the same trends of decreasing dust over northeast China and southwest Asia and increasing trends over the Arabian Peninsula (approximately 20°N, 70°E) and central Sahara (approximately 20°N, 10°E) (Fig. 11b). There are some regional changes that are more pronounced in this record, including a statistically significant increase in τd over Southern California (approximately 34°N, 119°W). There is also a large increase in dust over Oman (0.20; approximately 20°N, 57°E) that is not present in the τdM derived from Terra and a greater increase over Iran than is present in the τdM derived from Terra. Differences in regional trends between these two datasets could be due the differences in the length of record, as well as the differences in overpass time for the two satellites from which the datasets are derived as it relates to the daily cycle of dust emission (Heinold et al. 2013; Kocha et al. 2013). Terra has a morning orbit while Aqua has an afternoon orbit (Ichoku 2005). However, it has been shown that although there is no significant overall trend, the offset between the Terra and Aqua C6 total AOD records has oscillated direction over time, and its variability has increased (Levy et al. 2018). This is likely to impact the differences in τdM from Aqua and Terra.

Trends in overwater τd, from 1981 to 2018, were also calculated using monthly mean τdA (Fig. 11c). Over these longer time periods, there is a small increasing trend in τdA near the Arabian Peninsula [0.02 (10 yr)−1] and a decreasing trend in τdA in the tropical North Atlantic near to the African coast [0.01 (10 yr)−1]. The decreasing trend over offshore of northern Africa is consistent with the time series of dust at Cape Verde (16.5°N, 23.04°W) created using a similar method (Evan and Mukhopadhyay 2010), and with another study that utilized satellite and coastal surface measurements (Chin et al. 2014).

We note that Zhao et al. (2013) found a slight positive trend in the CDR AVHRR AOT record that is the result of residual cloud contamination. As such, trends in the tropical North Atlantic and Arabian Peninsula could be similarly biased.

b. Comparison with a modern reanalysis dust product

To understand whether this observation-based dataset is consistent with a modern reanalysis dust product, we compared τdM with MERRA-2 dust extinction aerosol optical thickness at 550 nm (https://giovanni.gsfc.nasa.gov/giovanni/) over two of the regions—Barbados and the southwestern United States—used previously in our comparisons with ground observations (Figs. 7 and 8). Our monthly mean τdM, with the seasonal cycle removed, was significantly correlated at the 99% confidence level (r value = 0.73) to the monthly mean MERRA-2 dust extinction aerosol optical thickness over Barbados (Fig. 12a). However, the annual mean τdM displays greater interannual variability and was much more similar to the ground observational record than the MERRA-2 dust extinction aerosol optical thickness over the southwestern United States (Fig. 12b). Over this location, the MERRA-2 dust record was not statistically significantly correlated (r value = 0.20) with τdM.

Fig. 12.

(a) Monthly MERRA-2 dust extinction aerosol optical thickness at 550 nm and monthly mean τdM over Ragged Point between 2001 and 2013 with the seasonal cycle removed. (b) Annual mean MERRA-2 dust extinction aerosol optical thickness at 500 nm and annual mean τdM over the southwestern United States (red-outlined box) from 2001 to 2016.

Fig. 12.

(a) Monthly MERRA-2 dust extinction aerosol optical thickness at 550 nm and monthly mean τdM over Ragged Point between 2001 and 2013 with the seasonal cycle removed. (b) Annual mean MERRA-2 dust extinction aerosol optical thickness at 500 nm and annual mean τdM over the southwestern United States (red-outlined box) from 2001 to 2016.

5. Conclusions

These global, daily, observation-based datasets of dust aerosol optical depth are rooted in observations of optical characteristics of the aerosol column from Terra Moderate resolution Imaging Spectroradiometer and from the Advanced Very High Resolution Radiometer, but incorporate observations of characteristic fine-mode fraction in dusty, clean marine, and anthropogenically dominated regions from the Aerosol Robotic Network as well as surface wind speed from the second Modern-Era Retrospective Analysis for Research and Applications. As previously mentioned, the long-term mean global τdM over the ocean was found to be 0.03 ± 0.06. Although there are few measurements of dust aerosol for comparison, τdM over land compares well to ground-based measurements of dust from the Interagency Monitoring of Protected Visual Environments network in the western United States and τdM over ocean compares well to independent ground-based measurements made at Ragged Point.

We had hoped to create a reliable dataset of τdA extending from 1981 to 2018 using AVHRR aerosol optical thickness from the Climate Data Record AOT dataset, but we found that the AVHRR AOT had limited coverage in dusty regions. Comparison of daily AVHRR AOT and MODIS AOD over the tropical Atlantic indicated that this limited coverage is at least partially explained by optically thick dust identified as cloud in the AVHRR cloud masking process. This limited coverage results in a low offset in seasonal and monthly mean τdA when compared with τdM. We found that τdM increased in the central Sahara and the Middle East and decreased over northern China and southwestern Asia between 2001 and 2018. The largest increase in τd occurred over the Aral Sea, which has receded, leaving emissive playa surfaces exposed. We found decreasing trends in τdA near Africa in the equatorial Atlantic and increasing trends near the Arabian Peninsula from 1981 to 2018.

