Impacts of Soot, Ash, Sand, and Haze on Snow Albedo in Sierra Nevada, Spain

Abstract

Snow covers are greatly affected by particles deposited on their surface. In this work, an experimental campaign was carried out in the Sierra Nevada (Granada, Spain). The optical effect of different contaminating particles on the snow covers was measured using a field spectroradiometric system composed of three upwelling spectroradiometers and three downwelling ones. Sand collected from a Mediterranean beach (Spain), ash collected from the La Palma volcano eruption, haze collected from an event that occurred in Spain, and soot collected from a diesel vehicle were employed for contaminating the snow. Soot, ash, and sand were analysed with X-ray diffraction to obtain their mineralogical composition or their structural characteristics, whereas haze's mineralogical composition was obtained from the literature. From this information, the refractive index of each material was weigh-averaged, considering the refractive indices of their components. After measurements, snow samples were filtered and weighted to evaluate the particle concentrations in the snow. Previous contamination with soot was observed due to the existence of a nearby road. Snow albedo was calculated with the OptiPar model. The experimental and modelled results show that contaminating with sand decreases the snow albedo in the visible range whereas it increases the albedo in the infrared range. However, the rest of the materials lead to a decrease in the albedo in the whole spectrum, although with different intensities depending on the wavelength range.

1. Introduction

Particles can be generated from combustion processes such as those vehicles, aircrafts, and boilers or from forest burning [1]. In these processes, the most common particles generated is soot. In addition, due to strong haze events and gusts of wind, sand and haze particles from soils can be moved and suspended in the atmosphere [2]. Furthermore, volcanoes constitute a natural source of ash which can also remain suspended in the atmosphere [3]. The residence time of particles in the atmosphere depends on their size. Very small and very large particles have low residence times since they are controlled by sedimentation processes and diffusive losses, respectively [4,5].

Some strong haze events took place last spring (2022) in Spain, where haze particles were blown from the Saharan dessert, as can be seen from satellite images of Aerosol Optical Depth and haze transportation in Figure 1a [6], as well as ash cloud transportation from La Palma eruption which started on September 2021 Figure 1b [7].

These particles from anthropogenic (soot) and natural (ash, haze, and sand) sources have an environmental effect on global heating [10], as well as on polluting surfaces as far as their suspension residence time is large enough [11,12]. An example of a house facade contaminated through a spring haze event in the area of Almería, Spain, is shown in Figure 2a, where the first floor remains coloured, and the ground floor has already been cleaned up. However, the surfaces that are mostly affected by particles deposited on their surface are snow covers [13], due to their high reflectance, and are the ones that contribute most to cryological climate change [14]. Additionally, an example of the soiling of the snow surface with haze in the south of the Iberian Peninsula (Sierra Nevada, Granada) is shown in Figure 2b.

Considering the environmental vulnerability of this region, experimental measurements have been carried out in the Sierra Nevada, employing different particles (soot, ash, haze, and sand) for snow artificial contamination to evaluate the effect of the surface albedo modifications. The reasons why these materials are selected are the following: soot is the main component of particles emitted from vehicles which often circulate close to ski resorts. Ash can be transported from eruptions located thousands of kilometres away. Sand can be lifted from bare soils and moved through strong winds or storms. Saharian haze is a fine mineral dust that often blows toward south-Europe regions. All these materials can be deposited in Sierra Nevada snow areas affecting the radiative balance and thus the hydrological cycles.

