Effect of spectral-dependent surface albedo on Saharan dust direct radiative forcing



[1] Direct radiative forcing (DRF) due to mineral dust has quite large uncertainty. Surface albedo is one of the most important factors affecting dust radiative forcing and climate. Here we investigate the effect of spectral-dependent surface albedo on dust DRF over the Saharan desert region using MODIS surface albedo data and a size-resolved global aerosol model. In one simulation, surface albedo in 7 wavebands from MODIS-retrieved data is interpolated to the corresponding 4 solar wavebands of the radiation transfer model (case 1). In another simulation, surface albedo for visible wavebands is applied to all 4 solar spectral-bands (case 2), which was previously used by many global model simulations. Our results show that the annual averaged DRF for all sky over the Saharan dust is −2.4 W m−2 and −5.6 W m−2 at TOA, and −9.9 W m−2 and −12.9 W m−2at surface for cases 1 and 2, respectively. Such a large difference highlights the importance of using accurate spectral-dependent surface albedo, and implies that previous studies employing only visible-band surface albedo might have significantly overestimated the dust cooling effect over the Saharan dust.

1. Introduction

[2] Mineral dust is one of the most abundant aerosol species in the atmosphere. It impacts the Earth's radiation balance by scattering and absorbing incoming solar radiation, and by absorbing and emitting terrestrial radiation. The Saharan desert is the largest source of mineral dust in the world, thus it can significantly influence the climate of North Africa. In addition, since dust particles can be lifted to high altitudes and transported over long distances from the source regions, Saharan dust can also play an important role in modifying climate on the global scale [Chin et al., 2007]. Due to the complexity in factors such as dust emission, size distribution, optical properties and surface conditions, the direct radiative forcing (DRF) of mineral dust and its climate effect remains to have large uncertainties and great efforts have been made in the past to improve the representation of mineral dust in the model [Ginoux et al., 2001; Zender et al., 2003a].

[3] Surface conditions, particularly surface albedo, significantly affect aerosol DRF. The dependence of DRF on surface albedo has been investigated in earlier studies using 1D column radiation transfer models [Liao and Seinfeld, 1998; Li et al., 2008]. The study of Liao and Seinfeld [1998]shows that dust DRF for clear-sky increases from negative (cooling) to positive (heating) as surface albedo increases, whereas TOA DRF for all-sky is positive for all surface albedo.Stier et al. [2007]examined the sensitivity of radiative forcing to surface albedo, and their results showed that the total aerosol DRF at TOA for clear-sky (all-sky) is −4.19 (−2.62) and −3.55 (−2.09) W m−2for AeroCom minimum surface albedo (global mean of 0.18) and maximum surface albedo (global mean of 0.36) respectively. In these previous modeling studies, surface albedo for solar broadband or visible albedo instead of spectral-dependent surface albedo is often used. Spectral-dependent surface albedo can now be derived from MODIS observation but the application of such data in global models to assess aerosol DRF is limited. It should be noted that, surface albedo is explicitly predicted by land surface models in some GCM models. However, there exist large uncertainties in the modeled surface albedo [Oleson et al., 2003; Stier et al., 2007]. Oleson et al. [2003]compared predicted land surface albedo with MODIS observation and found a large negative model bias for the Sahara Desert and Arabian Peninsula, particularly in the near-infrared.

[4] In this study we examine in detail the effect of spectral-dependent surface albedo on dust DRF over the Saharan desert, using MODIS surface albedo data and a global aerosol model. The paper is organized as follows: model description and extensions are presented inSection 2, the surface albedo used in the study and dust DRF over the Saharan desert and sensitivity to surface albedo is described and discussed in section 3. Section 4 is a summary.

2. Model Description and Extensions

[5] We employ the GEOS-Chem-APM model [Yu and Luo, 2009], in which a size-resolved aerosol microphysics model has been coupled with a global chemistry transport model (GEOS-Chem). The model has been extended by including a look-up table Mie code, to calculate aerosol optical properties [Yu et al., 2012], and a radiation transfer model to estimate radiation fluxes [Ma et al., 2012].

