Analysis of the Observation Results for Preferentially Oriented Particles in High-Level Clouds According to the EARLINET Lidar Network and MODIS Data

Abstract

This analysis presents the results of observation episodes for high-level clouds consisting of ice crystals with different orientations in space, using the EARLINET lidar network and MODIS data over Europe from 2015 to 2022. The technique used involves identifying specular reflecting layers through ground-based laser soundings of the atmosphere and finding synchronous images from space, followed by the retrieval of cirrus cloud parameters. The study considers several properties of high-level clouds, including reflection ratio (ρ), effective emissivity (ε), optical thickness, effective particle radius, water path, and top height, according to MODIS data. The results of retrieving these properties for cirrus clouds with different orientations of ice crystals over individual EARLINET lidar stations and over Europe as a whole are discussed. The study indicates that high-level clouds with ρ ≥ 0.15 and ε ≥ 0.4 definitely contain specular reflecting layers.

1. Introduction

High-level clouds (HLCs) are one of the main factors of uncertainty in predicting changes in the Earth’s climate [1,2,3]. On the one hand, cirrus clouds can scatter incoming solar radiation or reflect it back into outer space [4,5,6], preventing the underlying surface from heating up. On the other hand, HLCs can retain outgoing long-wave radiation in the Earth’s troposphere, thereby enhancing the greenhouse effect [7,8,9]. Considering the high frequency of cirrus clouds, which is 50% at mid-latitudes, and the horizontal extent of their fields, reaching several hundred kilometers, the extent of the existing uncertainty in predicting climate change with an underestimation of the contribution of HLCs becomes clear [10].

According to modern theories, the degree of influence of HLCs on radiative transfer depends on the spatial orientation of the ice crystals of which they are composed [11,12,13]. If the flat faces of these particles are preferentially oriented in the horizontal plane, then the effect of mirror reflection is observed. With random orientations of the surfaces of ice crystals relative to each other, there is multiple re-reflection (scattering) of the radiation coming to them from space and/or from the underlying surface. The effect of specular reflection of HLCs was first described in [14] using the results of laser polarization sounding. In the presence of layers of ice crystals preferentially oriented in the horizontal plane, strong backscattering and weak depolarization of scattered radiation are observed in cirrus clouds. Later, it was found that the specular reflecting layers in the HLCs can have a vertical extent of several hundred meters, and their lifetime can vary from tens of minutes to several hours [15,16,17]. In this case, the main reason for the alignment of crystal particles in the horizontal plane is considered to be gravitational settling with a minimal effect of turbulence [18].

The primary method for obtaining information about the spatial orientation of cloud particles is through ground-based lidar systems [19,20,21,22,23,24]. However, this approach has a significant disadvantage in that it only provides localized measurements, making it impossible to estimate the horizontal dimensions of the specular reflective layers in the HLCs. Recently, attempts have been made to identify features and conditions that facilitate the determination of the spatial orientation of crystal particles in cirrus clouds using data from other remote sensing tools [25].

In [26], an analysis was conducted on the results of retrieving the properties of the HLCs in the Tomsk region (56.46 N, 85.05 E) using the LOZA-M3 lidar [24] and the MODIS spectroradiometer (Aqua and Terra satellites). Cloud parameters that are promising from the perspective of assessing the orientation of ice crystals and retrieved from passive satellite sensing data were identified: reflection ratio ρ and effective emissivity ε. Here, the reflection coefficient refers to the fraction of solar radiation in the visible range reflected by clouds back into outer space [27]. The emissivity shows the proportion of outgoing long-wave radiation retained by the cloud [28]. Additionally, in [26], estimates of their threshold values were obtained in the presence/absence of specular reflecting layers in the HLCs. Thus, at ρ ≥ 0.15 and ε ≥ 0.5, it is assumed that the clouds observed on the MODIS image consist of preferentially oriented ice crystals. The reliability of the threshold values found was confirmed in [29] based on the results of an analysis of active and passive measurements of the HLC properties by the CALIOP lidar (CALIPSO satellite) and the MODIS spectroradiometer, respectively, obtained over Western Siberia from 2006 to 2007, when the lidar off-nadir angle was only 0.3°. It should be noted that the estimates of ρ and ε considered in [26,29] were obtained for a single region of the planet. Therefore, the question of regional features of the properties of specular reflecting layers remains open.

The purpose of this work was to analyze the results of long-term observations of specular reflecting layers in HLCs by the EARLINET (European Aerosol Research Lidar Network) lidar network and the MODIS spectroradiometer, as well as to find the regional features of their properties.

The work is organized as follows. In Section 2, we present the initial data. In Section 3, we present the method for the analysis of ground-based lidar measurements and satellite data. In Section 4, we discuss the obtained results of the comparison of the observations of specular reflecting layers in HLCs. The conclusion and future direction of this study are briefly summarized in Section 5.