In 2011, the Suomi National Polar-Orbiting Partnership (SNPP) satellite was launched carrying the Visible Infrared Imaging Radiometer Suite (VIIRS), which is intended to continue the observational records that have been provided by sensors such as MODIS and AVHRR (Sayer et al. 2017). VIIRS has been shown to be capable of identifying aerosol type and partitioning between fine- and coarse-mode aerosol. As such, observational studies that require a shorter record (beginning in 2012) may be able to rely on these products. In addition, Chen et al. (2018) demonstrated that an inversion technique, utilizing the GOES-Chem model, can be applied to certain satellite aerosol optical property retrievals in order to retrieve desert dust. This was applied to one year of satellite-derived aerosol information generated by the General Retrieval of Atmosphere and Surface Properties (GRASP) algorithm from Polarization and Anisotropy of Reflectances for Atmospheric Sciences coupled with Observations from a Lidar (POLDER/PARASOL) retrievals. This technique can provide daily dust at 2° × 2° resolution with uncertainty below 25.8%, but it is computationally costly and has not been applied to longer records of remotely sensed aerosol optical properties.

The current work provides daily τd generated from a simple algorithm and observed aerosol data. Previous estimates of τd have enhanced our understanding of the global dust cycle and have been a valuable tool for evaluating the performance of global models (Pu and Ginoux 2018a; Evan and Mukhopadhyay 2010; Kaufman 2005; Ginoux et al. 2012). This longer, higher-temporal-resolution record will facilitate additional advances in our understanding of the dust cycle.

Acknowledgments

This work was funded by the California Department of Water Resources Contract 4600010378, Task Order OSCOP215, and the Army Corps of Engineers USACE (CESU) W912HZ–15–0019. We thank the AERONET principal investigators and their staff for establishing and maintaining the 39 sites used in this investigation. We are grateful for the datasets and data-archiving centers that supported this work and appreciate those who made our study possible, including the MERRA-2 team at the GMAO and staff at GSFC, and the MODIS teams at NASA, as well as Tom Zhao and the NOAA Climate Data Record team. We thank Drs. Joseph Prospero, Jennifer Hand, and Daniel Tong for their contributions of long-term datasets for comparisons made in this work, and Drs. Martin Ralph, Leah Campbell, and Nora Mascioli for helpful comments on this paper. We also thank three anonymous reviewers for helpful comments on this paper. The datasets described in this paper will be made publicly available at Pangea Open Access (https://doi.pangaea.de/10.1594/PANGAEA.909140).

