Improved estimates and understanding of global albedo and atmospheric solar absorption


  • Dohyeong Kim,

    Corresponding author
    1. National Meteorological Satellite Center, Korea Meteorological Administration, Gwanghyewon, South Korea
    • Corresponding author: D. Kim, National Meteorological Satellite Center, Korea Meteorological Administration, 636-10, Gwanghyewon, Jincheon, Chungbuk 365-831, South Korea. (

    Search for more papers by this author
  • V. Ramanathan

    1. Center for Clouds, Chemistry and Climate, Scripps Institution of Oceanography, La Jolla, California, USA
    Search for more papers by this author


[1] This study integrates available surface-based and satellite observations of solar radiation at the surface and the top of the atmosphere (TOA) with a comprehensive set of satellite observations of atmospheric and surface optical properties and a Monte Carlo Aerosol-Cloud-Radiation (MACR) model to estimate the three fundamental components of the planetary solar radiation budget: Albedo at the TOA; atmospheric solar absorption; and surface solar absorption. The MACR incorporates most if not all of our current understanding of the theory of solar radiation physics including modern spectroscopic water vapor data, minor trace gases, absorbing aerosols including its effects inside cloud drops, 3-D cloud scattering effects. The model is subject to a severe test by comparing the simulated solar radiation budget with data from 34 globally distributed state-of-the art BSRN (Baseline Surface Radiation Network) land stations which began data collection in the mid 1990s. The TOA over these sites were obtained from the CERES (Cloud and Earth's Radiant Energy System) satellites. The simulated radiation budget was within 2 Wm−2for all three components over the BSRN sites. On the other hand, over these same sites, the IPCC-2007 simulation of atmospheric absorption is smaller by 7–8 Wm−2. MACR was then used with a comprehensive set of model input from satellites to simulate global solar radiation budget. The simulated planetary albedo of 29.0% confirms the value (28.6%) observed by CERES. We estimate the atmospheric absorption to be 82 ± 8 Wm−2 to be compared with the 67 Wm−2 by IPCC models as of 2001 and updated to 76 Wm−2by IPCC-2007. The primary reasons for the 6 Wm−2 larger solar absorption in our estimates are: updated water vapor spectroscopic database (∼1 Wm−2), inclusion of minor gases (∼0.5 Wm−2), black and brown carbon aerosols (∼4 Wm−2), the inclusion of black carbon in clouds (∼1 Wm−2) and 3-D effect of clouds (∼1 Wm−2). The fundamental deduction from our study is the remarkable consistency between satellite measurements of the radiation budget and the parameters (aerosols, clouds and surface reflectivity) which determine the radiation budget. Because of this consistency we can account for and explain the global solar radiation budget of the planet within few Wm−2.

1. Introduction

[2] Solar radiation budget plays a fundamental role in the global climate system. There have been numerous efforts to understand the global solar radiation budget by using satellite based measurements as well as models which include both GCMs and radiative transfer models [e.g., Kiehl and Trenberth, 1997; Li et al., 1997; Ohmura and Gilgen, 1993; Wild, 2005; Wild et al., 2006]. In spite of efforts, there are still large discrepancies between various models [e.g., Halthore et al., 2005; Wild, 2005; Wild et al., 2006] as well as between models and observations [e.g., Li et al., 1997; Ramanathan and Vogelmann, 1997]. In particular, invariably the simulated atmospheric solar absorption is smaller than observations by around 10 Wm−2.

