Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
 Atmospheric aerosols represent one of the most important components that attenuate solar radiation reaching the Earth's surface. The aerosol radiative forcing (ARF) at the surface is usually determined in the visible range of the solar spectrum. In contrast, there are few experimental works in the literature about the ARF in the ultraviolet (UV) region. Therefore, this paper focuses on quantifying the aerosol forcing efficiency in the UV erythemal range (AFEery), ARF per unit of aerosol optical depth (AOD). Simultaneous UV erythemal irradiance (UVER) and AOD measurements recorded between January 2006 and December 2008 in Granada (Spain) were used. In addition, an empirical model is utilized to estimate the UVER values for an atmosphere with very low aerosol loads (clean conditions). The AFEery varies from −62 to −26 mW/m2 per unit of AOD at 380 nm when the solar zenith angle (SZA) changes from 20° to 55°, showing a strong influence of the SZA on AFEery. The variations of the aerosol size and absorption properties also cause significant changes of this variable. Thus, 1 μm aerosols (related to desert dust particles) produce significantly higher AFEery (in absolute values) than submicrometer particles (associated with urban or industrial aerosols). For instance, AFEery varies from −52 mW/m2 per unit of AOD for Angström exponents smaller than 0.5 to −29 mW/m2 per unit of AOD for Angström exponents higher than 1.5. In addition, the AFEery values are −59 mW/m2 per AOD unit for single-scattering albedos (SSAs) smaller than 0.85 and −28 mW/m2 per AOD unit for SSAs larger than 0.85, showing that stronger aerosol absorption (low SSA) leads to a larger surface forcing efficiency (in absolute values). All these results highlight the outstanding role that atmospheric aerosol plays in the modifying levels of UV radiation reaching the surface.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 The analysis of the ultraviolet (UV) solar radiation at the Earth's surface (290–400 nm) has received considerable attention in recent years. A detailed knowledge of UV radiation is important because it affects both air temperature variability in the upper atmospheric layers and the origin and variability of stratospheric ozone [Iqbal, 1983]. In addition, UV radiation has a large influence on many biological, ecological, and photochemical processes, often being quite harmful to living organisms [Diffey, 1991, 2004]. Therefore, it is essential to further our understanding of factors modifying the UV radiation levels at the surface. Under cloud-free conditions, ozone and atmospheric aerosol are the main attenuating components for the UV radiation reaching the surface [United Nations Environment Programme, 2006]. Ozone column influences have been widely studied but conversely the effect of atmospheric aerosols is not clear yet, mainly because of their variability and difficulties associated with their characterization.
 The aim of this paper is to evaluate the surface AFE in the erythemal action range (AFEery), using simultaneous measurements of UV erythemal irradiance (UVER) and AOD data recorded between January 2006 and December 2008 in Granada (Spain). The dependency of the surface AFEery on the solar zenith angle (SZA) as well as the size and absorption properties of the aerosol particles are investigated. The erythemal action spectrum (280–400 nm) accounts for the effect of UV radiation on human skin, and it was proposed in its final form by McKinlay and Diffey  and later adopted as a standard by the Commission Internationale de l'Éclairage (CIE). Only a few studies on the analysis of AFE in the erythemal range have been published to date [e.g., di Sarra et al., 2002]. Therefore, this work is expected to improve the understanding of the influence of the atmospheric aerosol on UV radiation.
2. Instruments and Data
 Ground-based data were measured at the radiometric station located on the rooftop of the Andalusian Center for Environmental Studies (CEAMA, 37.17°N, 3.61°W, 680 m above sea level (asl)) in Granada, southeastern Spain. The station is operated by the Atmospheric Physics Group (GFAT) of Granada University
 Granada is a nonindustrialized medium-sized city with a population of 300,000 inhabitants that increases up to 600,000 when the metropolitan area is included. The city is located in a natural basin surrounded by mountains with elevations between 1000 and 3500 m asl. Near-continental conditions prevailing at this site are responsible for large seasonal temperature differences, providing cool winters and hot summers. The diurnal thermal oscillation is quite high throughout the year, often reaching up to 20°C.