REFERENCES

REFERENCES
Ai
,
N.
, and
K. R.
Polenske
,
2008
:
Socioeconomic impact analysis of yellow-dust storms: An approach and case study for Beijing
.
Econ. Syst. Res.
,
20
,
187
203
, https://doi.org/10.1080/09535310802075364.
Alfaro-Contreras
,
R.
,
J.
Zhang
,
J. S.
Reid
, and
S.
Christopher
,
2017
:
A study of 15-year aerosol optical thickness and direct shortwave aerosol radiative effect trends using MODIS, MISR, CALIOP and CERES
.
Atmos. Chem. Phys.
,
17
,
13 849
13 868
, https://doi.org/10.5194/acp-17-13849-2017.
Angstrom
,
A.
,
1929
:
On the atmospheric transmission of sun radiation and on dust in the air
.
Geogr. Ann.
,
11
,
156
166
, https://doi.org/10.1080/20014422.1929.11880498.
Ault
,
A. P.
,
C. R.
Williams
,
A. B.
White
,
P. J.
Neiman
,
J. M.
Creamean
,
C. J.
Gaston
,
F. M.
Ralph
, and
K. A.
Prather
,
2011
:
Detection of Asian dust in California orographic precipitation
.
J. Geophys. Res.
,
116
,
D16205
, https://doi.org/10.1029/2010JD015351.
Bedoya
,
A.
, and Coauthors
,
2016
:
Strong Saharan dust event detected at Lalinet LOA-UNAL station, over Medellín, Colombia by active and passive remote sensing. 27th Int. Laser Radar Conf., New York, New York, NOAA CREST, https://doi.org/10.1051/epjconf/201611908006
.
Bosilovich
,
M. G.
, and Coauthors
,
2015
:
MERRA-2: Initial evaluation of the climate. Technical Report Series on Global Modeling and Data Assimilation, Vol. 43, NASA Tech Rep. NASA/TM-2015-104606, 145 pp., https://gmao.gsfc.nasa.gov/pubs/docs/Bosilovich803.pdf
.
Bréon
,
F.-M.
,
A.
Vermeulen
, and
J.
Descloitres
,
2011
:
An evaluation of satellite aerosol products against sunphotometer measurements
.
Remote Sens. Environ.
,
115
,
3102
3111
, https://doi.org/10.1016/j.rse.2011.06.017.
Brown
,
J. K. M.
,
2002
:
Aerial dispersal of pathogens on the global and continental scales and its impact on plant disease
.
Science
,
297
,
537
541
, https://doi.org/10.1126/science.1072678.
Cazorla
,
A.
,
R.
Bahadur
,
K. J.
Suski
,
J. F.
Cahill
,
D.
Chand
,
B.
Schmid
,
V.
Ramanathan
, and
K. A.
Prather
,
2013
:
Relating aerosol absorption due to soot, organic carbon, and dust to emission sources determined from in-situ chemical measurements
.
Atmos. Chem. Phys.
,
13
,
9337
9350
, https://doi.org/10.5194/acp-13-9337-2013.
Chan
,
P. K.
,
X.-P.
Zhao
, and
A. K.
Heidinger
,
2013
:
Long-term aerosol climate data record derived from operational AVHRR satellite observations
.
Dataset Pap. Geosci.
,
140791
, https://doi.org/10.7167/2013/140791.
Chaubey
,
J. P.
,
K. K.
Moorthy
,
S. S.
Babu
, and
V. S.
Nair
,
2011
:
The optical and physical properties of atmospheric aerosols over the Indian Antarctic stations during Southern Hemispheric summer of the International Polar Year 2007–2008
.
Ann. Geophys.
,
29
,
109
121
, https://doi.org/10.5194/angeo-29-109-2011.
Chen
,
C.
, and Coauthors
,
2018
:
Retrieval of desert dust and carbonaceous aerosol emissions over Africa from POLDER/PARASOL products generated by the GRASP algorithm
.
Atmos. Chem. Phys.
,
18
,
12 551
12 580
, https://doi.org/10.5194/acp-18-12551-2018.
Chen
,
G.
, and Coauthors
,
2011
:
Observations of Saharan dust microphysical and optical properties from the eastern Atlantic during NAMMA airborne field campaign
.
Atmos. Chem. Phys.
,
11
,
723
740
, https://doi.org/10.5194/acp-11-723-2011.
Chen
,
J. L.
,
T.
Pekker
,
C. R.
Wilson
,
B. D.
Tapley
,
A. G.
Kostianoy
,
J.-F.
Cretaux
, and
E. S.
Safarov
,
2017
:
Long-term Caspian Sea level change
.
Geophys. Res. Lett.
,
44
,
6993
7001
, https://doi.org/10.1002/2017GL073958.
Chin
,
M.
, and Coauthors
,
2014
:
Multi-decadal aerosol variations from 1980 to 2009: A perspective from observations and a global model
.
Atmos. Chem. Phys.
,
14
,
3657
3690
, https://doi.org/10.5194/acp-14-3657-2014.
Creamean
,
J. M.
, and Coauthors
,
2013
:
Dust and biological aerosols from the Sahara and Asia influence precipitation in the western U.S
.
Science
,
339
,
1572
1578
, https://doi.org/10.1126/science.1227279.
Dubovik
,
O.
,
B.
Holben
,
T. F.
Eck
,
A.
Smirnov
,
Y. J.
Kaufman
,
M. D.
King
,
D.
Tanré
, and
I.
Slutsker
,
2002
:
Variability of absorption and optical properties of key aerosol types observed in worldwide locations
.
J. Atmos. Sci.
,
59
,
590
608
, https://doi.org/10.1175/1520-0469(2002)059<0590:VOAAOP>2.0.CO;2.
Evan
,
A. T.
, and
S.
Mukhopadhyay
