Atmospheric Chemistry and Physics Discussions

C. Spyrou, G. Kallos, C. Mitsakou, P. Athanasiadis, C. Kalogeri, and M. Iacono University of Athens, School of Physics, Division of Environmental Physics-Meteorology, BldgPHYS-V, 15784, Athens, Greece King’s College London, Environmental Research Group, Franklin-Wilkins Building, 150Stamford Street, London SE1 9NH, UK Centro Euro-Mediterraneo per i Cambiamenti Climatici, Viale Aldo Moro, 44 I, 40127Bologna, Italy Atmospheric and Environmental Research (AER), 131 Hartwell Avenue, Lexington, MA02421-3126, USA

As a second type of effect, aerosols function as particles on which water vapor accumulates during cloud droplet formation, acting as cloud condensation nuclei (Levin et al. 2005;Solomos et al., 2011). Modifications of the number concentration or the 10 hygroscopic properties of dust tend to modify the physical and radiative properties of clouds (Twomey, 1977;Khan et al., 2009) and the possibility and intensity with which a cloud can precipitate (e.g. Liou and Ou, 1989;Albrecht, 1989;Solomos et al., 2011). All the various changes in cloud processes imposed by aerosols are referred to as "aerosol indirect effects" (e.g. Solomos et al., 2011). Finally by absorbing solar radia- 15 tion, particles contribute to the reduction in cloudiness, a phenomenon referred to as the semi-direct effect (Khan et al., 2009).
The magnitude of the feedback on the radiative transfer depends strongly on the optical properties of particles (single scattering albedo, asymmetry parameter, extinction efficiency), which in turn depend on the size, shape and refractive indices of dust par-20 ticles (Tegen, 2003;Pace et al., 2006;Helmert et al., 2007). The mineral composition of the dust source areas, as well as the chemical composition and transformation of aerosols during their transportation, are all factors that influence the optical intensity of dust (Tegen, 2003;Wang et al., 2005;Astitha et al., 2010). Furthermore, the vertical distribution of dust, the presence of clouds and the albedo of the surface all contribute Introduction  Reddy et al. (2005) simulated a decrease of 0.28 W m −2 on the shortwave and an increase of 0.14 W m −2 on the longwave. Other studies and measurements have produced similar values for the global dust radiative effect (IPCC, 2007). However, little work has been done to incorporate the dust radiative feedback in regional and mesoscale models 5 (Stokowski, 2005).
If we look at the radiative impacts at a local scale during dust episodes, the DRE exhibits much stronger signals. Haywood et al. (2003) measured a decrease of incoming shortwave radiation up to 130 W m −2 (instantaneous value) at the coast of West Africa during the passage of a dust storm in September 2000. During a severe dust episode 10 in Niger on 3-12 March 2006 the ARM (Atmospheric Radiation Measurement) mobile facility detected a maximum shortwave decrease of 250 W m −2 on the surface, where at the same time the reflected radiation at the top of the atmosphere increased by 100 W m −2 (Slingo et al., 2006). For the same episode, the emitted longwave radiation at the surface decreased by approximately 100 W m −2 , corresponding to the "cooling" 15 imposed by dust shading. The local effects of airborne dust have been thoroughly examined in a number of other studies. The main purpose of this work is to examine the long-term implications of the dust-radiation interactions. In order to model these diverse feedbacks on the radiation balance, the SKIRON/Dust limited area model has been used (Kallos et al., 20 1997a(Kallos et al., 20 , b, 2006Papadopoulos et al., 2002;Spyrou et al., 2010). Continuing from these studies the model has been updated further to include the Rapid Radiative Transfer Model -RRTMG (Mlawer et al., 1997;Oreopoulos et al., 1999;Iacono et al., 2003;Pincus et al., 2003;Baker et al., 2003;Clough et al., 2005;Morcette et al., 2008;Iacono et al., 2008). The addition of the RRTMG scheme has made it possible to model and 25 study the effects of desert dust particles on the radiation balance of the atmosphere.
The following section summarizes the main characteristics of the modelling system while Sect. 4 outlines the new radiative transfer scheme that was incorporated in SK-IRON/Dust system. Section 5 depicts the optical properties of suspended particles and in particular dust aerosols that are discussed in the present study. The experimental setup that was designed for the present simulations is described in Sect. 6. The model evaluation and the sensitivity tests for the estimation of the energy impact are analysed in Sects. 7 and 8, respectively. Finally, the last section summarizes the main conclusions of the study.