 The ground-based station is equipped with a broadband UV radiometer, model UVB-1, manufactured by Yankee Environmental Systems, Inc. (Massachusetts), for measuring the UVER and a CM-11 pyranometer manufactured by Kipp & Zonen (Delft, Netherlands) for measuring the total solar irradiance. In order to guarantee the simultaneity of UVER and total solar irradiance data, both variables were recorded simultaneously (every minute) by the same data logger (CR10-X model, manufactured by Campbell Scientific, Inc). The CM-11 pyranometer complies with the specifications of the first-class World Meteorological Organization (WMO) classification of this instrument (resolution better than ±5 W/m2). This instrument was calibrated periodically with a reference CM-11 pyranometer. Output voltages provided by the UV radiometer were converted to UVER values by applying conversion factors obtained from the “two-step” calibration method, using the information derived from the first Spanish calibration campaign of broadband UV radiometers that took place at the “El Arenosillo” INTA station in Huelva (Spain) during September 2007 [Vilaplana et al., 2009]. The two-step method involves two stages: Initially, the output signal of the broadband UV radiometer is compared with the effective irradiance from the reference spectroradiometer; then, the effective response values are converted to erythemal units. A complete description of the “two-step” method is given by Webb et al. , Hülsen and Gröbner , and Antón et al. [2011a]. Particularly, Antón et al. [2011b] compared data provided by the UVB-1 radiometer installed in Granada using this calibration method with those estimated by a multilayer transfer model. These authors reported a mean absolute bias error (MABE) of (5.1 ± 0.5)%, highlighting the reliability of their UVER data.
 A CIMEL CE-318 Sun photometer, included in the NASA AERONET network [Holben et al., 1998] and located near both the CM-11 and UVB-1 radiometer, provided AOD data at 340, 380, 440, 500, 670, 870 and 1020 nm. In this work, the AERONET AOD data of level 2 (cloud screened and quality assured) were used [Dubovik and King, 2000]. These measurements have an estimated accuracy between 0.01 and 0.02 [Holben et al., 1998] except the AOD at 340 nm which present a larger error [Díaz et al., 2007]. Thus, since this work focuses on the UV region, the AOD at 380 and 440 nm were used. Furthermore the spectral dependency of the AOD has been considered through the Angström exponent evaluated in the range 440–870 nm.
 Cloud cover was characterized by the clearness index (kt) obtained from the ratio of the total solar irradiance to the extraterrestrial total solar irradiance on a horizontal surface [Alados-Arboledas et al., 2000]. In addition, an All-Sky Imager developed by the GFAT team was used to obtain the cloud cover in oktas (eighths of sky) [Cazorla et al., 2008a]. This instrument provides images of the whole sky dome in daytime, and it has been used for cloud cover characterization and retrieval of information on the atmospheric aerosol load [Cazorla et al., 2008b, 2009]. The All-Sky Imager is a custom adaptation of a scientific CCD camera, using a digital color video camera mounted with a fish-eye lens (180°FOV) pointing at the zenith. All data used in this work correspond to cloud-free conditions (kt higher than 0.65 and cloud cover smaller than one okta).
 Surface aerosol radiative forcing is defined as the instantaneous increase or decrease of the net radiation flux at the surface that is due to an instantaneous change of aerosol atmospheric content, in which the net flux is the difference between the downwelling (Fd) and the upwelling surface flux (Fu). Therefore, to determine the ARF it is necessary to have a reference case. In our case we have selected clear aerosol conditions (very low aerosol load) as the reference. Thus, ARF can be expressed as
where the superscript “0” denotes fluxes in clear aerosol conditions.
 The surface albedo (α) relates the downwelling and upwelling fluxes as
Consequently, the ARF in the UV erythemal range is evaluated as
where UVER represents the erythemal data recorded under cloud-free conditions in Granada and UVER0 corresponds to the erythemal data for the same SZA and atmospheric conditions but for clear aerosol conditions. The UV surface albedo (αUV) in Granada was derived from the UV albedo climatology over Europe, which was developed within the context of the European Union's action COST-726, “Long term changes and climatology of UV radiation over Europe” [Litynska et al., 2010]. The empirical expression proposed by Madronich  for cloud-free and unpolluted conditions was used for calculating UVER0:
where μ0 is the cosine of the SZA and TOC is the total ozone column in Dobson units (DUs). Other empirical expressions [e.g., Foyo-Moreno et al., 2007; Antón et al., 2009a] and artificial neural network models [e.g., Alados et al., 2004; 2007] to estimate UVER data can be found in the literature. Nevertheless, we have decided to use the Madronich expression because of its simplicity and excellent results [Antón et al., 2011c].