,
2010
:
African dust over the northern tropical Atlantic: 1955–2008
.
J. Appl. Meteor. Climatol.
,
49
,
2213
2229
, https://doi.org/10.1175/2010JAMC2485.1.
Evan
,
A. T.
,
J.
Dunion
,
J. A.
Foley
,
A. K.
Heidinger
, and
C. S.
Velden
,
2006
:
New evidence for a relationship between Atlantic tropical cyclone activity and African dust outbreaks
.
Geophys. Res. Lett.
,
33
,
L19813
, https://doi.org/10.1029/2006GL026408.
Evan
,
A. T.
,
C.
Flamant
,
S.
Fiedler
, and
O.
Doherty
,
2014
:
An analysis of aeolian dust in climate models
.
Geophys. Res. Lett.
,
41
,
5996
6001
, https://doi.org/10.1002/2014GL060545.
Formenti
,
P.
,
W.
Elbert
,
W.
Maenhaut
,
J.
Haywood
, and
M. O.
Andreae
,
2003
:
Chemical composition of mineral dust aerosol during the Saharan Dust Experiment (SHADE) airborne campaign in the Cape Verde region, September 2000
.
J. Geophys. Res.
,
108
,
8576
, https://doi.org/10.1029/2002JD002648.
Formenti
,
P.
, and Coauthors
,
2008
:
Regional variability of the composition of mineral dust from western Africa: Results from the AMMA SOP0/DABEX and DODO field campaigns
.
J. Geophys. Res.
,
113
,
D00C13
, https://doi.org/10.1029/2008JD009903.
Foster
,
M. J.
, and
A.
Heidinger
,
2013
:
PATMOS-x: Results from a diurnally corrected 30-yr satellite cloud climatology
.
J. Climate
,
26
,
414
425
, https://doi.org/10.1175/JCLI-D-11-00666.1.
Ge
,
J. M.
,
J. P.
Huang
,
C. P.
Xu
,
Y. L.
Qi
, and
H. Y.
Liu
,
2014
:
Characteristics of Taklimakan dust emission and distribution: A satellite and reanalysis field perspective
.
J. Geophys. Res.
,
119
,
11 772
11 783
, https://doi.org/10.1002/2014JD022280.
Gelaro
,
R.
, and Coauthors
,
2017
:
The Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2)
.
J. Climate
,
30
,
5419
5454
, https://doi.org/10.1175/JCLI-D-16-0758.1.
Ginoux
,
P.
,
2003
:
Empirical TOMS index for dust aerosol: Applications to model validation and source characterization
.
J. Geophys. Res.
,
108
,
4534
, https://doi.org/10.1029/2003JD003470.
Ginoux
,
P.
,
J. M.
Prospero
,
T. E.
Gill
,
N. C.
Hsu
, and
M.
Zhao
,
2012
:
Global-scale attribution of anthropogenic and natural dust sources and their emission rates based on MODIS Deep Blue aerosol products
.
Rev. Geophys.
,
50
,
RG3005
, https://doi.org/10.1029/2012RG000388.
Glantz
,
P.
,
E.
Nilsson
, and
W.
von Hoyningen-Huene
,
2009
:
Estimating a relationship between aerosol optical thickness and surface wind speed over the ocean
.
Atmos. Res.
,
92
,
58
68
, https://doi.org/10.1016/j.atmosres.2008.08.010.
Gong
,
S. L.
,
2003
:
A parameterization of sea-salt aerosol source function for sub- and super-micron particles
.
Global Biogeochem. Cycles
,
17
,
1097
, https://doi.org/10.1029/2003GB002079.
Griffin
,
D. W.
,
2007
:
Atmospheric movement of microorganisms in clouds of desert dust and implications for human health
.
Clin. Microbiol. Rev.
,
20
,
459
477
, https://doi.org/10.1128/CMR.00039-06.
Grythe
,
H.
,
J.
Ström
,
R.
Krejci
,
P.
Quinn
, and
A.
Stohl
,
2014
:
A review of sea-spray aerosol source functions using a large global set of sea salt aerosol concentration measurements
.
Atmos. Chem. Phys.
,
14
,
1277
1297
, https://doi.org/10.5194/acp-14-1277-2014.
Hand
,
J. L.
,
W. H.
White
,
K. A.
Gebhart
,
N. P.
Hyslop
,
T. E.
Gill
, and
B. A.
Schichtel
,
2016
:
Earlier onset of the spring fine dust season in the southwestern United States
.
Geophys. Res. Lett.
,
43
,
4001
4009
, https://doi.org/10.1002/2016GL068519.
Heidinger
,
A. K.
,
M. J.
Foster
,
A.
Walther
, and
X. T.
Zhao
,
2014
:
The Pathfinder Atmospheres–Extended AVHRR climate dataset
.
Bull. Amer. Meteor. Soc.
,
95
,
909
922
, https://doi.org/10.1175/BAMS-D-12-00246.1.
Heinold
,
B.
,
P.
Knippertz
,
J. H.
Marsham
,
S.
Fiedler
,
N. S.
Dixon
,
K.
Schepanski
,
B.
Laurent
, and
I.
Tegen
,
2013
:
The role of deep convection and nocturnal low-level jets for dust emission in summertime West Africa: Estimates from convection-permitting simulations
.
J. Geophys. Res.
,
118
,
4385
4400
, https://doi.org/10.1002/jgrd.50402.
Holben
,
B. N.
, and Coauthors
,
1998
:
AERONET—A federated instrument network and data archive for aerosol characterization
.
Remote Sens. Environ.
,
66
,
1
16
, https://doi.org/10.1016/S0034-4257(98)00031-5.
Holben
,
B. N.
, and Coauthors
,
2001
:
An emerging ground-based aerosol climatology: Aerosol Optical Depth from AERONET