SKIRON/Dust modelling system
The SKIRON/Dust modeling system is based on the atmospheric model SKIRON, which has been developed at the University of Athens from the Atmospheric Modelling and Weather Forecasting Group (Kallos et al., , 2006Nickovic et al., 2001;Spyrou et al., 2010) in the framework of a number of projects (SKIRON, MEDUSE, ADIOS, 10 CIRCE and recently MARINA Platform). A dust module that simulates the production and removal of the desert dust aerosol is directly coupled with the host model. The dynamical core of the model is based on the ETA concept, which was originally developed by Mesinger (1984Mesinger ( , 1988 and Janjic (1990Janjic ( , 1994. The SKIRON atmospheric model includes several sophisticated parameterization schemes, such as the OSU (Oregon

15
State University) scheme (Ek and Mahrt, 1991) for simulation of the surface processes, including a data assimilation scheme for soil temperature and soil wetness, and the option of choosing among Betts- Miller-Janjic (Betts, 1986) and Kain-Fritsch (Kain and Fritsch, 1990)  Recently, the modelling system was significantly upgraded in order to improve the model prediction efficiency meeting the current needs for accurate simulation of the mineral dust cycle and the interaction mechanisms with various atmospheric processes. New features include improvements in the description of the bottom boundary (ground or sea surface) characteristics of the atmospheric model and the dust aerosol 5 properties (Spyrou et al., 2010). The new model version includes a 16-category soil characteristics dataset (Miller and White, 1998) that provides detailed information on soil physical properties, such as porosity and available water capacity. A high-resolution (30 s) global land use/land cover database including urban areas and classified according to the 24-category USGS land use/land cover system (Anderson et al., 1976) is utilized. For the more accurate description of the topographic variability that determines the incoming solar radiation reaching the surface, in the upgraded SKIRON/Dust model a new preprocessor was developed that derives statistics for the slope steepness and orientation, from the high resolution topography datasets. The dust aerosol is described by using the three-modal lognormal function of D'Almeida (1987) for the 15 aerosol mass distribution at the source areas and the 8-size bin transport mode of Schulz et al. (1998) for the long-range transported particles. The dust particles are assumed to be mobilized through the process of saltation bombardment (Marticorena and Bergametti, 1995) and deposited via dry (diffusion, impaction, gravitational settling) and wet (in-cloud and below-cloud removal) mechanisms. More details on the Introduction model, RRTM, were developed in the context of continual comparison to Line-by-line Radiative Transfer Model -LBLRTM, which is an accurate and highly flexible model that continues to be validated with measured atmospheric radiance spectra from the sub-millimeter to the ultraviolet (Clough et al., 2005;Turner et al., 2004). This approach realizes the goal of providing an improved radiative transfer capability that is 5 directly traceable to measurements. Molecular absorbers in RRTMG include water vapor, ozone, carbon dioxide, nitrous oxide, methane, oxygen, nitrogen and four halocarbons (CFC11, CFC12, CFC22, and CCL4) in the longwave and water vapor, carbon dioxide, ozone, methane and oxygen in the shortwave. The water vapor continuum is based on CKD v2.4, and molecular line parameters are based on HITRAN 2000 for 10 water vapor and HITRAN 1996 for all other molecules (Iacono et al., 2008;Rothman et al., 2009). RRTMG uses sixteen spectral bands to represent the longwave region, while the shortwave band is represented by fourteen spectral intervals. Absorption and emission from aerosols and clouds are included in the longwave, and the shortwave treatment includes extinction (absorption plus scattering) from aerosols, clouds and 15 Rayleigh scattering. Aerosol radiative effects are treated in RRTMG through the specification of their optical properties within each spectral interval. RRTMG incorporates several modifications to RRTM (while retaining the same basic physics and absorption coefficients) in order to improve computational efficiency, to update the code formatting for easier application to global and limited area models, and 20 to represent sub-grid scale cloud variations. The complexity of representing fractional cloudiness and cloud overlap in the presence of multiple scattering is addressed in RRTMG with the use of McICA, the Monte-Carlo Independent Column Approximation (Pincus et al., 2003), which is a statistical technique for representing representing radiative impacts of sub-grid scale on cloud variability including cloud overlap. Although 25 this method introduces a certain random noise to the cloudy calculation of radiance, the result is unbiased. This approach provides a much needed flexibility in representing the vertical correlation of clouds (cloud overlap) in some detail by imposing an assumed relation (random or maximum-random) among the stochastic cloud arrays in the vertical Introduction dimension. The maximum-random cloud overlap assumption, in which adjacent cloud layers in the vertical are presumed to overlap maximally and non-adjacent cloudy layers are assumed to overlap randomly, is applied within RRTMG in the SKIRON/Dust model.