2. Materials

The EARLINET lidar network was established in 2000 primarily as a research project to create a comprehensive database of atmospheric and anthropogenic aerosol properties on a continental scale [30]. Currently, EARLINET includes over 30 ground-based laser atmospheric sounding stations located throughout Europe. The database created on the basis of their observational results is one of the main sources of information on the vertical structure of aerosols. The lidars used by EARLINET are not standardized instruments but are constantly being upgraded to improve the quality and quantity of measured parameters. In addition, measurements are not carried out regularly due to weather conditions, equipment maintenance, and more. Therefore, the volumes of information accumulated to date differ by different stations. We used the EARLINET data since 2015, when measurements began to be carried out quite regularly, including the determination of the volume linear depolarization ratio. Data were received from stations in Antikythera (7 episodes), Barcelona (44), Bucharest (33), Warsaw (3), Granada (1), Kuopio (7), Leipzig (58), Lille (29), Limassol (44), Palaiseau (3), Thessaloniki (4), and Évora (49). The geographical coordinates and characteristics of the lidars installed on them can be found in [31]. We have considered the following results of ground-based laser-sounding HLCs by the EARLINET network: their top height (h_CT), total attenuated backscatter at a wavelength of 532 nm (${\beta }_{532}^{\prime }$), and volume linear depolarization ratio (δ). Moreover, h_CT was estimated based on the “cloudmask” product if the lidar station provides it or by identifying the highest point with a high value of ${\beta }_{532}^{\prime }$ in the vertical profile.

The results of remote sensing by the MODIS spectroradiometer [32] were used as a source of satellite information. This device is installed on board two satellites, Terra and Aqua, which allows us to obtain 2–4 images over Europe during the daytime between 9:00 and 15:00 UTC. We used the following MODIS data products (Collection 6.1): images in the visible spectrum with a spatial resolution of 1000 m (MOD021KM and MYD021KM), files with reconstructed cloud properties (MOD06_L2 and MYD06_L2), and georeferencing (MOD03 and MYD03). It should be noted that only those episodes of HLC observations were considered when the time between their ground-based and satellite measurements did not exceed 15 min. At the same time, images in the visible range of the spectrum and georeferencing files were used by us to localize the studied fields of cirrus clouds. The products MOD06_L2 and MYD06_L2 served as a source of information about the ratios ρ and ε, as well as the values of optical thickness (τ) (two-channel retrieval using band 2.1 μm), effective particle radius (r_eff) (2.1 μm), water path (P) (2.1 μm), and h_CT (1 km resolution) [33]. In this case, only those parts of the MODIS images that were obtained at off-nadir angles less than ±40° were considered in order to level out the errors in retrieving the above cloud properties [34,35].

3. Method

The main essence of the method for analyzing active ground-based and passive satellite observations of the HLCs is as follows:

(1) Using data from the EARLINET network data, episodes of HLC registration with h_CT > 8000 m and a vertical extent of less than 1000 m are identified. These threshold values were chosen to exclude convective clouds and mid-level clouds from consideration, which can also consist of crystal particles [36];

(2) Furthermore, the episodes of HLC observation with ${\beta }_{532}^{\prime }$ > 5 × 10⁻⁶ m⁻¹·sr⁻¹ selected at stage 1 are divided into two groups: those containing specular reflecting layers and those consisting of randomly oriented ice crystals. Thus, if δ < 0.1, then cirrus clouds belong to the first group, and when δ > 0.4, they belong to the second group. It should be noted that the threshold values used by us for ${\beta }_{532}^{\prime }$ and δ are stronger than in the original work [37], which is necessary to obtain more reliable analysis results. Figure 1 shows examples of selected episodes of HLC observations over Limassol containing specular reflective layers (11 November 2020 from 12:00 to 12:45 UTC) and consisting of randomly oriented ice crystals (16 March 2022 from 09:45 to 12:00 UTC). In the first case, the cirrus clouds had a vertical extent of about 500 m, and 8800 < h_CT ≤ 9200 m, and in the second case, 900 m and 8800 < h_CT ≤ 10,000 m, respectively. From Figure 1b, it can be seen that the specular reflecting layer is located in the upper part of the observed cloud. The value of the coefficient δ for cirrus clouds with a random orientation of particles varies from 0.4 to 0.6 according to Figure 1b.

(3) Synchronous MODIS satellite scenes were found for selected episodes of HLC observation with different orientations of crystal particles using known coordinates of lidar stations and georeferencing files MOD03 and MYD03. In this case, only the image on which the studied field of cirrus clouds is closest to the nadir line is selected from all existing images for a specific day. Episodes with a discrepancy of more than 15 min between ground-based lidar measurements and satellite imagery are excluded from consideration. As a result of an analysis of active and passive observations, 132 cases of synchronous registration of specular reflecting layers in the HLCs and 97 episodes of observation of cirrus clouds with a random orientation of particles were selected. Most of these cases were recorded by lidar stations in Barcelona, Bucharest, Leipzig, Limassol, and Évora.