.
J. Geophys. Res.
,
106
,
12 067
12 097
, https://doi.org/10.1029/2001JD900014.
Hsu
,
N. C.
,
R.
Gautam
,
A. M.
Sayer
,
C.
Bettenhausen
,
C.
Li
,
M. J.
Jeong
,
S.-C.
Tsay
, and
B. N.
Holben
,
2012
:
Global and regional trends of aerosol optical depth over land and ocean using SeaWiFS measurements from 1997 to 2010
.
Atmos. Chem. Phys.
,
12
,
8037
8053
, https://doi.org/10.5194/acp-12-8037-2012.
Hsu
,
N. C.
,
M.-J.
Jeong
,
C.
Bettenhausen
,
A. M.
Sayer
,
R.
Hansell
,
C. S.
Seftor
,
J.
Huang
, and
S.-C.
Tsay
,
2013
:
Enhanced Deep Blue aerosol retrieval algorithm: The second generation
.
J. Geophys. Res.
,
118
,
9296
9315
, https://doi.org/10.1002/jgrd.50712.
Hsu
,
N. C.
,
A. M.
Sayer
,
C.
Bettenhausen
,
J.
Lee
,
R. C.
Levy
,
S.
Mattoo
,
L. A.
Munchak
, and
R.
Kleidman
,
2014
:
Comparing MODIS C6 ‘Deep Blue’ and ‘Dark Target’ aerosol data. NASA Tech. Rep., 1 pp., https://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/20140012657.pdf
.
Hsu
,
N. C.
,
J.
Lee
,
A. M.
Sayer
,
N.
Carletta
,
S.-H.
Chen
,
C. J.
Tucker
,
B. N.
Holben
, and
S.-C.
Tsay
,
2017
:
Retrieving near-global aerosol loading over land and ocean from AVHRR
.
J. Geophys. Res.
,
122
,
9968
9989
, https://doi.org/10.1002/2017JD026932.
Huang
,
H.
,
G. E.
Thomas
, and
R. G.
Grainger
,
2010
:
Relationship between wind speed and aerosol optical depth over remote ocean
.
Atmos. Chem. Phys.
,
10
,
5943
5950
, https://doi.org/10.5194/acp-10-5943-2010.
Huneeus
,
N.
, and Coauthors
,
2011
:
Global dust model intercomparison in AeroCom phase I
.
Atmos. Chem. Phys.
,
11
,
7781
7816
, https://doi.org/10.5194/acp-11-7781-2011.
Ichoku
,
C.
,
2005
:
Quantitative evaluation and intercomparison of morning and afternoon Moderate Resolution Imaging Spectroradiometer (MODIS) aerosol measurements from Terra and Aqua
.
J. Geophys. Res.
,
110
,
D10S03
, https://doi.org/10.1029/2004JD004987.
Indoitu
,
R.
,
G.
Kozhoridze
,
M.
Batyrbaeva
,
I.
Vitkovskaya
,
N.
Orlovsky
,
D.
Blumberg
, and
L.
Orlovsky
,
2015
:
Dust emission and environmental changes in the dried bottom of the Aral Sea
.
Aeolian Res.
,
17
,
101
115
, https://doi.org/10.1016/j.aeolia.2015.02.004.
Izhitskiy
,
A. S.
, and Coauthors
,
2016
:
Present state of the Aral Sea: Diverging physical and biological characteristics of the residual basins
.
Sci. Rep.
,
6
,
23906
, https://doi.org/10.1038/srep23906.
Jaeglé
,
L.
,
P. K.
Quinn
,
T. S.
Bates
,
B.
Alexander
, and
J.-T.
Lin
,
2011
:
Global distribution of sea salt aerosols: New constraints from in situ and remote sensing observations
.
Atmos. Chem. Phys.
,
11
,
3137
3157
, https://doi.org/10.5194/acp-11-3137-2011.
Kahn
,
R. A.
,
M. J.
Garay
,
D. L.
Nelson
,
K. K.
Yau
,
M. A.
Bull
,
B. J.
Gaitley
,
J. V.
Martonchik
, and
R. C.
Levy
,
2007
:
Satellite-derived aerosol optical depth over dark water from MISR and MODIS: Comparisons with AERONET and implications for climatological studies
.
J. Geophys. Res.
,
112
,
D18205
, https://doi.org/10.1029/2006JD008175.
Kaufman
,
Y. J.
,
2005
:
Dust transport and deposition observed from the Terra-Moderate Resolution Imaging Spectroradiometer (MODIS) spacecraft over the Atlantic Ocean
.
J. Geophys. Res.
,
110
,
D10S12
, https://doi.org/10.1029/2003JD004436.
King
,
M. D.
,
S.
Platnick
,
W. P.
Menzel
,
S. A.
Ackerman
, and
P. A.
Hubanks
,
2013
:
Spatial and temporal distribution of clouds observed by MODIS onboard the Terra and Aqua satellites
.
IEEE Trans. Geosci. Remote Sens.
,
51
,
3826
3852
, https://doi.org/10.1109/TGRS.2012.2227333.
Klaver
,
A.
, and Coauthors
,
2011
:
Physico-chemical and optical properties of Sahelian and Saharan mineral dust: In situ measurements during the GERBILS campaign
.
Quart. J. Roy. Meteor. Soc.
,
137
,
1193
1210
, https://doi.org/10.1002/qj.889.
Kleidman
,
R. G.
,
N. T.
O’Neill
,
L. A.
Remer
,
Y. J.
Kaufman
,
T. F.
Eck
,
D.
Tanré
,
O.
Dubovik
, and
B. N.
Holben
,
2005
:
Comparison of Moderate Resolution Imaging Spectroradiometer (MODIS) and Aerosol Robotic Network (AERONET) remote-sensing retrievals of aerosol fine mode fraction over ocean
.
J. Geophys. Res.
,
110
,
D22205
, https://doi.org/10.1029/2005JD005760.
Kocha
,
C.
,
P.
Tulet
,
J.-P.
Lafore
, and
C.
Flamant
,
2013
:
The importance of the diurnal cycle of aerosol optical depth in West Africa
.
Geophys. Res. Lett.
,
40
,
785
790
, https://doi.org/10.1002/grl.50143.