2.1. GEOS-Chem-APM Model

[6] The GEOS-Chem model is a global 3-D model of atmospheric composition driven by assimilated meteorological observations from the Goddard Earth Observing System (GEOS) of the NASA Global Modeling Assimilation Office (GMAO) [e.g.,Bey et al., 2001]. The APM model, incorporated into GEOS-Chem byYu and Luo [2009], is an advanced multi-type, multi-component, size-resolved microphysics model. The basic microphysical processes in the model include nucleation, condensation/evaporation, coagulation, thermodynamic equilibrium with local humidity, and dry and wet deposition. Currently, the APM model in GEOS-Chem employs 15 bins to represent dust particles covering a dry diameter size range from 0.02μm to 63 μm. Dust emission in GEOS-Chem is parameterized by a dust entrainment and deposition (DEAD) scheme developed byZender et al. [2003a, 2003b]. The scheme treats the vertical dust flux as proportional to the horizontal saltation flux, where the total horizontal saltation flux is based on the theory of White [1979]. The scheme permits regions that become seasonally devegetated to mobilize.

2.2. Mie Calculation of Aerosol Optical Properties

[7] The key particle optical properties used for aerosol direct radiative forcing (DRF) calculations include extinction efficiency, single scattering albedo, and asymmetry parameter. A computationally efficient scheme, in terms of lookup tables, has been designed and developed to calculate online the aerosol optical properties that take advantage of important particle information (sizes, compositions, etc) predicted by APM [Yu et al., 2012]. For dust particles, a wavelength dependence of the refractive index presented in Balkanski et al. [2007]is adapted to this study. The particles are assumed to be spherical and the non-spherical effect is not considered in this study. More detailed information about the calculation of aerosol optical properties in GEOS-Chem-APM can be found inYu et al. [2012].

2.3. Radiation Transfer Model

[8] The 1D radiation transfer model of the Canadian Center for Climate Modeling and Analysis (CCCma) has been adapted and incorporated into the GEOS-Chem-APM, to calculate shortwave fluxes over four solar bands (0.20–0.69, 0.69–1.19, 1.19–2.38, 2.38–4.00μm) and longwave fluxes over 9 infrared spectral intervals (4.0–4.5, 4.5–5.5, 5.5–6.5, 6.5–9.0, 9.0–10.0, 10.0–12.5, 12.5–18.5, 18.5–27.9, 27.9–40.0 μm). The model includes molecular Rayleigh scattering, gas absorption, cloud effects, and aerosol scattering and absorption. This radiation model is a correlated k-distribution scheme for gaseous transmission [Li and Barker, 2002]. For shortwave radiation, water vapor, CO2, O3, CH4 and O2 are considered for gaseous transmission. For longwave radiation, water vapor, CO2, O3, CH4, N2O, CFC11 and CFC12are considered for gaseous transmission. In the present study, the atmospheric vertical profiles of pressure, temperature and water vapor mixing ratios are provided by GEOS-Chem meteorological fields. O3concentration is taken from GEOS-Chem simulation. The solar zenith angle, cloud cover and cloud water content (both liquid and ice) are also taken from GEOS-Chem. The cloud optical properties: specific extinction, single scattering albedo and asymmetry parameter, are parameterized for each band as a function of cloud size and concentration. The effective radius for liquid cloud and ice cloud are taken from CERES satellite data [Ma et al., 2010]. With the cloud water contents and effective particle sizes, the cloud optical properties for each band are computed for liquid cloud particles at solar [Dobbie et al., 1999] and infrared [Lindner and Li, 2000] wavebands and for ice cloud particles at solar [Fu, 1996] and infrared [Fu et al., 1998] wavebands.

3. Saharan Dust Direct Radiative Forcing

3.1. Surface Albedo

[9] We used the surface albedo dataset from MODIS satellite retrievals (product MCD43C4). The data provides 8 day mean surface albedo values at 0.05° (5600 m) resolution, for MODIS spectral wavebands 1–7 (620–670 nm, 841–876 nm, 459–479 nm, 545–565 nm, 1230–1250 nm, 1628–1652 nm, 2105–2155 nm). The accuracy of MODIS retrieved surface albedo has been validated against ground measurements by previous studies [Liang et al., 2002; Jin et al., 2003]. These studies show that the MODIS albedo data are reasonably accurate, with typically less than 5% absolute error. Figure 1shows the area-averaged surface albedo over Saharan regions (10°–25° N, 10° W–30° E) at 7 wavebands from MODIS datasets. It is clearly shown that surface albedo is strongly dependent on waveband, with the lower surface albedo at visible bands (bands 1, 3 and 4) and higher at near-infrared bands. The spatial distribution of surface albedo at band 3, band 5, and the ratio of band 3 to band 5 are presented inFigure 2. It is seen that surface albedo at band 3 is generally lower than 0.3, while at band 5 it is higher than 0.5, so the ratio is around 0.2 to 0.3. This indicates that using visible surface albedo, which is commonly applied in many previous studies, may not be appropriate and may cause the bias in predicted radiative forcing and climate. The surface albedo in the different seasons (January, April, July and October) is also shown in Figure 1. No significant seasonal variability in the region is found since dust covers the surface in the whole year. In this study, the surface albedo in the MODIS bands 1–7 have been interpolated to the 4 bands corresponding to those used in the radiation transfer model and updated every 8 day.