5
The absorption and scattering of light by a spherical particle is a classic problem in physics, the mathematical formulation of which is referred to as Mie theory (Bohren and Huffman, 1983). The most important factors that govern these processes are the wavelength of the incident radiation, the size of the particle, the ratio of the circumference of the particle to the wavelength and the complex refractive index, expressed 10 as RI=n+i k. The real part n expresses attenuation due to scattering (non-absorbing), while the imaginary part is related to the absorption of light by the medium (Helmert et al., 2007). The refractive indices of various spectral ranges, for all the particle sizes, have been determined using the OPAC (Optical Properties of Aerosols and Clouds) software pack- 15 age (Hess et al., 1998), which in turn utilizes the Mie theory for its calculations (Quenzel and Muller, 1978). The same package was used for the definition of the aerosol single scattering albedo ω and the asymmetry parameter g for the same wavelengths. For the 550 nm spectral window, where the extinction of the incoming solar radiation is most intense, a single scattering albedo value of 0.95 was used instead of the smaller value 20 (0.83) provided by OPAC, for transported mineral dust. Field measurements suggest this value as more realistic for desert dust in the mid-visible (Kalashnikova et al., 2005;R. Kahn personal communication, 2009).
For the calculation of the extinction efficiencies Q ext,i for each size bin i , at each wavelength λ, the Wiscombe Mie algorithm was used (Wiscombe, 1980;Mishchenko 25 et al., 2002). Although dust particles are known to be non-spherical (Nakajima et al., 1989;Meloni et al., 2004), Mie calculations can be used to compute the radiative ACPD 13,2013 Radiative effects of desert dust on weather and regional climate C. Spyrou et al. parameters for equivalent volume spheres and provide very good representation for non-spherical scattering (Tegen and Lacis, 1996). The aerosol optical thickness at each wavelength τ(λ) was calculated by the following formula: where N = 8 is the number of particle size bins, ρ = 2.65 Kg m −3 is the particle density, r is the particle radius, DL is the layer dust load and Q ext,i (λ) is the extinction efficiency calculated from the Mie theory (Tegen and Lacis, 1996;Perez et al., 2006).

Experimental design
In order to evaluate and quantify the effects of desert dust particles on the radiation 10 budget in the atmosphere, a series of test runs were carried out. The model was integrated for a period of 6 yr (2002)(2003)(2004)(2005)(2006)(2007) over an extended area that covers the European continent, the Mediterranean Sea and northern Africa, as well as a major part of the Middle East and Turkey. The horizontal grid increment was 0.25. A vertical dimension of 38 levels was used stretching from the surface up to 20 km. The ECMWF operational 15 analysis dataset (horizontal resolution of 0.25 degree) was used for the initial and lateral boundary conditions for the meteorological parameters. The timestep was set at 60 s and the radiation driver was invoked every 15 min. The model runs were carried out under two different setups: (a) by neglecting the effects of dust particles on the radiative parameters (NDE) and (b) by including the dust-20 radiation interaction mechanisms (WDE). Between the two setups, the differences in the radiative fluxes and heating/cooling rates, as well as other atmospheric parameters, are calculated and discussed in the following paragraphs.

1335
ACPD 13,2013 Radiative effects of desert dust on weather and regional climate C. Spyrou et al.

Model evaluation -qualitative study and statistical techniques
To estimate the impact on model performance of dust-radiation interaction processes, a specific case of desert dust transport towards the SE Europe was examined. On the 5 April 2006 strong winds originating from across the Southeastern side of the Atlas mountain range injected large quantities of mineral dust particles into the atmosphere.