[3] The main objective of this study is to provide state-of-the estimates which are based on radiative transfer model calculations for planetary albedo and atmospheric solar absorption and surface solar absorption. The approach of this study is to use the best available theory of solar absorption in cloudy atmospheres and the most comprehensive global data sets of model input parameters (aerosols, clouds, surface properties, and others) to simulate global solar radiation budget. The present study uses Monte Carlo Aerosol-Cloud-Radiation (MACR) model which relies on satellite and ground-based measurements for the model input parameters such as clouds, aerosols, water vapor, surface albedo and ozone. The details of the model as well the source of input parameters are explained in detail by us earlier [Kim and Ramanathan, 2008, hereinafter KR08]. This study is basically an update of the results published in KR08. This update was necessitated by the following improvements in the treatment of radiation physics in MACR: (1) The inclusion of minor gaseous absorption, such as NO2, N2O and CH4 in the MACR model. (2) Consideration of the effect of black carbon (BC) and cloud interaction within and above low clouds in the MACR model. To determine the amount of solar radiation absorption by the clouds containing BC, the number concentration of BC particles is estimated from the fraction of aerosol absorption which is dependent on Aerosol Optical Depth (AOD) from MISR (Multiangle Imaging Spectroradiometer) and Single Scattering Albedo (SSA) from GOCART (Georgia Tech/Goddard Global Ozone Chemistry Aerosol Radiation and Transport) discussed in Session 2 (cf. auxiliary material, Text S1, section 1). In addition to the baseline profile of aerosol, constant from surface to 3 km and then exponentially decrease above the height maintaining about 15% of aerosols between 3 km and 8 km, several different vertical profiles of aerosol are tested to consider the interaction between BC and clouds within and above the low clouds. (3) Consideration of 3-D effect in global solar radiation absorption using the advantage of the Monte Carlo radiative transfer model whose capability for investigating the effects of 3-D clouds distributions on global solar radiation calculation, and those effects are also discussed inSection 3.

[4] In order to assess the validity of the model and the satellite input, the simulated fluxes are assessed at two global networks of surface radiation measurements: BSRN (Baseline Surface Radiation Network) and GEBA (global energy balance archive) sites. Our primary emphasis in assessing the performance of the model is on the BSRN data since it has well established protocols for data quality. However, we include GEBA data sets for completeness.

2. Model and Data

2.1. Model Description and Gaseous Absorption

[5] The MACR, developed at the Center for Clouds, Chemistry, and Climate (C4), Scripps Institution of Oceanography, uses 32 bands to cover the solar spectrum from 0.25 to 5.0 μm with 50 layers [Podgorny and Ramanathan, 2001; Vogelmann et al., 2001]. MACR adopted the correlated k-distributions from Kato's database based on HITRAN 2000 database to incorporate gaseous absorption by water vapor (including continuum absorption), ozone, oxygen, and carbon dioxide [Kato et al., 1999] (cf. Text S1, section 1.1). In addition to the gases included in KR08, the present study includes absorption by NO2, N2O, and CH4 [e.g., Solomon et al., 1999]. The contributions of atmospheric absorption by these additional gases are discussed in Session 3. To consider the spatial variations of NO2absorption, recently released satellite-based columnar NO2measurement data (Aura OMI NO2 Level 2G Global Data Product-OMNO2G) are used [Boersma et al., 2007]. Further information on the MACR model details and datasets is available in the auxiliary material.

2.2. The Effect of Black Carbon on Cloud SSA

[6] It is now well established that BC by internally mixing with organic and sulfate aerosols nucleate cloud drops [e.g., see Chen et al., 2010]. In addition, BC is removed by precipitation wither as nuclei or by scavenging [Hadley et al., 2010]. Thus it is important to include the effects of BC inside a cloud drop or ice crystal since BC inside a cloud drop can reduce cloud drop SSA and therefore increase the atmospheric absorption by clouds [Chýlek et al., 1996]. We focus on evaluating the importance in atmospheric absorption by BC in low water clouds by calculating atmospheric absorption with and without BC (cf. Text S1, section 1.2).

2.3. The Effect of 3-D Clouds

[7] One of the major advantages of Monte Carlo radiative transfer model is its ability to treat the 3-D radiative transfer within clouds [e.g.,Podgorny, 2003]. Recently we developed the 3-D treatment of clouds, which generates random distribution of clouds in 3-D space by given cloud fraction and cloud size probability as model inputs, and has been successfully used to reproduce the observed 3-D radiation fields [Kim et al., 2007; Ramanathan et al., 2007]. The MACR model was carried out to examine the effects of 3-D clouds distribution on atmospheric absorption, which is obtained by the flux difference between 3-D and IPA (Independent Pixel Approximation) method over the globe. However, due to the constraint of flat shape cloud assumption and small domain size (∼50 km), only limited 3-D cloud effect can be examined in this study.