 Coefficients a, b, and c in equation (5) were obtained from regression analysis using data from January 2006 to December 2007. The UVER and μ0 data were hourly average values, while TOC corresponded to daily estimates from the Ozone Measuring Instrument-Total Ozone Mapping Spectrometer (OMI-TOMS) on the Nimbus 7satellite [Antón et al., 2009b]. The regression analysis was performed for clear-sky conditions (hourly clearness index higher than 0.65, hourly oktas smaller than 1, and hourly AODs at 380 and 440 nm smaller than 0.1). Thus, we have implicitly assumed that the atmospheric aerosol observed on a clear day is the natural background. The values of UVER, TOC, and μ0 recorded during 2008 under identical clear-sky conditions were used for validating the empirical model.
Bush and Valero  defined the AFE as the rate at which the irradiance at a certain wavelength range is “forced” per unit of AOD. Therefore, this variable in the erythemal range (AFEery) can be expressed as
 In order to obtain AFE it is necessary to have a wide range of AODs with high and low aerosol loads [García et al., 2006]. Table 1 shows the mean values (± one standard deviation) of the AOD values at 380 and 440 nm for each month along the monitoring period in Granada. Tthe monthly averages of the Angström exponent (440–870 nm) have also been added to the table. This last quantity characterizes the AOD spectral dependence and gives information about the size of predominant particles that comprise the atmospheric aerosol. For the AOD at 380 nm (440 nm), the lowest monthly median value of 0.14 (0.12) was reached in November, while the highest 0.26 (0.24) was obtained in May. It can be observed that summer AOD values are higher than winter ones, in agreement with the seasonal results shown by Alados-Arboledas et al.  and Lyamani et al.  at the same site.
Table 1. Monthly Mean Values (±1 SD) of AODs (at 380 nm and 440 nm) and the Angström Exponent (440–870 nm)
AOD at 380 nm
AOD at 440 nm
0.21 ± 0.11
0.17 ± 0.09
1.4 ± 0.4
0.23 ± 0.11
0.20 ± 0.10
1.2 ± 0.5
0.17 ± 0.10
0.14 ± 0.09
1.3 ± 0.4
0.24 ± 0.14
0.21 ± 0.12
1.0 ± 0.4
0.26 ± 0.12
0.24 ± 0.12
0.9 ± 0.4
0.25 ± 0.13
0.22 ± 0.12
0.9 ± 0.3
0.25 ± 0.13
0.23 ± 0.12
0.6 ± 0.3
0.23 ± 0.12
0.21 ± 0.11
0.6 ± 0.3
0.25 ± 0.13
0.22 ± 0.12
1.0 ± 0.4
0.23 ± 0.16
0.21 ± 0.16
0.9 ± 0.4
0.14 ± 0.05
0.12 ± 0.05
1.4 ± 0.3
0.20 ± 0.02
0.17 ± 0.02
1.4 ± 0.2
 In contrast to the aerosol optical depth, the monthly Angström exponents present an opposite pattern with smaller values during summer than winter. The outstanding negative correlation between the AOD and the Angström exponent in combination with the reduced value of this last variable is clear evidence of the presence of a large contribution of coarse particles to the atmospheric aerosols included in the vertical column [Lyamani et al., 2005]. The CIMEL Sun photometer used in this work is located in a region frequently affected by Saharan mineral dust intrusions because of its proximity to North Africa. In this region, most of these Saharan dust outbreaks occur in summer at high altitudes in the atmosphere [Guerrero-Rascado et al., 2008], and rising particles load in the troposphere and therefore increase the columnar AOD [Lyamani et al., 2005; Guerrero-Rascado et al., 2009]. Moreover, forest fires represent an additional source of aerosol particles during summer in this region [Lyamani et al., 2006b; Alados-Arboledas et al., 2011]. Thus, the advection of Saharan dust and/or biomass burning aerosols to Granada may explain both large values of the AOD and small values of the Angström exponent measured in summer.
 The hourly AOD histograms for cloud-free conditions in Granada (not shown) indicated that the majority of the AOD data (66% at 440 nm and 63% at 380 nm) were in the range 0.1–0.3. In addition, there was also a large percentage of data (16% at 440 nm and 11% at 380 nm) corresponding to clean conditions (AOD smaller than 0.1) and a small but significant number of cases (3.3% at 440 nm and 4.1% at 380 nm) with high aerosol loads (AOD higher than 0.5). These results emphasize that the distribution of the AOD in Granada presents a broad range of values, feasible for determining the AFE.