Some experimental campaigns and simulations about the effect of aerosols deposited onto snow surfaces all over the world have been reported in the literature. Cereceda-Balic et al. [15] measured the reduction in broadband snow albedo due to local emissions of soot from high vehicular traffic in the surroundings of a road in the Chilean Andes. Further studies about the effect of soot deposition from vehicles have shown that this reduction is different depending on the spectral range, such as those by González-Correa et al. [16] in Sierra Nevada, and Lapuerta et al. [17] in Cotos Port, Madrid (both in Spain). Skiles and Painter [18] measured daily black carbon and dust, and snow optical parameters in the Rocky Mountains, Colorado, where reductions of snow albedo down to 0.3 were observed as a consequence of dust deposition. They also observed different reductions depending on the spectral range. Niu et al. observed reductions in the broadband albedo of aged snow in Mount Yulong from 2% to 10% derived from the deposition of black/organic carbon and dust [19]. Dang et al. [20] computed the reduction of snow spectral albedo by black carbon and mineral dust using a radiative transfer model. Simulations with Snow, Ice, and Aerosol Radiative (SNICAR) for different Saharan dust sizes and with ash from the Eyjafjallajökull volcano were presented by Flanner et al. [13]. Painter et al. [21] used remote sensing, and processed results with the MODIS Dust Radiative Forcing in Snow (MODDRFS) model, retrieving surface radiative forcing by dust and carbonaceous particles in snow and observing snow spectral albedo reductions mainly in the visible range in Upper Colorado River Basin and Hindu Kush-Himalaya. Constantin et al. [22] measured and simulated with SNICAR a large impact on snow albedo in the Argentinian Andes as a consequence of ash deposition from volcanic eruptions. In most of these studies, the effects on albedo depended not only on the spectral range, but also on the composition, size, shape, and crystalline structure of the deposited particles [23].

Different materials lead to different spectral effects on snow albedo and thus to different radiative forcings. However, they have not been compared under similar conditions. The main novelty of this work lies in the comparison of the effect of various contaminating materials (to which the Iberian Peninsula is frequently exposed) on the snow albedo, for which in-field measurements were made at the same site. The compositional characterization of these materials, together with the modelling of these effects has been revealed as helpful to interpret such effects.

2. Methodology

Experimental testing was carried out in different stages, as shown in Figure 3. First, the materials for snow surface contamination were selected (light brown box in Figure 3). These materials were obtained from different sources: (1) soot was collected from experimental tests from an automotive diesel engine in the University of Castilla-La Mancha (UCLM) installations [24]; (2) ash was collected on 26 September 2021 by members of Geomorphology, Territory, and Landscape in Volcanic Regions (GEOVOL) research group at UCLM from Tenecuita centre, 14 km away from the eruption of the volcano Cumbre Vieja in La Palma island, in Canary Islands, Spain, occurred on 19 September 2021 [3]; (3) sand was collected from a Mediterranean beach and d) haze particles arising from haze events held in Spain in March 2022 was collected on 15 March 2022 from depositions in the area of UCLM campus in Ciudad Real [25].

Materials collected were analysed (purple box in Figure 3). For soot, Transmission Electron Microscopy (TEM) analysis was made to obtain their morphology and fractal parameters [26,27,28]. For the rest of the materials, X-ray diffraction was performed to characterize their mineralogical components [29].

Afterward, field measurements were carried out (green box in Figure 3). A field-hyperspectral spectroradiometer system designed by Antares Instrumentation S.L. employing Avantes instruments was used to perform spectral radiance of natural and contaminated snow surface and solar spectral irradiance measurements. The equipment is composed of two groups of three spectroradiometers for each wavelength range: UV-VIS (300–1100 nm), NIR (900–1700 nm), and IR (1700–2500 nm). These spectroradiometers are connected to two optical fibres supported by an extended tripod with two cosine-corrector optical detectors placed in an upwelling and downwelling direction for irradiance and radiance measurements, respectively. More details of this equipment and setting up can be found in Lapuerta et al. [17] and González-Correa et al. [16].

The experimental campaign took place on two different dates: 16 December 2021 and 8 April 2022, in the proximities of Sierra Nevada ski resort, Granada, Spain. Measurements were carried out at two different sites. The first one was carried out close to a ski slope (37.09° N, 3.39° W, 2500 m.a.s.l.) and the second one on a helicopter landing pad (37.05° N, 3.23° W, 2500 m.a.s.l.) (see Figure 4). Measurements were made on “clean” and artificially contaminated snow. “Clean” contamination refers to surfaces exposed to continuous contamination with soot particles from a nearby road from the last snowfall that occurred on 9 December 2021 and on 6 April 2022 [30] and also with haze from haze events that happened on 25 March 2022 [6] and artificially contaminated refers to the manual deposition with a sieve of collected particles on the analysed area.