5
The particles were transported over the Mediterranean Sea and two days later, on the 7 April 2006, reached Crete on the Eastern Mediterranean ( Fig. 1). To evaluate the effects of dust on the radiative transfer, solar radiation data retrieved from a pyranometer located at the Technological University (TEI) of Crete were used. Model runs with the two different setups (NDE, WDE) were carried out for the simulation of this test case.
During the passage of the dust plume, the incoming solar radiation decreased by approximately 200 W m −2 (Fig. 2). This abrupt change is depicted by the WDE simulation to a certain extent, but cannot be attributed solely on the shading effect of desert dust particles. One viable explanation is that the change in the radiation balance of the atmosphere, due to the presence of dust particles, led to cloud formation in the area 15 that is blocking the incoming solar radiation observed at the station in Crete.
To further investigate this hypothesis the vertical profiles of the cooling rates have been derived. In Fig. 3, the cooling rates together with the vertical extent of dust concentration are shown. The cooling rates of the WDE simulation exhibit a sharp decrease of 4 K day −1 just above the dust plume in a layer about 9 km high. The dust particles 20 seem to have an albedo effect, partly reflecting back the incoming solar radiation. As a side effect of this reflection, cooling of the adjacent atmospheric layer over the dust plume is simulated. This temperature decrease drives the air temperature closer to the dew point temperature and leads to the formation of a cirrus cloud over the station in Crete, as shown in Fig. 4. The SKIRON/Dust model (WDE run) appears to capture this ACPD 13,2013 Radiative effects of desert dust on weather and regional climate For a more thorough evaluation of the model performance, a series of statistics were derived. At first, in order to evaluate the radiation calculations, the model outputs were compared with measurements obtained from three of the AERONET network stations (Crete, Sede Boker and Moldova) for the year 2006. The characteristics of the stations (geographical location and altitude), as well as the statistical parameters (regression 5 trend lines, correlation coefficients) determined from the analysis, are shown in Table 2. The flux comparisons were performed for the days of increased values of aerosol optical depth (AOD). Thus, AOD measurements at 500 nm from the AERONET stations were also obtained and the 95th and 99.5th percentiles were calculated for each dataset in order to detect the periods of increased aerosol optical depth in the atmo-10 sphere. The 95th percentile of AOD was found at about 0.3, while the 99.5th percentile fluctuated between 0.5 and 0.6 for the three studied sites. In general, the SKIRON/Dust model produces accurate simulations of the incoming radiation, since the correlation coefficients calculated for the compared datasets have high values and the linear regression slopes of the trend lines are close to unity. 15 However, the intercept of the regression lines, which indicate the difference between the simulated and the observed values, are very high and exceed 50 when the dust effects on radiation are not considered. This overestimation of the observed radiation fluxes by the model in the NDE setup, is directly related to the neglection of the radiative effects of the dust aerosols. On the other hand, the inclusion of dust-radiation 20 interaction processes appears to diminish significantly this overestimation at Crete and Sede Boker, with a complete elimination noticed at Sede Boker (−0.8) for AOD values greater than 0.6. The decrease of the model overestimation in Moldova (the most northern site out of the three) with the incorporation of dust effects appears to be negligible. Since the dust load that reaches Moldova is reduced compared to the amounts that 25 affect the more south sites, the radiative forcing due to the suspension of dust particles is expected to be reduced.
To evaluate the impact of dust particles on the model calculation of air temperature, simulation runs for a 6 yr period (2002)(2003)(2004)(2005)(2006)(2007)  In general, the inclusion of dust-radiation interaction processes appears to enhance the model performance in estimating the radiative fluxes and near surface air temperature. 20 The life cycle of dust is defined as the emission, atmospheric loading and surface deposition of the desert dust particles. Quantification of the dust mass over the selected model domain is essential for the determination of the modified amounts in the energy budget due to the suspension of desert dust particles. To estimate the desert dust impact on the energy budget over the studied area, the 6 yr model runs covering the conditions, while the WDE simulations represent the dust transport events by taking into consideration the dust-radiation interaction processes. Firstly, the areas that are mostly affected by the emission and transport of desert dust particles were detected. In Fig. 6a-d, the dust load (integrated column) values for the different seasons of the year are depicted across the studied area. Significant dust amounts appear to be over Chad with dust load values exceeding 1 g m −2 and at the northwestern part of the African continent over Mauritania and Algeria. The suspension of desert dust particles is also present across and around the Red Sea in the spring and summer periods. As reported by many studies (e.g. Moulin et al., 1997;Nickovic et al., 2001), the Saharan dust transport events are characterized by a strong seasonal variability. In particular, maximum dust impact is observed in central-eastern Mediterranean in spring, while the maximum is shifted in summer (and early autumn) to the central-western Mediterranean (Querol et al., 2009).

