Direct radiative impacts of desert dust on atmospheric water content

ABSTRACT

The direct and indirect radiative impact of naturally produced dust particles influences climate from regional to global scale, introducing one of the largest uncertainties in future climate projections. By absorbing and scattering solar radiation, aerosols reduce the amount of energy reaching the earth's surface, while at the same time they enhance the greenhouse effect by absorbing and emitting longwave radiation (direct dust effect). In this study an attempt is made to quantify the feedback of this energy redistribution in the atmospheric water content in the Arabian Peninsula (one of the main sources of atmospheric mineral dust globally). To this end the SKIRON/dust modeling system was used for 2 years (2014–2015) and two sets of simulations were performed: in the first one the dust particles exert no feedback on the radiative transfer due to dust particles (control run), while in the second set dust interacts with radiation (direct radiative effect). Both simulations have been evaluated in their ability to describe the impacts on surface humidity, with the simulations including the dust feedback showing improved results. These direct feedbacks lead to an increase in the mass of water in the atmospheric column that can reach a maximum daily average of 0.5 g per kg of dry air. Water vapor is the most important greenhouse gas and through this process dust enhances its own greenhouse effect, further increasing the surface temperature and humidity, making life difficult for people living in an already harsh desert climate.

© 2018 American Association for Aerosol Research

EDITOR:

• Nicole Riemer

1. Introduction

The Arabian Peninsula is a major source of atmospheric mineral dust. Alongside with the Sahara desert, arid and semiarid regions in Central Asia and the Gobi desert in East Asia they form a wide geographic zone known as the ‘dust belt’ (Prospero et al. 2002). The large amounts of airborne dust lifted in the atmosphere from these sources can cause severe air pollution, reduced visibility, crop damage, decreased efficiency of solar devices (e.g., solar panels), and others (Zheng 2009). Through the absorption and scattering of solar radiation, aerosols reduce the amount of energy reaching the surface (Tegen 2003; Spyrou et al. 2010, 2013). Also, dust particles enhance the greenhouse effect by absorbing and emitting longwave radiation towards the surface (Heinold et al. 2008). Impacts on human health include respiratory and cardiovascular diseases, as well as several pathogenic conditions due to the microbiota that dust can carry with it (Goudie 2013; Esmaeil et al. 2014; Tong et al. 2017).

The topography of the Arabian Peninsula is, in its largest part, a flat desert-soil terrain with few mountainous areas Southwest and Northwest (Powers et al. 1963). The dust storms over the Arabian Peninsula are frequent evenlts, with the frequency increasing in spring (especially over northern Arabian Peninsula) and summer months (overall Peninsula; Mashat et al. 2008). On a diurnal scale, the peak of dust activity is noticed during daytime due to the generation of high turbulence and pressure gradients by the intense solar heating of the ground. Regarding precipitation, the dry season over the region extends from June to September. During this period the Peninsula is almost rainless. From November to nearly April, the wet season, there is more precipitation, especially over the northern Peninsula (Almazrouia et al. 2012).

The direct and indirect radiative impact of dust particles influence climate from regional to global scale, introducing one of the largest uncertainties in future climate projections (Myhre et al. 2013). By modifying the radiative forcing, dust aerosols affect local climatic parameters and specifically temperature and precipitation (Spyrou et al. 2010; Solomos et al. 2011; Rap et al. 2013; Liu et al. 2014). The radiative impact of mineral dust on the thermodynamic state of the atmosphere has been the subject of a large number of studies. For instance Slingo et al. (2006) measured a reduction in the incoming solar radiation to the surface reaching up to -250 Watt/m² during a severe dust episode in Niger on March 2000. Stanelle et al. (2010) found a reduction up to 4°K in surface temperature during a severe dust storm in West Africa. Chen et al. (2017) examined the thermodynamic impact of desert dust during the summer of 2006, showing, among others, an increase of heating rates up to 0.5°K/day, during a severe dust transport episode. However, the vast majority of these studies were conducted on the Western and Central North Africa. Modeling studies, as well as observations, focusing on Red Sea and the Arabian Peninsula, are far more limited. For example Mohalfi et al. (1998) used the Florida State University Model (FSU) at a 0.5 degree grid resolution to examine the radiative impacts of dust over a 6 day period in June 1979 and found that the dust layer results in warming of the middle troposphere and cooling of the lower troposphere. Islam and Almazroui (2012) again used a coarse resolution setup (50 Km) of the RegCm model to examine the impacts of dust during the wet season. They found an increase in both precipitation and evaporation and thus an acceleration of the water cycle. Due to the relatively coarse resolution of previous studies the local features of the hydrological cycle, and its daily variations, may not be satisfactorily simulated. The more recent work of Prakash et al. (2015) made use of the WRF-Chem model at a fine resolution of 10 Km, but only for a small period of time, examining a dust storm that occurred from 18 to 20 March 2012.