Although homogeneity was searched during artificial contamination, wind gusts, the size distribution of the materials, and other additional external factors generated non-homogeneous surface contamination. This issue has been addressed in a few experiments on artificially contaminated snow [31], arising new concerns on the representativity of the deposition process onto the natural snowpack, such as perturbance of the snowpack physical properties, scavenging of small particles into deeper layers due to an airflow input for deposition, or melting acceleration caused by a temperature increase when a chamber for deposition is used, among others [32,33]. For this reason, the mass concentration obtained from laboratory analysis was corrected and weighted considering the contamination observed in the photographs (see Figure 5). The vision area from the downwelling cosine receptor is a circle of radius 1.5 m, which comprises a vision factor of 95% considering that the upwelling detector height was 35 cm with respect to the snow surface. In Figure 5, contaminated areas with three different darkness levels in the analysed zone are shown, to exemplify the different concentrations of materials artificially deposited on snow surfaces. Each concentric circle corresponds to a visual area, and the average mass concentration is calculated by radially weighting the darkened areas.

Field measurements of snow grain size, cover depth, cloudiness, rugosity, surface inclination, and solar zenith angle were taken (see

Table 1. Parameters for snow contamination characterisation (previously contaminated refers to “clean” snow).

Date	Snow Surface Previously Contaminated with:	Snow Surface Artificially Contaminated with:	Corrected Zenith Angle (°)	Concentration (mg/kg)	Snow Density (kg m−3)	Snow Grain Radius (µm)	Depth (cm)
16 December 2021	Soot	-	70.3	0.9	250	309.8 ± 53.4	Semi-infinite
	-	Soot		0.8
	Soot	-	61.0	0.5
	-	Ash		90
	Soot	-	65.0	0.3
	-	Sand		100
8 April 2022	Soot + Haze	-	48.6	0.2 Soot + 25 Haze	404.78	270.3 ± 40.6	11.46
	-	Haze		25

). Photographs of the snow size were made on a calibrated card for a size distribution characterization with ImageJ (National Institutes of Health, Bethesda, MD, USA.) [34] (see Figure 6). No cloudiness was appreciated on the dates of the campaign (as observed in Figure 4) and rugosity of the snow surfaces and humidity were considered neglectable. The solar zenith angle was corrected using the surface inclination. In addition, snow samples are used in laboratory analysis for snow density and mass concentration (mg/kg) determination. The procedure to obtain the particle mass concentration is explained in González-Correa et al. [16].

With all these experimental parameters, simulations were made with OptiPar [17] (blue box in Figure 3), which contains a radiative transfer model for the determination of optical properties of some materials and ice particles, as well as snow surfaces contaminated with deposited light-absorbing particles (LAPs). Among the various options available in OptiPar, the following methods and parameters were selected. Snow albedo of a finite (considering the underground albedo of the soil) or semi-infinite cover was determined through the model from Wiscombe and Warren [35] and with the refractive index of Warren and Wiscombe proposed in 2008 [36]. For snow albedo characterization, snow density, depth, grain, and the corrected zenith angle were taken for the simulations. Particles of the different polluting materials were simulated with the Mie theory or the Rayleigh-Debye-Gans approximation (using the correction for multiple scattering proposed by Mountain and Mullholland [37]) depending on the size of the deposited particles were used.

After the materials and the snow are optically characterized separately, the snow surface contamination was simulated using both the concentration of previously deposited soot (and eventually haze) and the artificial contamination with soot, ash, sand, and haze concentrations.

3. Particle Characteristics

Snow and materials characteristics were summarized in Table 1 and

Table 2. Materials average grain radius and densities (previously deposited refers to “clean” snow).

Material		Average Grain Radius (nm)	Densities (kg m−3)
Soot		97	1850
Ash		450	2600
Sand		10,000	1500
Haze	Previously deposited	1500	2500
	Artificially deposited	400

. From the observed particle size distributions, the following average particle radius was obtained for each material: 300 nm for ash, 10 µm for sand, 1.5 µm for haze deposited in Sierra Nevada, and 400 nm for haze deposited in Ciudad Real.