4.2. Clear-Sky Model
 In order to obtain the AFE in the erythemal range, the UVER values at the surface under a clean atmosphere (cloud-free and clear aerosol conditions, UVER0) were estimated. Multiple regression analysis was performed for evaluating coefficients from equation (5). The analysis showed satisfactory results with a coefficient of determination R2 ∼ 0.99 and a root mean square error (RMSE) of 3.9%. The results indicated that around 99% of the UVER variability under clear-sky conditions in Granada may be attributed to changes in the solar zenith angle and the total ozone column. Measured and estimated UVER0 values were compared in order to validate the reliability of the empirical model, using data recorded under clear-sky conditions during 2008 (not previously used for calculating fitting coefficients). Figure 1 shows the correlation between measured and modeled UVER0 data for Granada. The solid line, representing the zero-bias line, fits to the data, confirming the high degree of proportionality. A linear regression analysis between measured and modeled UVER0 was performed; R2 values higher than 0.99 indicated the similarity among the clear-sky cases sampled. The statistical analysis renders slopes very close to unity, supporting the validity of the empirical model used. In order to analyze relative differences between measured and modeled UVER0 values, the statistical parameters mean bias error (MBE) and mean absolute bias error (MABE) were calculated, (UVERexp − UVERmod)/UVERexp. The results indicated that the empirical model estimates UVER0 with a MABE of 2.4%. The small uncertainty (standard error 0.015%) of this parameter indicates the statistical significance of the reported value. In addition, the MBE showed a small positive value (+0.18%) resulting in a slight overestimation of UVER0 arising from the empirical model.
4.3. Aerosol Forcing Efficiency in the Erythemal Range
 Besides experimental and estimated UVER values, the UV surface albedo is the other variable necessary to obtain the ARF in the erythemal range. Because the surface albedo for spectral solar irradiances in the UV region is low (typical values between 0.02 and 0.08) compared with the surface albedo in the visible spectral range [Feister and Grewe, 1995], it is usual to calculate the ARF in the UV region assuming a null value of surface albedo [García et al., 2006; Díaz et al., 2007; Kazadzis et al., 2009]. This assumption can introduce a significant overestimation (2% to 8%) of the real ARF values in the UV range. To overcome this, the surface UV albedo for Granada was derived from the COST 726 UV albedo climatology over Europe. This product provides daily values of surface albedo at 360 nm for the period 1958–2002 in a 1.0° (longitude) × 1.0° (latitude) grid. Thus, an annual climatology of the surface UV albedo in Granada was used to infer aerosol radiative forcing in the erythemal range at this location. This climatology is calculated as the monthly average of the daily UV albedo values provided by the COST 726 data set, which ranged from 0.032 (April) to 0.045 (October).
Figure 2 shows ARF in the erythemal range as a function of AOD at 380 nm (Figure 2, top) and 440 nm (Figure 2, bottom) for the period 2006–2008. ARF has been calculated every minute using instantaneous UVER data and the corresponding UVER0 estimates (equation (4)). Simultaneous ARF and AOD values are grouped in two data sets according to their SZA: 25° ± 2.5° in red and 55° ± 2.5° in blue. It can be seen that there is a large spread of the data for each SZA data set that can be associated with the size of the SZA interval used in the analysis (5°). Additionally, this spread may be also related to the variability of the size and the composition of the particles within each data set. The two plots clearly show that ARF is more negative as AOD increases. Moreover, it can be seen that this relationship strongly depends on the SZA. Thus, for a fixed AOD, ARF is smaller (in absolute values) for 55° than for 25° SZA because of the longer path through the atmosphere, attenuating the direct beam component of the solar irradiance but also increasing the diffuse component that is due to multiple scattering. This is the case for UV wavelengths since scattering increases more rapidly with decreasing wavelength than absorption. Additionally, the diffuse component presents a large contribution to the global UV irradiance as a consequence of the strong wavelength dependence of Rayleigh scattering. For instance, approximately 50% of global UV irradiance is diffuse for low SZA, and the percentage rises with increasing SZA. Thus, the diffuse component is dominant at high SZA, and consequently, the measured UV irradiance at the surface becomes less sensitive to changes in the aerosol load. Our results agree with the work of Horvath et al.  which analyzed the wavelength dependence of the surface ARF for different SZAs. These authors showed that the largest effect of the SZA on ARF occurs for short wavelengths because of a notable higher vertical optical depth related to the wavelength dependence of the scattering or extinction.