In this work we aim at quantifying the impact of dust produced over Arabian Peninsula, on an inter-annual scale on the precipitation and the mass of water in the atmospheric column. Water vapor is one of the main greenhouse gases in the Earth's atmosphere and its maximum amount is controlled by the temperature (Myhre et al. 2013). Therefore changes incurred by the dust radiative feedback will have an impact on the water mass inside the atmospheric column, a phenomenon that deserves further examination. For this purpose the desert dust cycle for the Arabian Peninsula and its impact on both shortwave and longwave, were simulated using the SKIRON/Dust model with an online dust production scheme (Spyrou et al. 2010; Mesinger et al. 2012; Spyrou et al. 2013; Campos et al. 2017). This area was selected, over the more commonly studied Saharan desert, due to close approximation of two very warm sea bodies: the Red Sea and the Arabian Gulf, which supply the area with evaporated water.

The rest of the manuscript is organized as follows: the description of the experimental set-up and the model characteristics is given in Section 2. The results of the simulations with and without the impact of dust on the radiation fluxes are presented in Section 3 while Section 4 includes the conclusions of this work.

2. Experimental design—Model evaluation

The SKIRON/Dust modeling system has been developed at the University of Athens from the Atmospheric Modelling and Weather Forecasting Group (Kallos et al. 1997; Nickovic et al. 2001; Spyrou et al. 2010) in the framework of several funded research projects (SKIRON, MEDUSE, ADIOS, CIRCE and MARINA Platform). A dust module that simulates the production, transport and removal of desert dust aerosols is directly coupled with the atmospheric model. The dynamical core of the model is based on the ETA concept, which was originally developed by Mesinger et al. (1988) and Janjic (1994). Longwave and shortwave radiative transfer, as well as the feedbacks of aerosols in the radiative transfer, in the SKIRON/Dust model are parameterized using RRTMG (Mlawer et al. 1997; Iacono et al. 2008), a broadband correlated k-distribution radiation model developed at AER, Inc. with support from the U.S. Department of Energy. More details on the specific characteristics of the atmospheric model and the dust module capabilities and performance are provided in Spyrou et al. (2010, 2013) and Mesinger et al. (2012). Therefore an evaluation on the dust module capabilities is not performed here.

The model domain was setup to cover the entire Arabian Peninsula and Eastern Africa with a grid spacing of 0.1^o x 0.1^o. The resolution was chosen in order to be able to represent in a more accurate way the meteorological and topographical features of the area of interest, in order to understand their response to mineral dust. A coarser resolution would not be able to properly describe the mountainous areas of the domain, especially near the coasts of the Red Sea, whose morphology is essential to the findings of this work. Additionally a finer resolution is not needed, since the average effects of dust on the water content over a long period of time (2 years) will be examined and more detailed phenomena (e.g., sea breeze) can be neglected without major errors. On the vertical 38 levels were used stretching from the surface up to 20 Km. ECMWF reanalysis fields with a 0.5^o x 0.5^o resolution were used as initial and boundary conditions.

The optical properties of dust particles (single scattering albedo, asymmetry parameter, refractive index) for the entire radiation spectrum have been determined using the Optical Properties of Aerosols and Clouds (OPAC) software package (Hess et al. 1998). Specifically for the 550 nm spectral window, where the extinction of the incoming solar radiation is most intense, a more realistic single scattering albedo value of 0.95 was selected for transported mineral dust (Kalashnikova et al. 2005; Spyrou et al. 2013).

The simulation period lasts from December 2013 to December 2015 with the first month acting as spin up time in order for the system to build an adequate dust background. The remaining months (January 2014 to December 2015) are referred hereafter as the simulation period. Two sets of simulations were performed: one control simulation where dust particles are inert and don't interact with the solar radiation (from here on named NDE – No Dust Effects) and one simulation where dust direct feedbacks are included in the radiative forcing calculations (from here on named WDE – With Dust Effects).