2.4. Data for Assessment

[8] Observational data from BSRN and GEBA form the basis for assessment of model simulations at the surface. The radiative fluxes at 35 BSRN sites are measured with well-calibrated instruments of high accuracy producing 1-minute averaged data covering the period from 1992 to 2002 [Ohmura et al., 1998]. The GEBA database, created from measurements taken at 1500 surface stations, contains monthly mean shortwave irradiances since the 1950s [Gilgen et al., 1998]. The quality of the GEBA data has been rigorously controlled since the GEBA database was redesigned and updated in 1994 and 1995. The estimated relative random errors of solar radiation values in the GEBA are 5% for the monthly means and 2% for annual means [Gilgen et al., 1998].

[9] The satellite climatologies of the radiative fluxes at TOA above the GEBA sites are obtained from the Earth Radiation Budget Experiment (ERBE) [Barkstrom et al., 1989; Rieland and Raschke, 1991] covering the period 1985–1989 (2.5° by 2.5° resolutions) as well as the Cloud and Earth's Radiant Energy System (CERES) [Wielicki et al., 1996] to consider the TOA flux for period 2000 to 2002. The uncertainties in the monthly averaged ERBE and CERES data are estimated at ±5 Wm−2 and ±2 Wm−2, respectively.

[10] Reference is also made to a recent dataset of 10 models (cf. auxiliary material, Figure S1) which participated in the experiments carried out for the 4th assessment report of the Intergovernmental Panel on Climate Change (IPCC-2007).

3. Global Mean Radiation Budget

[11] Figure 1shows the global annual mean radiation budget comparisons. First we compare the radiation budget between IPCC-2007 [Intergovernmental Panel on Climate Change (IPCC), 2007] and IPCC-2001 [IPCC, 2001]. IPCC-2007 shows increased atmospheric absorption by 9 Wm−2, and decreased surface absorption by 6 Wm−2compared with IPCC-2001. The updated water vapor spectroscopic database and inclusion of absorbing aerosols have contributed to the overall increased atmospheric absorption in IPCC-2007 models [Barker et al., 2003; Kinne et al., 1998; Wild, 2005; Wild et al., 1998]. Nevertheless, Standard Deviation (SD) among the models participating in the IPCC-2007 shows larger difference in the simulation of surface absorption (∼7.0 Wm−2) than that in TOA (∼3.9 Wm−2) or the atmosphere (∼4.8 Wm−2). Due to the GCMs' tendency toward underestimation of atmospheric absorption as shown in Section 3, Wild [2005] related the mean excessive surface insolation of GCMs (∼8 Wm−2) from GEBA to GCMs' deficiency in simulating atmospheric absorption. In spite of the assumption that the global mean bias is uniform and same as the regional mean bias over GEBA sites, Wild [2005] proposed that atmospheric solar absorption should be around 81 Wm−2 (average of 20 GCMs is 74 Wm−2) in the atmosphere, and 154 Wm−2 (average of 20 GCMs is 162 Wm−2) at the surface, respectively.

Figure 1.

Global annual mean solar radiation budget comparisons: IPCC-2001 [Kiehl and Trenberth, 1997], IPCC-2007 [IPCC, 2007], Wild et al. [2005] for 20 GCMs average incorporating the bias from GEBA, MACR (2000∼2002), KR08 (2000–2002), and CERES (average from 2000 to 2002) at TOA. KR08 represents the results from Kim and Ramanathan [2008].