Figure 1.

MODIS-retrieved surface albedo over the Saharan dust region at 7 wavebands in 2006.

Figure 2.

MODIS-retrieved annual mean surface albedo over the Saharan dust region at (a) band 3, (b) band 5, and (c) the ratio of band 3 to band 5.

3.2. Saharan Dust DRF

[10] The simulations start from November 2005, with the first two months as spin up. One year results in 2006 are used for analysis. The assimilated meteorological data for this year are used as input. As MODIS satellite data provides surface albedo at 7 wavebands covering from 459 nm to 2155 nm, a linear interpolation is firstly used to calculate surface albedo at wavelengths corresponding to those in the radaitive transfer model and then a weighted averaged, weighting by the solar spectrum at the top of atmosphere, is applied to obtain the surface albedo at 4 broader bands. Two simulations are conducted in the study. In case 1, surface albedo is taken from 4 solar wavebands. In case 2, surface albedo is obtained only from visible-bands and applied to all 4 wavebands in the radiation transfer model. The calculated dust aerosol burden and radiative forcing are presented in this section.

[11] Figure 3ashows the GEOS-Chem-APM simulated dust burden over the Saharan desert in January and July. It is clear that the model could successfully capture the seasonality of Saharan dust, i.e., relatively low burden in January and high burden in July. The calculated dust aerosol optical depth (AOD) in both seasons are also presented inFigure 3b and compared with satellite retrievals from MISR (Figure 3c) and MODIS (Figure 3d). MISR (Multi-angle Imaging SpectroRadiaometer) aerosol AOD [Martonchik et al., 1998] data used here is the monthly level-3 product with a resolution of 0.5° × 0.5°. Annual averaged AOD at 555 nm (green band) for the year 2006 is applied for comparison with the model simulation. MODIS (Moderate Resolution Imaging Spectroradiometer) AOD [Kaufman et al., 1997; Remer et al., 2005] data is taken from the monthly level-3 product from Aqua (MYD08_M3.051) with 1° × 1° degree resolution, and combined with the deep blue product, which is a separate product specifically retrieved for the AOD over desert regions. The annual averaged AOD at 550 nm is used for comparison. It should be noted that the AOD values from satellite retrievals are the total extinction caused by all types of aerosol particles, but the model simulated AOD shown in theFigure 3b is just for dust particles. However, the dominant aerosol particles in the region are from dust. Overall, the spatial distribution and seasonal change of the simulated AOD are quite consistent with the satellite retrievals.

Figure 3.

(a) GEOS-Chem-APM simulated dust burden in mg m−2, (b) dust optical depth AOD at 550 nm, (c) AOD from MISR at 555 nm, and (d) AOD from MODIS at 550 nm in January and July. AOD values from satellite retrievals are the total extinction caused by all types of aerosol particles, while the modeled AOD shown in Figure 3b is just for dust particles.

[12] The simulated annual mean DRF at TOA is presented in Figure 4a. It is shown that the results are quite different if the spectral-dependent surface albedo is taken into account (case 1, top plot), i.e., positive DRF is found over some of the area. In contrast, DRF is always negative in the case with only visible-band surface albedo (case 2, middle plot). The difference between two simulations (case 1 minus case 2) is significant (bottom plot), with the magnitude over 1 W m−2 for most of the area and up to 4 W m−2 for the center of the Saharan desert. These results will be further discussed as below.

Figure 4.

The results over Saharan with the (top) spectral-dependent surface albedo, (middle) visible-band surface albedo applied, and (bottom) their difference for (a) dust SW DRF at TOA for all sky, (b) clear-sky DRF at surface, and (c) atmospheric absorption for all sky. Unit: W m−2.