In order to examine the impact of the dust radiative feedback on humidity, the modeled relative humidity is compared with station values at 6 regional stations (Figure 1). Four of these stations (Doha, Jeddah, Gizan and Kuwait) are located close to the coast of the Red Sea and the Gulf, while two of them (Al-Jouf and Hail) are located inland. This way we will be able to see whether the changes in radiative transfer (incurred by desert dust) have any significant effect on the surface humidity.

Figure 1. Locations of the stations used for model evaluation using surface relative humidity data.

Relative humidity values were extracted from the SYNOP weather reports of the stations presented above, and from the two different model simulations (NDE – WDE) for the entire period. Using this data, the BIAS, the root mean square error (RMSE), the correlation coefficient r, the Normalized Mean Error (NME) and the Normalized Mean Bias (NMB) were computed for the entire simulation period (Table 1), as described in Wilks (2006). In Table 1 the improved scores are marked with bold.

Table 1. Statistical parameters: BIAS, RMSE, correlation coefficient, NME and NMB for the entire period of simulations (January 2014–December 2015) with dust effects (WDE) and without dust effects (NDE) for relative humidity, for 6 different surface stations (Doha, Jeddah, Gizan, Al-Jouf, Kuwait, Hail).

Station		BIAS	RMSE	Correlation	NME (%)	NMB (%)
Doha	WDE	8.43	15.85	0.819	23.77	14.25
	NDE	10.53	17.47	0.804	25.97	18.56
Jeddah	WDE	−0.06	14.41	0.681	18.7	−8.8
	NDE	0.21	15.07	0.677	18.8	−9.7
Gizan	WDE	0.4	10.39	0.67	12.69	1.41
	NDE	1.53	11	0.681	13.44	3.04
Al-Jouf	WDE	−4.75	10.74	0.858	23.65	−14.43
	NDE	−4.96	10.83	0.858	23.86	−14.96
Kuwait	WDE	1.09	12.18	0.865	25.82	−12.79
	NDE	1.24	12.37	0.863	26.58	−13.91
Hail	WDE	−13.2	17.86	0.838	33.93	−32.58
	NDE	−13.6	18.01	0.841	35.18	−32.59

In all cases including the radiative feedback of dust (WDE) improves the simulation of relative humidity at all stations. The BIAS and RMSE are reduced in all stations, with Doha showing the biggest improvement by almost 25%, whereas Al-Jouf has the smallest impact approximately 4%. Accordingly the NME and NMB have the same behavior with improvements in all locations. In general stations that are located close to the coast are mostly affected by changes in the relative humidity due to dust radiative impact. Coastal areas in the Arabian Peninsula exhibit high relative humidity values, due to the proximity to the sea and the increased surface temperatures all year long. Also the constant presence of dust clouds over the entire area means that the “greenhouse effect” of natural particles is significant, therefore altering the amount of water vapor in the atmosphere (as explained in more detail in Section 3).

In contrast the two inland stations (Al-Jouf and Hail) show the least improvement, as these are very dry areas away from water sources. Therefore the feedback of dust is reduced, as there is not enough vapor to make a significant difference through the radiative feedback. As far as the correlation coefficient is concerned a relatively small improvement was observed in all coastal stations. This is to be expected since dust feedbacks are expected to affect the total amount of water in the atmosphere rather than its diurnal and seasonal variability. In order for the correlation coefficient to be significantly affected, dust particles need to change the way relative humidity is modified (increasing or decreasing) during the day, which cannot be caused in such a degree by the radiative feedback.

3. Estimating the impact on the simulated water budget

Desert dust cycle (production, transport and deposition) is a complicated process, but essential in order to properly understand and quantify the feedbacks on radiative transfer and therefore the water content of the area of interest. As seen by the average dust load in Figure 2, significant dust amounts are simulated all over the Arabian Peninsula (reaching even 1 g/m²) and the highest amounts are found over the Red Sea. The dust load distribution over the area originates from the WDE run, since it is considered the more realistic, as it includes the dust radiative feedbacks. Such high dust loads over the Red Sea occur due to the trapping of dust along the topographic barriers in this area which is affected by both Saharan and Arabian sources. High dust loads are also found over Iran, which are mostly attributed to the local sources there. The difference in dust production between the two simulations is not discussed in detail since it is almost negligible.