[12] MACR (243 ± 5 Wm−2) estimates annual mean absorbed solar flux at TOA within 2 Wm−2 error compared with CERES (2000–2002) of 244 ± 2 Wm−2 while around 4 Wm−2 error from ERBE (1985–1989) of 240 ± 4 Wm−2. However, the prescription of solar insolation at the TOA (1361 to 1367 depending on the data source) can bias the value by about 1 to 1.5 Wm−2and so we rely less on the TOA flux; but depend more on the albedo which is a normalized quantity. The newly retrieved planetary albedo of 28.6 ± 0.6% by CERES is consistent with MACR model (∼29.0 ± 0.6%) and surface data while IPCC-2001 shows planetary albedo around 30.8%. The overestimation of planetary albedo in model simulations can cause additional uncertainties besides inadequate energy distributions between the surface and the atmosphere.

[13] The MACR-derived global atmospheric absorption is 82 ± 5 Wm−2 and the corresponding surface radiation is 161 ± 6 Wm−2. The atmospheric solar absorption is larger than previous values (IPCC-2001) by 15 Wm−2. When it compares with the SW radiation budget from KR08, global atmospheric absorption is increased by 3 Wm−2 (79 to 82 Wm−2) and global surface absorption is decreased by 3 Wm−2 (164 to 161 Wm−2). The uncertainties of TOA and surface solar flux retrieved from MACR are estimated by taking the difference between maximum and minimum values (rounded off to nearest higher integer value) as our 2-sigma uncertainty value. This procedure yields 5 Wm−2 for TOA flux, 8 Wm−2 for atmospheric absorption, and 6 Wm−2 for surface fluxes, respectively. Subsequently IPCC [2007] models rectified several deficiencies in climate model treatment of solar absorption and revised the atmospheric absorption upwards to 76 Wm−2, still 6 Wm−2 smaller compared with our value of 82 Wm−2. To investigate the source of this difference i.e., the excess absorption, we undertook several sensitivity studies, and identified the role of the new physics in atmospheric solar absorption: 1) Updated water vapor spectroscopic database and minor gases: As shown in Table 1, the atmospheric absorption by updated spectroscopic database and inclusion of minor gases (i.e., NO2, N2O, and CH4) increase absorption by 1.5 Wm−2 over the globe and by more than 2.0 Wm−2 over the land areas (see Figure S1 in the auxiliary material). The atmospheric forcing of 0.5 Wm−2 by inclusion of minor gases in Table 1 is comparable to the effect of 0.4 Wm−2 due to greenhouse gases without CO2 [Ramanathan and Carmichael, 2008]. 2) Aerosols: Table 1 shows that the largest contribution to atmospheric absorption is caused by absorbing aerosols (about 4 Wm−2). Among the total contribution of aerosols to atmospheric absorption, 15% can be attributed to the absorbing aerosols in mid- to upper-troposphere. 3) Cloud impurity: The effect of BC inclusion in clouds is investigated as described inSection 2.2. The global atmospheric absorption increases by 0.4–1.0 Wm−2 due to the inclusion of BC in cloud drops. This result is consistent with that of Chýlek et al. [1996] who suggested the upper boundary of enhanced absorption of solar radiation due to BC is between 1 and 3 Wm−2over the absorption of pure water clouds. 4) 3-D clouds distribution and clouds morphology: The global annual mean increased absorption due to the 3-D cloud effect (difference between 3-D and IPA calculation with flat shape clouds) as described inSection 2.3 is about 1.0 Wm−2. Marshak et al. [1998]suggested that the 3-D effect increased atmospheric absorption by 3–4 Wm−2 for bumpy morphology, while ∼1 Wm−2 for flat shape cloud. Similarly, Fu et al. [2000]showed that 3-D effect is less than 4 Wm−2 in a larger horizontal domain (e.g., 512 km), while larger than 20 Wm−2in a smaller domain (e.g., 75 km) for the tropical convective cloud system. They also showed that 3-D effect is less than 1 Wm−2for other cloud system regardless of domain size. Since 3-D clouds effects was investigated by using flat shape clouds without considering tropical convective clouds, the effect of ∼1 Wm−2 in this study should be considered as a lower boundary.

Table 1. The Contribution of Atmospheric Absorption by Gases, Aerosols and Clouds in MACR Calculationsa
  • a

    Updated water vapor spectroscopic database (HITRAN 96 to HITRAN 2000) and minor gases are considered. Atmospheric absorptions by aerosols in lower boundary as well as mid and upper troposphere (3–8 km) are considered. The effect of BC in clouds and 3-D effect are presented. BC and OC represent Black Carbon and Organic Carbon, respectively.