Model sensitivity analysis -estimation of the impact on energy budget
The seasonal differences in the incoming shortwave and longwave radiation fluxes reaching the surface are illustrated in Figs. 7 and 8. The differences in the shortwave fluxes appear to be negative during all periods of the year and greater during the warm spring and summer periods approaching the value of 70 W m −2 over the extended desert areas of Chad (this value corresponds to an area of approximately 40 000 Km 2 ), a very strong dust source. On the other hand, the depicted deviations over Europe and Mediterranean are negligible during the cold periods, because there is no signif-20 icant dust transport. However, in spring, the discrepancies over European land areas is found as high as 10 W m −2 over the south and 5 W m −2 near the central European countries. Another region of increased differences in the incoming shortwave radiation that are at a maximum during summer is the northwestern African continent and in particular over the southern desert areas of Morocco, where deviations greater than 25 40 W m −2 are modeled (Fig. 7c). As expected, the deviations between the two model setups (NDE and WDE) are more noticeable during the periods and in the areas with high dust productivity. In all cases, the suspension of dust aerosols leads considerable reduction of the incoming shortwave radiation due to scattering and absorption of the Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | solar radiation. As illustrated in Fig. 8a the total annual reduction of the incoming solar radiation (temporally integrated throughout the year) can be as high as 500 KWh m −2 , which represents a significant portion of the total available solar radiation (by temporally integrating the hourly flux values - Fig. 8b). This reduction should be taken into consideration when modeling radiative fluxes for solar energy applications.

5
According to the model results the incoming longwave radiation fluxes with the presence of dust particles are increased during all periods of the year (Fig. 9) with greater differences noticed in the spring. The longwave radiation emitted by the dust layer towards the surface during night-time is superimposed with the incoming solar radiation during the day. As a result the incoming flux that reaches the surface is enhanced. This To better understand the annual variability of desert dust feedback at regional scales, the monthly average radiative forcing has been calculated over the entire model do- 15 main, for the incoming shortwave (Fig. 10a) and incoming longwave (Fig. 10b) radiation. As expected the most profound effects are modeled during the spring, especially in May, reaching a maximum of approximately −14W m −2 for the shortwave, and +6 W m −2 for the longwave.
The net atmospheric forcing by the mineral dust aerosols in the model simulations 20 can be estimated as the difference in the radiative flux absorbed by the atmosphere between the two model configurations (WDE-NDE) over the model domain. This was derived for each season and the result is illustrated in Fig. 11a-d. The radiative flux that is absorbed by the atmosphere can be determined through: (2) 25 If we consider that the dust particles exert a negligible effect on the incoming radiation at the top of the atmosphere, then F can be expressed as F = F TOA ↑ NDE −F TOA ↑ WDE . In ACPD 13,2013 Radiative effects of desert dust on weather and regional climate C. Spyrou et al.  Fig. 11a-d denote an increase in the atmospheric absorption due to dust feedback and positive values a decrease. The modeled absorbed fluxes increase over the desert areas of the northern African continent, while during the summer months differences can be also noticed over southern and central European areas. The energy amounts absorbed by the dust aerosols 5 approach 10 W m −2 over land areas, while over the Red Sea the increase in the absorbed energy exceeds 20 W m −2 during the summer period. This can be attributed to the fact that the warm water of the Red Sea emits a large amount of longwave radiation that is trapped between the sea surface and the dust layer when there are increased concentrations of dust particles (greenhouse effect). On the contrary, the cold water of the North Atlantic emits less longwave radiation and the difference in atmospheric absorption detected is attributed to the emission from dust clouds (especially during the summer where dust particles move towards the Atlantic). The absorbed energy amounts by the dust particles can influence significantly the atmospheric stability by heating the lower tropospheric layer.  (Fig. 12a), which is related to the existence of more wide-ranging dust source areas over this part. On the other hand, the high mountains in the central region of northern Africa ( Fig. 13a) limit the 25 extent of the dust source areas. The vertical cross-sections of the temperature difference between the two model setups (WDE-NDE), as illustrated in Fig. 12b for the summer months and Fig. 13b for the spring months of the 6 yr study period, reveal the importance of the dust radiative ACPD 13,2013 Radiative effects of desert dust on weather and regional climate feedback: first of all dust particles reflect incoming radiation at high altitudes. This mechanism is reproduced by the model as a mid-tropospheric cooling of ∼ 0.1 K at ∼ 300 hPa. At the same time these particles increase the temperature of the dust layer through atmospheric absorption of the incoming SW radiation. According to the model results this increase in the air temperature profile reaches ∼ 0.2 K at about 600 hPa (mid-tropospheric heating). Temperature decreases in the lower troposphere (925 hPa up to 825 hPa) up to 0.3 K are due to extinction of the incoming solar radiation, which is more profound in the western part of the African continent (Fig. 12b). An interesting feedback is that near source areas the modeled temperature of the atmospheric layer near the ground in-10 creases by ∼ 0.4 K during the summer (Fig. 12b) and even up to ∼ 0.5 K during spring (Fig. 13b), even though the extinction of the incoming solar radiation is more profound there. This can be attributed to two superimposed factors: (1) during the day the dust particles absorb significant amounts of incoming solar radiation (as seen also in Fig. 9), raising the temperature of the near surface layer and (2) during the night this dust layer 15 "traps" longwave radiation emitted from the ground (greenhouse effect). As a result, the air temperature close to the surface increases, while a cooling of up to −0.2 K is produced by the model inside the dust layer.
Far from the desert areas over the Mediterranean Sea the surface temperature decreases due to scattering and absorption by the dust cloud. According to the model 20 results this reduction reaches −0.2 K on average (Figs. 12b and 13b).