Figure 2. Dust load as average during the entire simulation period (January 2014–December 2015), from the WDE simulation.

Analysis of the two model simulations (NDE and WDE) is used to identify the impact of dust on atmospheric processes. First the differences in incoming Shortwave (Figure 3a) and Longwave (Figure 3b) radiation have been plotted. As expected, the most profound feedbacks are found in areas with the highest particle concentrations. Dust particles scatter and absorb shortwave radiation (Myhre et al. 2013; Kushta et al. 2014) leading to a maximum decrease of approximately -60 Watt/m² at the coast of the Red Sea and Iran. In contrast, as seen in Figure 3b, dust particles emit longwave radiation increasing the surface flux, up to approximately +50 Watt/m², also known as the “greenhouse” effect of dust particles (Myhre et al. 2013; Spyrou et al. 2013). The feedbacks of dust particles on the radiative balance of the atmosphere are significant (compared to other greenhouse gases like CO₂) and cause changes in the surface temperature (Figure 4a) and the water load of the atmospheric column (vertically integrated water mass – Figure 4b).

Figure 3. Averaged difference (WDE–NDE) in the incoming shortwave (a) and longwave (b) radiation at the surface.

Figure 4. Averaged difference of the model simulations (WDE – NDE) in the temperature at 2 m (a), and column integrated water load (b).

The surface temperature is increased throughout the whole domain due to the emission of longwave radiation from the dust cloud directly above. This change can even reach on average +0.8° K. In enclosed areas, like the Red Sea, longwave radiation is “trapped” between the sea surface and the dust layer (Spyrou et al. 2013). The largest differences in temperature were found over water mainly in 2 areas: the Red Sea and the Gulf of Oman. The reason for that is twofold: first of all both sea bodies are surrounded by mountainous areas, which leads to the “entrapment” of desert particles and increases the “dust greenhouse effect,” in contrast to the areas with peak shortwave and longwave radiation values (eastern part of the domain). Additionally the warm waters of both the Red Sea and the Gulf of Oman emit higher amounts of longwave radiation (as seen also in Spyrou et al. 2013) than the mountains of Iran, therefore amplifying this phenomenon. This “greenhouse effect” increases the surface temperature which in turn increases the evaporation rate of these areas. At the same time dust particles enhance the mid tropospheric warming, thus increasing atmospheric stability and reducing convection with the total effect of decreasing precipitation (Brooks 2000). Accordingly the increase in the water load of the atmospheric column follows the patterns of temperature change. This is due to the fact that increased temperatures will lead to enhanced evaporation at the sea surface. Additionally the precipitation reduction due to the increase in atmospheric stability and convection reduction (as discussed above) leads to the increase of the water mass inside the atmosphere. This feedback increases the daily average integrated amount of water in the atmospheric column (water load) by up to 1 g/m². This extra mass of water is transported further towards the Arabian Sea, following the typical flow patterns that dominate the area. A significant increase in water mass is found on the eastern boundary of the domain (around 27°N), that can also be attributed to local dust feedbacks, but one should be cautious since it is right on the model bounds and thus cannot be considered reliable information.

In order to have a clearer understanding of these effects, changes in the temperature and water mass need to be examined inside the atmospheric column. To this end the average dust distribution and the feedbacks on the temperature, rain ratio and water mass (between the two simulations) were plotted along the 15°N pathway. This area was selected due to the significant changes that were observed in temperature and water mass in Figures 4a and b.

It is clear that high dust particle concentrations are trapped close to the surface inside the Red Sea due to the specific topographic variation (Figure 5a). Also there is substantial dust load over the Arabian Gulf extending even up to 6 Km vertically all year long. The presence of these particles forces radiative re-distribution inside the atmospheric column and the surface. More specifically the daily average temperature on the surface layer is increased more than ∼0.6 °K due to the absorption of solar radiation and the re-emission of longwave radiation towards the surface as discussed above. In addition there is low-tropospheric cooling due to the emission of longwave radiation and the reduction in solar radiation from the dust cloud above. In the mid-troposphere dust particles increase the temperature of the layer by absorbing shortwave radiation while at the same time the reflected incoming solar radiation reduces the temperature on the layer above the dust cloud (Figure 5b). All these processes are well known and recorded in several other publications as well (Heinold et al. 2008; Solomos et al. 2011; Spyrou et al. 2013; Kushta et al. 2014; Saeed et al. 2014; Chen et al. 2017). In addition the averaged rain ratio is reduced in the atmospheric column, especially close to the surface (Figure 5c), reflecting the expected precipitation reduction due to the increase in atmospheric stability and convection reduction. It is interesting to note here that the increased water vapor in the atmospheric column does not cause an increase in rain. This can be attributed to the fact that the precipitation reduction, due to the increase in atmospheric stability and convection reduction, is more potent in modifying rain than the resulted water vapor increase, in the area of interest.