Kok
,
J. F.
,
2010
:
A scaling theory for the size distribution of emitted dust aerosols suggests climate models underestimate the size of the global dust cycle
.
Proc. Natl. Acad. Sci. USA
,
108
,
1016
1021
, https://doi.org/10.1073/pnas.1014798108.
Kok
,
J. F.
, and Coauthors
,
2017
:
Smaller desert dust cooling effect estimated from analysis of dust size and abundance
.
Nat. Geosci.
,
10
,
274
278
, https://doi.org/10.1038/ngeo2912.
Lehahn
,
Y.
,
I.
Koren
,
E.
Boss
,
Y.
Ben-Ami
, and
O.
Altaratz
,
2010
:
Estimating the maritime component of aerosol optical depth and its dependency on surface wind speed using satellite data
.
Atmos. Chem. Phys.
,
10
,
6711
6720
, https://doi.org/10.5194/acp-10-6711-2010.
Levy
,
R. C.
,
S.
Mattoo
,
L. A.
Munchak
,
L. A.
Remer
,
A. M.
Sayer
,
F.
Patadia
, and
N. C.
Hsu
,
2013
:
The Collection 6 MODIS aerosol products over land and ocean
.
Atmos. Meas. Tech.
,
6
,
2989
3034
, https://doi.org/10.5194/amt-6-2989-2013.
Levy
,
R. C.
,
S.
Mattoo
,
V.
Sawyer
,
Y.
Shi
,
P. R.
Colarco
,
A. I.
Lyapustin
,
Y.
Wang
, and
L. A.
Remer
,
2018
:
Exploring systematic offsets between aerosol products from the two MODIS sensors
.
Atmos. Meas. Tech.
,
11
,
4073
4092
, https://doi.org/10.5194/amt-11-4073-2018.
Mahowald
,
N.
,
2011
:
Aerosol indirect effect on biogeochemical cycles and climate
.
Science
,
334
,
794
796
, https://doi.org/10.1126/science.1207374.
Malm
,
W. C.
,
J. F.
Sisler
,
D.
Huffman
,
R. A.
Eldred
, and
T. A.
Cahill
,
1994
:
Spatial and seasonal trends in particle concentration and optical extinction in the United States
.
J. Geophys. Res.
,
99
,
1347
, https://doi.org/10.1029/93JD02916.
Mani
,
M.
, and
R.
Pillai
,
2010
:
Impact of dust on solar photovoltaic (PV) performance: Research status, challenges and recommendations
.
Renewable Sustainable Energy Rev.
,
14
,
3124
3131
, https://doi.org/10.1016/j.rser.2010.07.065.
Mårtensson
,
E. M.
,
E. D.
Nilsson
,
G.
de Leeuw
,
L. H.
Cohen
, and
H.-C.
Hansson
,
2003
:
Laboratory simulations and parameterization of the primary marine aerosol production
.
J. Geophys. Res.
,
108
,
4297
, https://doi.org/10.1029/2002JD002263.
Micklin
,
P.
,
1988
:
Desiccation of the Aral Sea: A water management disaster in the Soviet Union
.
Science
,
241
,
1170
1176
, https://doi.org/10.1126/science.241.4870.1170.
Miller
,
R. L.
,
I.
Tegen
, and
J.
Perlwitz
,
2004
:
Surface radiative forcing by soil dust aerosols and the hydrologic cycle
.
J. Geophys. Res.
,
109
,
D04203
, https://doi.org/10.1029/2003JD004085.
Moorthy
,
K. K.
, and
S. K.
Satheesh
,
2000
:
Characteristics of aerosols over a remote island, Minicoy in the Arabian Sea: Optical properties and retrieved size characteristics
.
Quart. J. Roy. Meteor. Soc.
,
126
,
81
109
, https://doi.org/10.1002/qj.49712656205.
Mulcahy
,
J. P.
,
C. D.
O’Dowd
,
S. G.
Jennings
, and
D.
Ceburnis
,
2008
:
Significant enhancement of aerosol optical depth in marine air under high wind conditions
.
Geophys. Res. Lett.
,
35
,
L16810
, https://doi.org/10.1029/2008GL034303.
Norris
,
J. R.
, and
A. T.
Evan
,
2015
:
Empirical removal of artifacts from the ISCCP and PATMOS-x satellite cloud records
.
J. Atmos. Oceanic Technol.
,
32
,
691
702
, https://doi.org/10.1175/JTECH-D-14-00058.1.
O’Neill
,
N. T.
,
2003
:
Spectral discrimination of coarse and fine mode optical depth
.
J. Geophys. Res.
,
108
,
4559
, https://doi.org/10.1029/2002JD002975.
Pandey
,
S. K.
,
V.
Vinoj
,
K.
Landu
, and
S. S.
Babu
,
2017
:
Declining pre-monsoon dust loading over South Asia: Signature of a changing regional climate
.
Sci. Rep.
,
7
,
16062
, https://doi.org/10.1038/s41598-017-16338-w.
Prospero
,
J. M.
,
1999
:
Long-range transport of mineral dust in the global atmosphere: Impact of African dust on the environment of the southeastern United States
.
Proc. Natl. Acad. Sci. USA
,
96
,
3396
3403
, https://doi.org/10.1073/pnas.96.7.3396.
Prospero
,
J. M.
,
2014
:
Characterizing the temporal and spatial variability of African dust over the Atlantic
.
Past Global Changes Mag.
,
22
,
68
69
, https://doi.org/10.22498/pages.22.2.68.
Prospero
,
J. M.
, and
R. T.
Nees
,
1986
:
Impact of the North African drought and El Niño on mineral dust in the Barbados trade winds
.
Nature
,
320
,
735
738
, https://doi.org/10.1038/320735a0.
Prospero
,
J. M.
, and
O. L.
Mayol-Bracero