Conclusions
This work represents a modelling study of the direct radiative effects of desert particles on the radiation budget of the atmosphere across the greater Mediterranean Region. The DRE is investigated with the aid of the SKIRON/Dust modelling system. New radia- data. Using the updated system the effects of desert dust particles on radiative transfer have revealed important findings, as summarized below: the presence of dust particles in the atmosphere has a significant regional influence on the radiative transfer and energy distribution. Therefore, variations in dust particle production can have impacts on radiative properties, cloud formation and water budget. These interactions are not one-5 way; there are feedbacks that are critical for both regional and climatological -scale phenomena.
An extensive statistical evaluation with temperature data from more than 600 stations of the ECMWF operational analysis has shown the improvement in the description of atmospheric processes when the dust-radiation feedback is included. This improvement 10 is evident in all the statistical scores used and is most profound during the transitional seasons.
According to the model the suspension of desert dust particles decreases the amount of radiative energy reaching the surface. At the same time the atmospheric absorption is increased throughout the simulation domain, as the dust particles absorb 15 both the incoming solar radiation and part of the longwave radiation emitted from the surface during the night.
In the vertical dimension, the phenomena are more complicated, as the redistribution of the energy has profound effects on the temperature profile: at high altitudes cooling is evident due to reflection from the dust clouds below. At the same time the temperature 20 of the mid-tropospheric layers is increased due to absorption and the near-surface layer temperature drops due to extinction from above.
A notable attribute of dust is the greenhouse effect on areas with high particle concentrations. Even thought the shading effect is dominant in these areas the longwave trapping imposed by dust leads to an increase of the near surface temperature, rather 25 than the expected reduction. This phenomenon is superimposed to the emission of infrared radiation by the dust cloud itself, further intensifying the heating effect. The latter is an indication that the links and feedbacks between dust and climate are very complicated. It must also be noted that in this study the indirect effects of natural particles are not included and the total contribution of desert dust on a regional scale remains to be identified. In general, the presence of desert dust particles in the atmosphere plays a key role in the energy distribution and the formation of the regional climate over the affected areas. The effects of airborne dust on the shortwave and longwave radiative fluxes 5 at the surface should be carefully considered, including indirect effects, when model data is used to assess the available solar energy for power plant applications in desert areas. Finally the dust influence on the energy parameters has a seasonal variability that is in line with the seasonality of the dust cycle events. Introduction

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version

1363
ACPD 13,2013 Radiative effects of desert dust on weather and regional climate C. Spyrou et al.   Vertical cross-sections of (a) dust concentration at 2 W and of (b) temperature difference at 2 W (WDE-NDE), as averaged for the summer months of the 6 yr period from the model simulations before and after incorporating dust-radiation interactions. The white color denotes the orography.
ACPD 13,2013 Radiative effects of desert dust on weather and regional climate C. Spyrou et al.    . 13. Vertical cross-sections of (a) dust concentration at 18 • E and of (b) temperature difference at 18 • E (WDE-NDE), as averaged for the summer months of the 6 yr period from the model simulations before and after incorporating dust-radiation interactions. The white color demotes the orography.