   Updated WV0.921.631.52
   Minor Gases (NO2, N2O, CH4)0.500.640.64
   Total (BC+OC+dust) aerosols4.255.594.08
   Mid- to upper- tropospheric aerosols0.651.090.69
   BC in clouds0.39∼1.610.38∼1.000.26∼1.30
   3-D effect1.141.121.25
Total (Wm−2)7.2∼7.89.4∼9.97.8∼8.2

[14] Moreover, the assessment of MACR over BSRN and GEBA sites for all-sky cases showed that the difference of atmospheric absorption between model simulation and observation was less than 3 Wm−2. This indicates that uncertainties between model and observation in atmospheric solar absorption have been significantly reduced by our study. The discussion on the flux assessment with GEBA and BSRN is available in the auxiliary material (cf. Text S1, section 2).

4. Summary

[15] MACR employs a comprehensive set of surface based and satellite borne instrumental data to estimate the global and regional solar radiation budget at the surface and TOA. In order to assess the uncertainties in the simulated radiation budget, MACR simulations were compared with observations at BSRN and GEBA sites. Over both BSRN and GEBA sites MACR simulates annual mean atmospheric absorption within 3 Wm−2under all-sky, while the biases in the IPCC-2007 simulations of atmospheric absorption are larger than 7–8 Wm−2.

[16] The planetary albedo of 29.0 ± 0.6% estimated by us supports the CERES value of around 28.6 ± 0.6%. The calculations are consistent with the observations within the combined uncertainties of the calculations and observations. We provide an improved and self-consistent estimate of 82 ± 5 Wm−2 for atmospheric solar absorption. It is larger by 15 Wm−2compared with classical and text-book estimates of 67 Wm−2(e.g., see IPCC-2001). There are many factors that contributed to the 15 Wm−2difference. Roughly half of it is due to deficiencies in the treatment of solar physics by earlier generation of climate models. More modern models (e.g., IPCC-2007) have revised it upwards to 76 Wm−2. The difference of 6 Wm−2 between these newer models and our estimate of 82 Wm−2 is due to several improved physics in our model: First of all, the updated water vapor spectroscopic database (HITRAN 96 to HITRAN 2000) and inclusion of minor gases (i.e., NO2, N2O, and CH4) increases global atmospheric absorption by 1.5 Wm−2. The largest contribution of increased atmospheric absorption (>4 Wm−2) is due to absorbing aerosols. About 15% of this increase is due to absorbing aerosols in the mid- to upper-troposphere. In addition, the BC in cloud drops increases the absorption by the range of 0.4–1.0 Wm−2due to the change of the single scattering co-albedo and asymmetry factor of the cloud drops. The increase volume fraction of BC to cloud drop can increase the absorbing characteristics of cloud drop embedded BC, but the increased value of atmospheric absorption is not larger than 1 Wm−2. 3-D cloud effect is another contribution to atmospheric absorption and its amount is around 1 Wm−2in this study whose value can be considered as a lower boundary since 3-D clouds effects were investigated by using flat shape clouds without considering tropical convective clouds.

[17] It is noted that among the increase in atmospheric absorption by 3 Wm−2 compared with KR08, the increased value in atmospheric absorption over ocean is larger than that over land by around 0.8 Wm−2, even though the major sources of BC emission are located in East Asia and developing nations in the tropics over land. This suggests that globally distributed BC by transportation from source region can significantly affect the global radiation budget as Ramanathan and Carmichael [2008]pointed out the contribution of BC in regional and global climate such as retreat of the Artic sea ice, ice-core records of Greenland, perturbation of monsoon.


[18] This study is supported by the National Meteorological Satellite Center (Project No. 153-3100-3137-302-210-13) and by the U.S. National Science Foundation (Grant AGS1016496).

[19] The Editor thanks the two anonymous reviewers for their assistance in evaluating this paper.