 The AFEery values were derived following the approach described by García et al. . The regressions shown in Figure 2 were calculated by averaging all ARF and AOD data contained inside 0.1 AOD bins and are shown as black dots in the plots. Error bars are also displayed, showing one standard deviation. The slope of the lines represents AFEery,as explained in Section 3. As can be seen from Figure 2, this slope also depends on the SZA. Thus, AFEery is larger (in absolute values) for the SZA at 25° than at 55°, in agreement with recent results found in literature [e.g, di Sarra et al., 2008; Di Biagio et al., 2009], which show that the AFE in the shortwave spectral range decreases (in absolute values) for increasing SZAs for several aerosol types. Table 2 shows the AFEery obtained over ±2.5° intervals of SZAs at 20°, 25°, 30°, 35°, 40°, 45°, 50°, and 55°. When the SZA changes from 20° to 55°, the AFEery varies from −62 ± 5 to −26 ± 2 mW/m2 per unit of AOD at 380 nm and from −65 ± 1 to −30 ± 1 mW/m2 per unit of AOD at 440 nm. We would like to highlight that for SZAs smaller than 25°, an increase of one unit in the AOD causes an UVER decrease larger than 60 mW/m2 in Granada. The highest UVER values measured for these SZAs at this location are reached in summer under cloud-free conditions, being around 250 mW/m2 [Antón et al., 2011b]. Therefore, the atmospheric aerosol load can work as an outstanding filter of UV erythemal irradiance leading to a notable reduction (∼25%) of this variable at the surface when an increase of one unit in the AOD is observed.
Table 2. Aerosol Radiative Forcing Efficiency in the Erythemal Range (AFEery) Obtained Over ±2.5° Intervals of SZA at 20°, 25°, 30°, 35°, 40°, 45°, 50°, and 55°
Solar Zenith Angle (deg)
AFEery (mW/m2 per unit AOD at 380 nm)
AFEery (mW/m2 per unit AOD at 440 nm)
−62 ± 5
−65 ± 1
−60 ± 3
−60 ± 2
−56 ± 4
−54 ± 3
−55 ± 2
−55 ± 1
−49 ± 1
−49 ± 3
−41 ± 2
−41 ± 3
−35 ± 1
−30 ± 5
−26 ± 2
−30 ± 1
4.4. Variability of ARFery With Particle Size and Single-Scattering Albedo
 In addition to the solar zenith angle, the evaluation of the AFEery is also affected by changes of the size and the composition of the particles throughout the study period [Conant et al., 2003; Kim et al., 2005; Ramana and Ramanathan, 2006; di Biago et al., 2009]. The aerosol size determines the AOD dependence on the wavelength through the Angström exponent. By contrast, the composition of the aerosol particles determines the relative weight of scattering and absorption on extinction, as described by the single-scattering albedo (SSA).
 To evaluate the effect of aerosol particles size in the AFEery, simultaneous values of instantaneous ARF and AOD at 380 nm data were divided into four groups according to the Angström exponent (<0.5, 0.5–1.0, 1.0–1.5, and >1.5). The influence of the SZA in the AFEery is removed by selecting simultaneous ARF, AOD, and Angström exponent data for a narrow range of SZA values (40° ± 2.5°). Figure 3 shows the correlation between ARF and AOD data for each selected group. The slope of the linear fit (AFEery) increases as the Angström exponent decreases. Thus, AFEery changes from (−52 ± 4) mW/m2 per unit of AOD for Angström exponents smaller than 0.5 to (−29 ± 3) mW/m2 per unit of AOD for Angström exponents higher than 1.5. We also performed this analysis for the AOD at 440 nm and other SZA intervals (not shown). Although the numbers change mainly because of the AFEery dependence on SZA, the influence of the Angström exponent is the same: The largest aerosol particles (Angström exponent <0.5), which are mainly related to desert dust, are responsible for a higher radiative forcing efficiency (in absolute value) compared with the smallest particles (Angström exponent >1.5) associated with urban and industrial aerosols. This behavior may be related to the fact that desert dust aerosols attenuate UV radiation more strongly than other aerosol types with the same AOD [Krotkov et al., 1998; Meloni et al., 2003a]. In addition, several authors have indicated enhanced absorption for coarse particles in the UV range compared with the visible because of their lower SSA (ratio between aerosol scattering and extinction) at UV wavelengths [Mattis et al., 2002; Meloni et al., 2004]. Therefore, besides size, aerosol absorption is also expected to play an outstanding role on the aerosol forcing efficiency in the UV erythemal range.