,
2013
:
Understanding the transport and impact of African dust on the Caribbean Basin
.
Bull. Amer. Meteor. Soc.
,
94
,
1329
1337
, https://doi.org/10.1175/BAMS-D-12-00142.1.
Prospero
,
J. M.
, and Coauthors
,
1996
:
Atmospheric deposition of nutrients to the North Atlantic basin. Nitrogen Cycling in the North Atlantic Ocean and Its Watersheds, Springer, 27–73, https://doi.org/10.1007/978-94-009-1776-7_2
.
Pu
,
B.
, and
P.
Ginoux
,
2016
:
The impact of the Pacific decadal oscillation on springtime dust activity in Syria
.
Atmos. Chem. Phys.
,
16
,
13 431
13 448
, https://doi.org/10.5194/acp-16-13431-2016.
Pu
,
B.
, and
P.
Ginoux
,
2018a
:
Climatic factors contributing to long-term variations in surface fine dust concentration in the United States
.
Atmos. Chem. Phys.
,
18
,
4201
4215
, https://doi.org/10.5194/acp-18-4201-2018.
Pu
,
B.
, and
P.
Ginoux
,
2018b
:
How reliable are CMIP5 models in simulating dust optical depth?
Atmos. Chem. Phys.
,
18
,
12 491
12 510
, https://doi.org/10.5194/acp-18-12491-2018.
Qian
,
W.
,
L.
Quan
, and
S.
Shi
,
2002
:
Variations of the dust storm in China and its climatic control
.
J. Climate
,
15
,
1216
1229
, https://doi.org/10.1175/1520-0442(2002)015<1216:VOTDSI>2.0.CO;2.
Ralph
,
F. M.
, and Coauthors
,
2016
:
CalWater field studies designed to quantify the roles of atmospheric rivers and aerosols in modulating U.S. West coast precipitation in a changing climate
.
Bull. Amer. Meteor. Soc.
,
97
,
1209
1228
, https://doi.org/10.1175/BAMS-D-14-00043.1.
Remer
,
L. A.
, and Coauthors
,
2008
:
Global aerosol climatology from the MODIS satellite sensors
.
J. Geophys. Res.
,
113
,
D14S07
, https://doi.org/10.1029/2007JD009661.
Remer
,
L. A.
,
S.
Mattoo
,
R. C.
Levy
,
A.
Heidinger
,
R. B.
Pierce
, and
M.
Chin
,
2012
:
Retrieving aerosol in a cloudy environment: Aerosol product availability as a function of spatial resolution
.
Atmos. Meas. Tech.
,
5
,
1823
1840
, https://doi.org/10.5194/amt-5-1823-2012.
Ridley
,
D. A.
,
C. L.
Heald
,
J. F.
Kok
, and
C.
Zhao
,
2016
:
An observationally constrained estimate of global dust aerosol optical depth
.
Atmos. Chem. Phys.
,
16
,
15 097
15 117
, https://doi.org/10.5194/acp-16-15097-2016.
Rosenfeld
,
D.
,
Y.
Rudich
, and
R.
Lahav
,
2001
:
Desert dust suppressing precipitation: A possible desertification feedback loop
.
Proc. Natl. Acad. Sci. USA
,
98
,
5975
5980
, https://doi.org/10.1073/pnas.101122798.
Ryder
,
C. L.
, and Coauthors
,
2013
:
Optical properties of Saharan dust aerosol and contribution from the coarse mode as measured during the Fennec 2011 aircraft campaign
.
Atmos. Chem. Phys.
,
13
,
303
325
, https://doi.org/10.5194/acp-13-303-2013.
Ryder
,
C. L.
, and Coauthors
,
2018
:
Coarse mode mineral dust size distributions, composition and optical properties from AER-d aircraft measurements over the tropical eastern Atlantic
.
Atmos. Chem. Phys.
,
18
,
17 225
17 257
, https://doi.org/10.5194/acp-18-17225-2018.
Satheesh
,
S. K.
,
J.
Srinivasan
, and
K. K.
Moorthy
,
2006
:
Contribution of sea-salt to aerosol optical depth over the Arabian Sea derived from MODIS observations
.
Geophys. Res. Lett.
,
33
,
L03809
, https://doi.org/10.1029/2005GL024856.
Sato
,
M.
,
J. E.
Hansen
,
M. P.
McCormick
, and
J. B.
Pollack
,
1993
:
Stratospheric aerosol optical depths, 1850–1990
.
J. Geophys. Res.
,
98
,
22 987
22 994
, https://doi.org/10.1029/93JD02553.
Sayer
,
A. M.
,
N. C.
Hsu
,
C.
Bettenhausen
, and
M.-J.
Jeong
,
2013
:
Validation and uncertainty estimates for MODIS Collection 6 “Deep Blue” aerosol data
.
J. Geophys. Res.
,
118
,
7864
7872
, https://doi.org/10.1002/jgrd.50600.
Sayer
,
A. M.
,
N. C.
Hsu
,
J.
Lee
,
N.
Carletta
,
S.-H.
Chen
, and
A.
Smirnov
,
2017
:
Evaluation of NASA Deep Blue/SOAR aerosol retrieval algorithms applied to AVHRR measurements
.
J. Geophys. Res. Atmos.
,
122
,
9945
9967
, https://doi.org/10.1002/2017JD026934.
Schmeisser
,
L.
, and Coauthors
,
2017
:
Classifying aerosol type using in situ surface spectral aerosol optical properties
.
Atmos. Chem. Phys.
,
17
,
12 097
12 120
, https://doi.org/10.5194/acp-17-12097-2017.
Shao
,
Y.
,
2008
:
Physics and Modelling of Wind Erosion. Springer Science and Business Media, 456 pp
.
Smirnov
,
A.
,
B. N.
Holben
,
T. F.
Eck
,
O.
Dubovik
, and
I.
Slutsker
,
2003
:
Effect of wind speed on columnar aerosol optical properties at Midway Island