Figure 5. Averaged vertical distribution of (a) dust concentration from the WDE simulation, (b) differences in vertical temperature, (c) differences in vertical rain ratio, and (d) differences in water mass between the model simulations (WDE–NDE) along the 15°N pathway for the entire simulation period (January 2014–December 2015).

In the case studied here the most important aspect is the increase of the surface temperature which in turn enhances the evaporation from the sea (Red Sea). Adding this to the reduced precipitation rate expected from dust (due to the mid-tropospheric warming which increases atmospheric stability) it can be seen that the increase in water mass inside the atmospheric column can reach a daily average of 0.5 g per kg of dry air (Figure 5d). The extra amount of water is then transported over the coastal and inland areas of Saudi Arabia, further enhancing the greenhouse effect there, but also making daily life harder for the local population, since the surface humidity is increased and the atmosphere becomes even more suffocating. A very small reduction in water mass observed near the surface at the Red Sea can be explained by the increase of the boundary layer depth due to heating, thus expanding the mixing layer. This is also a mechanism that transports the added water mass at greater heights.

4. Conclusions

The presence of dust particles in the atmosphere has significant regional impacts on the radiative transfer and energy distribution of the atmospheric column. These interactions are not one-way; there are feedbacks that are critical for both regional-scale and climatological phenomena. In this study an attempt is made to quantify the direct feedback of natural desert dust particles to the water content in the atmospheric column in the Arabian Peninsula. The reason it was selected over the Saharan desert is the close proximity of two very warm sea bodies: the Red Sea and the Arabian Gulf. Two sets of simulations were performed: one where the dust particles are inert and the other where dust particles are active and interact with radiation. Through scattering and absorption natural aerosols reduce the incoming solar radiation at the surface, by up to -60 Watt/m² on average, but at the same time they emit longwave radiation towards the surface, increasing the incoming amount by up to +50 Watt/m². Also the layer containing the highest concentrations of dust is prone to increased temperature due to absorption. These combined effects lead to an increase in the surface temperature, with maximum values reaching +0.8°K, and as a consequence the evaporation from the large sea bodies surrounding the Arabian Peninsula. Therefore the model showed an increase of the water load in the atmospheric column in almost all the computational domain, reaching even +1gr/m² in the Red Sea area. Through statistical evaluation it became clear that including these effects leads to significant improvements in simulation results, especially in coastal stations, where the mass of water plays a more dominant role.

Additionally the precipitation is reduced in the area of interest, as shown by the reduction in the rain ratio between the two runs, which agrees with previous studies which have shown that the direct feedback of dust particles suppresses the precipitation rate. These phenomena, when combined, lead to an increase in the mass of water in the atmospheric column that can reach a maximum daily average of +0.5 g per kg of dry air. Since water vapor is a very potent greenhouse gas, the added mass in the atmospheric column will lead to a further increase in surface temperatures, especially during the nighttime. Basically through this process dust enhances its own greenhouse effect and further increases the surface temperature and humidity, making the life difficult for people living in an already harsh desert climate. It is important to note here that the indirect effect of dust particles was left out on purpose, in order to focus and quantify the direct radiative feedback of dust particles on the water content of the atmosphere. By including the feedback of dust particles the issue becomes more complex: dust acts as CCN, which can lead, either to precipitation reduction, therefore increasing the amounts of water in the atmosphere, or precipitation increase which will remove water from the atmospheric column. Therefore additional work is needed in order to quantify the indirect effects, which will be assessed in a future work.

Acknowledgments

This work was supported by computational time granted from the Greek Research & Technology Network (GRNET) in the National HPC facility – ARIS – under project ID-PA001011-metocean_db.