[13] Figure 4b shows the clear sky DRF at surface for the case 1 (top plot), case 2 (middle plot), and their difference (bottom plot). These results present that the net shortwave (SW) radiation at surface due to dust particles is less negative in case 1 than in case 2. This is because the more incoming solar radiation is reflected by the higher surface albedo (see Figure 1), and thus less incoming radiation available for dust particles. Therefore the reflected radiation by dust particles decrease and DRF at surface becomes smaller. The magnitude and spatial distribution of difference in DRF at TOA (Figure 4a, bottom) is almost the same as that of the difference in DRF at surface for clear-sky (Figure 4b, bottom), indicating that the DRF at TOA is mostly determined by the reflected radiation at surface, which is controlled by surface albedo.

[14] Mineral dust particles not only scatter solar radiation, but also absorb solar radiation. The imaginary part of refractive index in the study ranges from 0.005 to 0.03 depending on wavebands [Balkanski et al., 2007]. We can see from Figure 4c that the annual averaged atmospheric absorption due to dust is generally higher than 2 W m−2, and up to 8 W m−2in the majority of the region. The atmospheric absorption in the case 1 (top plot) is a bit higher than in the case 2 (middle plot) since high-albedo surfaces lead to enhanced absorption of reflected radiation, so the difference (bottom plot) is slightly positive. The annual averaged vertical profiles of relative solar heating rates due to dust also show (figure omitted) that the solar heating rate with the spectral-dependent surface albedo is slightly higher than that with the visible-wavelength surface albedo.

[15] In this study, surface albedo from MODIS satellite retrieval is for daily averaged values of solar zenith angle. Earlier study [Wang et al., 2005] shows that desert surface albedo is dependent on solar zenith angle. We used a simple one-parameter formulation [Wang et al., 2005] to consider the dependence of albedo on solar zenith angle, and the difference of all-sky DRF at TOA between the test simulation and the original one is within 10%.

[16] It should be noted that there still exist large uncertainties in refractive index of dust particles, specifically imaginary part, which significantly impact dust optical properties and thus AOD and DRF. In this study the dust refractive index of Balkanski et al. [2007] is used, which is based on the AERONET observations. Based on the simulated size distribution the calculated single scattering albedo (SSA) is about 0.93–0.96, which is close to the upper limit of the SSA values reported in previous studies (0.85–0.97). To investigate the dependence of forcing on SSA, we conducted the sensitivity experiments in which SSA for dust is set to be 0.85 in both case simulations. The results indicate that the forcing will become positive due to stronger absorption for both cases, but still with the more positive forcing in the case 1 than case 2, which is consistent with our conclusion.

4. Summary

[17] In this study, two sets of surface albedo data are used in the GEOS-Chem-APM model, to examine the DRF due to dust particles, particularly the effect of surface albedo on DRF overt the Saharan desert. One simulation uses a spectral-dependent surface albedo, while the other uses the visible-band surface albedo, which has also been used in most previous studies. Our study shows that the surface albedo plays a significant role in modulating DRF over the Sarahan dust. If only the visible-band surface albedo is taken into account, the DRF at TOA is always negative. However, a spectral-dependent surface albedo used in the study gives less negative DRF values, or even positive in part of the region. This is because the surface reflects more incoming solar radiation due to the high surface albedo for the spectral-dependent case, so the dust particles scatter less. It can be concluded that previous studies which used a visible-band surface albedo overestimated the reflected solar radiation and DRF at TOA over the Saharan region. To predict the dust DRF more accurately, it is necessary to take a spectral-dependent surface albedo into account.


[18] This study is supported by NSF under grant AGS-0942106 and DOE under grant DE-SC0002199. The data for AOD from MODIS and MISR were downloaded using the GES-DISC Interactive Online Visualization and Analysis Infrastructure, a part of the NASA's Goddard Earth Sciences Data and Information Services Center and AERONET data were obtained from NASA Goddard Space Flight Center. The MODIS L1B data were obtained through the online Data Pool at the NASA Land Processes Distributed Active Archive Center (LP DAAC), USGS/Earth Resources Observation and Science (EROS) Center, Sioux Falls, South Dakota (http://lpdaac.usgs.gov/get_data). The GEOS-Chem model is managed by the Atmospheric Chemistry Modeling Group at Harvard University with support from NASA's Atmospheric Chemistry Modeling and Analysis Program.

[19] The Editor wishes to thank two anonymous reviewers for assisting with the evaluation of this paper.