 The available SSA in Granada given by the Aerosol Robotic Network (AERONET) is the most relevant parameter to define the absorbing properties of aerosols at this location. Simultaneous values of instantaneous ARF and AOD at 440 nm and SSA at 440 nm were selected to derive AFEery from the slope of ARF versus AOD considering two intervals for SSA (0.7 < SSA < 0.85 and 0.85 < SSA < 1.0). The simultaneous ARF, AOD, and SSA data were selected for a reduced range of SZA values (between 50° and 60°) to remove the influence of the geometry parameter in the analysis. Figure 4 shows the correlation between ARF and SSA data for the two selected groups. It can be seen that the slopes of the two linear fits are clearly different. Thus, the AFEery values are −59 ± 9 mW/m2 per AOD unit for SSA smaller than 0.85, and −28 ± 1 mW/m2 per AOD unit for SSA larger than 0.85, showing that stronger aerosol absorption leads to a larger surface forcing efficiency (in absolute value). These results are in agreement with recent works, for instance, Di Biagio et al.  have shown that the surface forcing efficiency at Lampedusa (central Mediterranean) decreases with increasing the SSA for several aerosol types. Additionally, the mean SSA value at 440 nm for the four aerosol size categories defined by the Angström exponent was obtained. These values were 0.85 ± 0.05, 0.85 ± 0.06, 0.87 ± 0.06, and 0.89 ± 0.05 for those cases with Angström exponents smaller than 0.5, between 0.5 and 1, between 1 and 1.5, and larger than 1.5, respectively. It can be seen that the differences between the four SSA values are not significant but it is appreciated that the smallest aerosols (alpha larger than 1.5), which presented the smallest ARFery (in absolute value), have the highest SSA values (lower absorption properties).
 Finally, the seasonal analysis of the AFEery was performed in Granada. Simultaneous values of instantaneous ARF and AOD data at 380 and 440 nm were selected for each season of the year in order to derive the seasonal AFEery from the slope of the regression line between ARF and AOD values. Table 3 shows the values of this variable for spring (March to May), summer (June to August), autumn (September to November), and winter (December to February). It can be observed that the AFEery values obtained in spring and summer present higher values (in absolute values) than those obtained in autumn and winter. This behavior may be partially related to the SZA influence on AFEery described above, since the smallest SZAs are recorded in summer. In addition, the Angström exponent presents a clear seasonal pattern in Granada with the following values: 1.06 (spring), 0.70 (summer), 1.10 (autumn), and 1.38 (winter). The aerosol particles present, on average, a larger size in summer months than during the rest of the year, resulting in higher AFEery values (in absolute values) during this season. The changes in AFEery are also sensitive to variations in the absorbing characteristics of atmospheric aerosols. Thus, the seasonal mean values of the SSA in Granada during the period of study were calculated with values of 0.86 (spring), 0.85 (summer), 0.86 (autumn), and 0.91 (winter). Therefore, the seasonal changes obtained in the AFEery suggest that spring-summer months in Granada present larger atmospheric aerosols with higher absorption properties in the UV region than those found during the rest of the year.
Table 3. Aerosol Radiative Forcing Efficiency in the Erythemal Range (AFEery) Obtained for Each Season
AFEery (mW/m2 per unit AOD at 380 nm)
AFEery (mW/m2 per unit AOD at 440 nm)
−50 ± 3
−56 ± 7
−59 ± 4
−63 ± 4
−45 ± 5
−47 ± 4
−22 ± 2
−23 ± 4
 In the previous section, factors influencing the AFEery were examined: solar zenith angle, aerosol size distribution by means of the Angström exponent, and aerosol absorption using the single-scattering albedo. This section includes a discussion regarding the relevance of each term. The relative variability of AFEery with respect to the relative variability of each variable has been analyzed by means of the following expression:
where X corresponds to each of the three factors influencing AFEery and n is the number of AFEery-X values. This study has been performed using the bin analyses described in the previous sections.