Soot fractal parameters were considered to simulate the morphology of soot particles. In both cases, “clean” and artificial contamination for all the simulations, agglomerates composed of 80 primary particles of 12.5 primary particle radius (nm) each and with a fractal dimension of 1.85 was used. These were the reference conditions in a previous study with a diesel engine [38]. From these parameters, the average radius of gyration was obtained for soot agglomerates, using the power law relationship [26], resulting in 97 nm.

Densities of contaminating materials were taken from the literature. For haze, a density value of 2500 kg m⁻³ was used as proposed by Dang et al. [20]. For sand, a value of 1500 kg m⁻³ was used, which is within the range proposed by Allan et al. [39]. For ash, a value of 2600 kg m⁻³ proposed by Flanner et al. [13], was used. For soot, 1850 kg m⁻³ inherent density was used based on the model proposed by Belenkov [40]. For the determination of the particle mass for each material, a spherical shape was assumed except in the case of soot agglomerates, for which it was obtained as the product of a number of primary particles and the mass of the mean primary particle.

Sokolil and Toon [41] showed that the mineral composition of the materials must be incorporated into the radiative models to estimate their optical properties. For this reason, an effective medium model [40] was employed for the determination of the dielectric constant (ε_particle). From the dielectric constant of all the minerals of the material (ε_i), their volumetric fraction (f_vi), and the dielectric constant of the medium (ε_m), the material dielectric constant was determined as follows:

$$\frac{{\mathsf{\epsilon }}_{\mathit{p}\mathit{a}\mathit{r}\mathit{t}\mathit{i}\mathit{c}\mathit{l}\mathit{e}}\mathsf{-}{\mathsf{\epsilon }}_{\mathit{m}\mathit{e}\mathit{d}\mathit{i}\mathit{u}\mathit{m}}}{{\mathsf{\epsilon }}_{\mathit{p}\mathit{a}\mathit{r}\mathit{t}\mathit{i}\mathit{c}\mathit{l}\mathit{e}}\mathsf{+}\mathbf{2}{\mathsf{\epsilon }}_{\mathit{m}\mathit{e}\mathit{d}\mathit{i}\mathit{u}\mathit{m}}}\mathbf{=}{\displaystyle \sum }_{\mathit{i}\mathbf{=}\mathbf{1}}^{\mathit{N}}{\mathit{f}}_{{\mathit{v}}_{\mathit{i}}}\frac{{\mathsf{\epsilon }}_{\mathit{i}}\mathbf{-}{\mathsf{\epsilon }}_{\mathit{m}\mathit{e}\mathit{d}\mathit{i}\mathit{u}\mathit{m}}}{{\mathsf{\epsilon }}_{\mathit{i}}\mathbf{+}\mathbf{2}{\mathsf{\epsilon }}_{\mathit{m}\mathit{e}\mathit{d}\mathit{i}\mathit{u}\mathit{m}}}$$

(1)

This general equation is particularized assuming the hypothesis that the medium is a vacuum as proposed by Lorentz-Lorenz [42], and, therefore, the dielectric constant of the medium is unity.

$$\frac{{\mathsf{\epsilon }}_{\mathit{p}\mathit{a}\mathit{r}\mathit{t}\mathit{i}\mathit{c}\mathit{l}\mathit{e}}\mathbf{-}\mathbf{1}}{{\mathsf{\epsilon }}_{\mathit{p}\mathit{a}\mathit{r}\mathit{t}\mathit{i}\mathit{c}\mathit{l}\mathit{e}}\mathbf{+}\mathbf{2}}\mathbf{=}{\displaystyle \sum }_{\mathit{i}\mathbf{=}\mathbf{1}}^{\mathit{N}}{\mathit{f}}_{{\mathit{v}}_{\mathit{i}}}\frac{{\mathsf{\epsilon }}_{\mathit{i}}\mathbf{-}\mathbf{1}}{{\mathsf{\epsilon }}_{\mathit{i}}\mathbf{+}\mathbf{2}}$$

(2)

Therefore, ash, sand, and haze were analysed with X-ray diffraction to obtain their mineralogical composition by volume (see Figure 7). With this information, the dielectric constants of the particles were determined and, hence, their refractive index for optical properties calculations was obtained (see Figure 8). In the case of sand, 7.4% content of acetoguanamine was detected with X-Ray diffraction. However, this mineral was not used for sand refractive index determination because no absorption coefficient was found for this material. Dielectric constants of each mineral were obtained from a database [43] and from the references shown in

Table 3. Source for the mineral’s dielectric constants.