.
J. Geophys. Res.
,
108
,
4802
, https://doi.org/10.1029/2003JD003879.
Stith
,
J. L.
, and Coauthors
,
2009
:
An overview of aircraft observations from the Pacific Dust Experiment campaign
.
J. Geophys. Res.
,
114
,
D05207
, https://doi.org/10.1029/2008JD010924.
Strong
,
J. D. O.
,
G. A.
Vecchi
, and
P.
Ginoux
,
2018
:
The climatological effect of Saharan dust on global tropical cyclones in a fully coupled GCM
.
J. Geophys. Res.
,
123
,
5538
5559
, https://doi.org/10.1029/2017JD027808.
Sun
,
J.
,
M.
Zhang
, and
T.
Liu
,
2001
:
Spatial and temporal characteristics of dust storms in China and its surrounding regions, 1960–1999: Relations to source area and climate
.
J. Geophys. Res.
,
106
,
10 325
10 333
, https://doi.org/10.1029/2000JD900665.
Swap
,
R.
,
M.
Garstang
,
S.
Greco
,
R.
Talbot
, and
P.
Kallberg
,
1992
:
Saharan dust in the Amazon Basin
.
Tellus
,
44B
,
133
149
, https://doi.org/10.3402/tellusb.v44i2.15434.
Textor
,
C.
, and Coauthors
,
2006
:
Analysis and quantification of the diversities of aerosol life cycles within AeroCom
.
Atmos. Chem. Phys.
,
6
,
1777
1813
, https://doi.org/10.5194/acp-6-1777-2006.
Tong
,
D. Q.
,
J. X. L.
Wang
,
T. E.
Gill
,
H.
Lei
, and
B.
Wang
,
2017
:
Intensified dust storm activity and valley fever infection in the southwestern United States
.
Geophys. Res. Lett.
,
44
,
4304
4312
, https://doi.org/10.1002/2017GL073524.
Toth
,
T. D.
, and Coauthors
,
2013
:
Investigating enhanced Aqua MODIS aerosol optical depth retrievals over the mid-to-high latitude Southern Oceans through intercomparison with co-located CALIOP, MAN, and AERONET data sets
.
J. Geophys. Res. Atmos.
,
118
,
4700
4714
, https://doi.org/10.1002/jgrd.50311.
Toth
,
T. D.
,
J.
Zhang
,
J. R.
Campbell
,
E. J.
Hyer
,
J. S.
Reid
,
Y.
Shi
, and
D. L.
Westphal
,
2014
:
Impact of data quality and surface-to-column representativeness on the PM1/satellite AOD relationship for the contiguous United States
.
Atmos. Chem. Phys.
,
14
,
6049
6062
, https://doi.org/10.5194/acp-14-6049-2014.
Toth
,
T. D.
,
J.
Zhang
,
J. R.
Campbell
,
J. S.
Reid
, and
M. A.
Vaughan
,
2016
:
Temporal variability of aerosol optical thickness vertical distribution observed from CALIOP
.
J. Geophys. Res.
,
121
,
9117
9139
, https://doi.org/10.1002/2015JD024668.
Yumimoto
,
K.
,
K.
Eguchi
,
I.
Uno
,
T.
Takemura
,
Z.
Liu
,
A.
Shimizu
,
N.
Sugimoto
, and
K.
Strawbridge
,
2010
:
Summertime trans-Pacific transport of Asian dust
.
Geophys. Res. Lett.
,
37
,
L18815
, https://doi.org/10.1029/2010GL043995.
Zábori
,
J.
,
R.
Krejci
,
A. M. L.
Ekman
,
E. M.
Mårtensson
,
J.
Ström
,
G.
de Leeuw
, and
E. D.
Nilsson
,
2012
:
Wintertime Arctic ocean sea water properties and primary marine aerosol concentrations
.
Atmos. Chem. Phys.
,
12
,
10 405
10 421
, https://doi.org/10.5194/acp-12-10405-2012.
Zhao
,
T. X.-P.
,
L. L.
Stowe
,
A.
Smirnov
,
D.
Crosby
,
J.
Sapper
, and
C. R.
McClain
,
2002
:
Development of a global validation package for satellite oceanic aerosol optical thickness retrieval based on AERONET observations and its application to NOAA/NESDIS operational aerosol retrievals
.
J. Atmos. Sci.
,
59
,
294
312
, https://doi.org/10.1175/1520-0469(2002)059<0294:DOAGVP>2.0.CO;2.
Zhao
,
T. X.-P.
,
I.
Laszlo
,
W.
Guo
,
A.
Heidinger
,
C.
Cao
,
A.
Jelenak
,
D.
Tarpley
, and
J.
Sullivan
,
2008
:
Study of long-term trend in aerosol optical thickness observed from operational AVHRR satellite instrument
.
J. Geophys. Res.
,
113
,
D07201
, https://doi.org/10.1029/2007JD009061.
Zhao
,
T. X.-P.
,
P. K.
Chan
, and
A. K.
Heidinger
,
2013
:
A global survey of the effect of cloud contamination on the aerosol optical thickness and its long-term trend derived from operational AVHRR satellite observations
.
J. Geophys. Res.
,
118
,
2849
2857
, https://doi.org/10.1002/jgrd.50278.
Zhao
,
X.
, and
NOAA CDR Program
,
2017
:
NOAA Climate Data Record (CDR) of AVHRR daily and monthly Aerosol Optical Thickness (AOT) over global oceans, Version 3.0. NOAA/National Centers for Environmental Information, accessed 26 February 2019, https://doi.org/10.7289/V5BZ642P
.

Footnotes

Supplemental information related to this paper is available at the Journals Online website: https://doi.org/10.1175/JAMC-D-19-0194.s1.

© 2020 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Supplemental Material