Table 2 shows the AFE obtained at ±2.5° intervals for SZA between 20° and 55°. Equation (7) has been applied to these values (n = 8), obtaining δAFESZA equal to 0.93 (at 380 nm) and 0.83 (at 440 nm). These results indicate that when the SZA increases by 1% in Granada, AFEery decreases by 0.83%–0.93% (in absolute terms) approximately. On the other hand, the influence of alpha on AFE was analyzed by selecting four alpha intervals (see Figure 3). Thus, the central value of alpha for each interval has been used with equation (7): 0.25, 0.75, 1.25, and 1.75 (n = 4). The value obtained for δAFEalpha is 0.25. Finally, the influence of SSA on AFE has been quantified using the two AFE values obtained for the two SSA intervals analyzed in this paper (0.7 < SSA < 0.85 and 0.85 < SSA < 1.0). The two SSA values used in expression (7) correspond to the central values for each interval (n = 2). Therefore, the value obtained for δAFESSA is 2.7, indicating that when the SSA increases by 1% in Granada, AFE decreases by 2.7% (in absolute terms) approximately. In summary, the variable with the largest influence on AFE is the SSA, followed by the SZA, while alpha is the variable presenting the smallest influence.
 Furthermore, it is interesting to note that computations on SSA were performed at 440 nm, far from the maximum sensitivity of the UVER range (around 305 nm). The spectral dependence of absorption properties of aerosols is under debate. Barnard et al.  showed that the SSA has only a little wavelength dependence between 300 and 400 nm for atmospheric aerosols in the Mexico City area. In the same city, Corr et al.  also found no significant SSA variability between 332 and 368 nm. Conversely, other works have indicated enhanced absorption in the UV compared with the visible. For instance, Meloni et al.  showed that in order to reproduce UV spectral irradiances measured by a Brewer spectrophotometer using a radiative transfer model, SSA values smaller than those in the visible (down to 0.78) are needed. A significant enhancement of the SSA between 400 and 500 nm was shown by Barnard et al.  and associated with increased absorption at UV wavelengths that was due to organic aerosols. Alados-Arboledas et al.  reported differences in the spectral dependency of the SSA for a contamination episode and a Saharan dust outbreak; while the SSA sharply decreased with wavelength during the contamination episode, this variable was enhanced with wavelength during dust events. Cachorro et al.  also showed that SSA values at 870 nm were larger than at 440 nm during desert dust episodes. Therefore, a larger decrease in SSAs is expected for categories affected by mineral dust, i.e., a lower Angström exponent.
 A detailed evaluation of the AFE in the UV erythemal range at an urban location in southeastern Spain during the period 2006–2008 drew important features in relation to the UV radiation reaching the surface. An increase of one unit in AOD reduces harmful UVER more than 60 mW/m2 for SZAs lower than 25° at the study site. This shows the great influence of aerosols on UV radiation and argues for the need for concurrent measurements of UV radiation and atmospheric aerosols at midlatitudes.
 AFEery presents a strong dependence on SZAs, with values between −62 and −26 mW/m2 per unit of AOD at 380 nm when the SZA varies from 20° to 55°. In addition, changes in the aerosol size and absorption properties in Granada throughout the year also produce a significant effect on AFEery. Thus, both large aerosols and strong aerosol absorption lead to a larger surface forcing efficiency (in absolute value). According to our results, a large load of mineral dust can lead to appreciable reductions in the UVER at the surface level. This is important, considering that the large particles associated with North African outbreaks are frequent in summer, when the UVER levels can reach their highest values. Thus, the mineral aerosol load could partially filter out UV erythemal irradiance, leading to a notable reduction (∼25%) in their values reaching the Earth's surface.
 Manuel Antón thanks Ministerio de Ciencia e Innovación and Fondo Social Europeo for the award of a postdoctoral grant (Juan de la Cierva). This work was partially supported by the Andalusian Regional Government through projects P08-RNM-3568 and P10-RNM-6299, the Spanish Ministry of Science and Technology through projects CGL2010-18782 and CSD2007–00067, and by the European Union through ACTRIS project (EU INFRA-2010-1.1.16–262254).