Aerosol	Mineral	Reference
Ash	Olivine	Fabian et al. 2001 [47]
	Cordierite	Classic gems [48]
	Franklinite
	Albite
Sand	Dolomite	Querry 1987 [49]
	Calcium carbonate	Posch et al. 2007 [50]
	Silicon oxide	Rodriguez de Marcos et al. [51]
Haze	Hematite	Triaud 2005 [41]
	Calcite	Posch et al. 2007 [50]
	Quartz	Calingaert 1936 [52]
	Montmorillonite	Querry 1987 [49]
	Kaolinite
	Illite

. Soot refractive index was characterized by the correlation proposed by Chang and Charalampompoulos [44] based on experimental data obtained at 300 K.

Refractive indices obtained by the Effective Medium were mostly in good agreement with the literature. The real and imaginary parts of the sand refractive index are of the magnitude of order as the values of Koepke et al. [53], around 1.5 and 0.001, respectively. For ash, the real part of the refractive index obtained was about 1.6 in accordance with Ball et al. [54] and Redd et al. [55] from Etna and Eyjafjallajökull volcanos data, respectively. For the imaginary part, also the data for Ball et al. [54] agrees with the Effective Medium results. For haze, the value of refractive index agrees with the Patterson et al. [56] of pure Saharan haze and Volz et al. [57] values of rainout haze aerosol collected in Germany. In addition, Balkanski et al. [58] determined that mineral haze absorbed in short wavelengths spectra is based mainly on an iron oxide formed as Hematite, underlying the necessity of considering this mineral component in a haze.

However, ash, sand, and haze refractive indices were corrected since (1) the dielectric constants of the minerals that contain the materials were not obtained from the exact conditions, (2) absorption coefficients were often missing and therefore they were roughly estimated, and (3) the observed characteristics with the experimental snow contamination (see Results and Discussion section) showed different behaviour as the predicted with the refractive indices weigh-averaged. Considering this, for sand, the refractive index proposed by Longtin et al. [46] was employed, with a higher real part (around 1.65) and more differences in the spectral range for the imaginary part from the results obtained from the Effective Medium. For ash, values from Deguine 2020 [45] considered for the Etan volcano were selected. Before 600 nm a constant value from the initial refractive at this wavelength was established since no data beneath this wavelength was available. In addition, a correction of the imaginary part from 1120 nm was made to a value of 0.5, because differences were seen in this range. This was also based on the low imaginary part of the minerals contained in the ash obtained with the weigh-averaging. For haze, the imaginary part of the refractive index was raised up and established at 0.53 from 1000 nm because the weight averaging decreased too much for the imaginary part. The real part was not modified.

4. Results and Discussion

With the conditions and parameters specified above, the optical properties of soot, ash, sand, and haze aerosols were determined. The MAC (Mass Absorption Cross Section), MSC (Mass Scattering Cross Section), and SSA (Single Scattering Albedo) are represented in Figure 9. As expected, the three optical results depend mainly on the refractive index, and the scattering efficiency is the dominant effect on characterizing the albedo of the particles. Particularly, for each material, different effects are observed. Soot SSA is in agreement with values obtained by Yon et al. [59] and MAC values agree with those simulated by Kelesidis et al. [60]. Ash SSA also is consistent with values obtained from Piontek et al. [61] if their results are extrapolated to lower wavelengths. Sand SSA has a similar behaviour as the values obtained by Takemura et al. [62] for soil dust, consisting of an increase of albedo as the wavelength increases. However, SSA for haze is lower than most of the values found in the literature. This can be explained by the high variability of the types of haze evaluated.

From the optical properties of the particles used for artificial contamination, albedo results modelled with OptiPar were compared to the experimental data (see Figure 10). “Clean” albedo is represented with blue lines in all cases as a reference before contamination. Artificial contamination is represented with different colours depending on the contaminating material (black for soot, grey for ash, orange for haze, and brown for sand). Differences observed in the “clean” albedos are due to different zenith angles, previous deposition of soot, and hour of the day. Modelled results approximately fit the experimental data observed spectrally. However, values below 400 nm are not reliable since the correlation proposed by Chang and Charalampompoulos [44] slightly underestimates the soot refractive index. Discrepancies can also be observed in the range 1700–2500 nm, where both solar irradiation and snow radiation are very weak (close to zero), and the ratio between them becomes very noisy and erratic, with any minor disturbance (e.g., any minor deviation of the detectors from parallelism) being highly magnified. However, the contribution of this discrepancy to the broadband albedo is minor because this is obtained as the integral of spectral irradiation divided by the integral of snow radiation [17], and both are very low from 1700 nm onwards.

Figure 11 shows albedo reductions from “clean” to a contaminated snow surface. It can be observed that there is a higher albedo reduction in the visible range rather than in the NIR range, as expected. Considering individually each material contamination, soot is by far the one that reduces the most snow albedo, due to its high absorbing light capacity: only 0.7 ppm concentration reduces the albedo in the same order as in the rest cases with much higher concentrations (see Table 1). In absolute values, ash leads to high albedo reductions due to the high concentration of deposited material. The albedo reduction, in this case, reaches 25% in the UV-VIS range (300–800 nm), 13% in the near-infrared (NIR: 800–1400 nm) and it is insensitive in the middle infrared (MIR: 1400–2500 nm). Ash refractive index obtained for this last range has high uncertainty which explains that modelled results do not fit the experimental ones. Artificial soot contamination decreases the albedo by around 14% in the UV-VIS range, whereas in the NIR the decrease is only 7% and it becomes insensitive in the MIR range. Haze has a similar behaviour as soot and ash, but the decrease in the albedo remains more uniform. Reductions on the UV-VIS are reduced from 27 to 12% (as wavelength increases), but in the NIR range, a 9% reduction remains uniform and again becomes insensitive in the MIR range. As in the case of ash, modelled results are affected by the uncertainty of the imaginary part. On the contrary, sand albedo reductions are only observed in the visible range (from 10 to 0%), while from 800 nm upwards the tendency changes and the albedo starts to increase (up to 3%) due to the high reflectivity of sand (as observed in the sand SSA in Figure 9).

Figure 12 shows the broadband snow albedo, calculated as described in Lapuerta et al. [17], of the “clean” and artificially contaminated snow surfaces, obtained from both the experimental and modelled results. Modelled results fit well the experimental results of both “clean” and contaminated snow broadband albedo. These results show that sand has the opposite behaviour to the other materials, where the spectral albedo moves from reductions (UV-VIS) to increases (NIR and MIR). This behaviour is self-compensated, leading to similar values of broadband albedo. For all other materials, snow albedo reductions remain (with variable intensity along the range) leading to significant broadband albedo reductions in all cases.

5. Conclusions

Particles deposited in snow surfaces decrease the snow albedo differently depending on their composition and concentration. Soot has the highest potential to reduce snow albedo for a given concentration of deposited particles. It decreases substantially the snow albedo in the UV-VIS range (300–800 nm), but such reduction drops down in near-infrared (NIR: 800–1400 nm) becoming insensitive in the middle infrared range (MIR: 1400–2500 nm). Ash has lower albedo reduction potential than soot, although similar distributed along the wavelength range. However, in this case, the uncertainty of the imaginary part of the refractive index makes these results unreliable. Haze has a similar qualitative effect as soot: they both show gradual albedo reduction in the UV-VIS, although soot leads to a sharper reduction in the NIR range. The opposite of this effect can be observed with sand since the observed reduction in the UV-VIS range turns into an increase in the snow albedo in the NIR and MIR range because of the high single scattering albedo of this material. However, this behaviour for sand is self-compensated leading to invariable values of broadband albedo.

Modelling reproduces qualitatively spectral tendencies, but they are affected by the hour, site, day of the year, and snow metamorphism. However, despite spectral inaccuracies, broadband albedo simulations agree with experimental results. In general, the refractive index of the materials (soot, ash, sand, and haze) was proved to be the most important parameter for simulations of the optical effects of different contaminating materials. Therefore, a precise determination of the refractive index has been revealed as essential for an accurate simulation.