(4) Then, for each episode of synchronous observation of the HLCs with different orientations of crystal particles, the properties τ, r_eff, P, ρ, and ε are restored at the points of location of the lidar stations using the products MOD06_L2 and MYD06_L2. The obtained values of the properties of cirrus clouds were used by us to analyze the regional features of the parameters of the specular reflecting layers, the results of which are presented in the next section.

4. Results

Figure 2 shows the scatterplots of the values of ρ and ε for HLCs with different orientations of ice crystals, obtained through long-term synchronous observations by EARLINET lidar stations and the MODIS spectroradiometer. Figure 2 shows that the values ρ < 0.15 and ε < 0.5 are typical for the majority of episodes of observation of cirrus clouds with the presence/absence of specular reflecting layers. At ρ ≥ 0.15 or ε ≥ 0.5, the number of cases of observation of HLCs with randomly oriented ice crystals is slightly lower than those with particles preferentially oriented in the horizontal plane. These threshold values are taken from our previous paper [26]. These differences are most pronounced over Barcelona, Leipzig, and Évora. Over Limassol, cirrus clouds with different orientations of ice crystals are practically indistinguishable from each other based on the parameters ρ and ε and generally have low values of these parameters.

Figure 3 shows scatterplots of the values of ρ and ε, ρ and τ, ρ and r_eff, and ρ and P for the HLCs with different orientations of crystal particles across all the lidar stations considered in this study. These plots were constructed based on the results of analyzing their data with the long-term MODIS satellite imagery for the period from 2015 to 2022. It can be seen from Figure 3a that at ρ ≥ 0.15 and ε ≥ 0.4, we can confidently speak about the presence of specular reflective layers in cirrus clouds and use these threshold values to detect them from passive satellite sensing data over Europe. The HLC isolated in this case is capable of reflecting more than 15% of solar radiation and retaining more than 40% of the long-wave radiation leaving the underlying surface. The information obtained can be used in solving various scientific and practical problems of climatology and climate modeling related to the study of radiative transfer. At ρ < 0.15, cirrus clouds with different orientations of ice crystals are virtually indistinguishable from each other according to passive satellite sounding data. It can be seen from Figure 3b that the specular reflecting layers can be located in both thin and denser HLCs. At the same time, a region (shown by a dotted line) with a high concentration of episodes of observation of predominantly oriented particles is quite clearly distinguished, compared with randomly oriented with 0.15 ≤ ρ ≤ 0.25 and 2 ≤ τ ≤ 3.5. Figure 3c shows that for ρ < 0.15, the spread of r_eff values is larger than for ρ ≥ 0.15 and decreases even more with increasing ρ values. At the moment, we cannot explain this feature, which can be considered as a promising direction for the development of this work. It can be seen from Figure 3d that the water path in the HLC, both containing specularly reflecting layers and not containing ρ ≥ 0.15, is almost the same and varies in the range from 10 to 100 g/m². Moreover, there is an almost identical linear relationship between ρ and P with an almost complete absence of emissions from clouds with different particle orientations, as evidenced by the almost coinciding trend lines in Figure 3d.

Table 1. Mean values of the properties of high-level clouds with preferential (P) and random (R) orientation of crystal particles for the period from 2015 to 2022 at various lidar stations in Europe.

Type of Particle Orientation	<ρ>	<ε>	<τ>	<reff>, μm	<P>, g/m2	<hCT>, m
Barcelona
P	0.12	0.45	1.6	33	30	9800
R	0.09	0.43	1.7	32	32	9300
Leipzig
P	0.14	0.49	2.0	29	35	10,400
R	0.10	0.38	1.8	30	29	10,100
Lille
P	0.12	0.44	2.1	27	27	10,400
R	0.11	0.40	1.8	22	22	10,900
Limassol
P	0.11	0.44	1.5	32	29	10,200
R	0.10	0.42	1.5	34	30	9900
Évora
P	0.13	0.50	1.6	30	26	10,000
R	0.09	0.48	1.1	37	26	8900
All stations
P	0.13	0.49	1.8	30	31	10,100
R	0.10	0.43	1.7	31	28	9900

shows the average values of the HLC properties with different ice crystal orientations, retrieved from MODIS satellite data, over individual ground-based lidar stations and across Europe as a whole for the period from 2015 to 2022. Table 1 shows that cirrus clouds with specular reflective layers have higher mean values of reflection coefficients <ρ> and effective emissivity <ε> than those without. The remaining estimates of the means <τ>, <r_eff>, <P>, and <h_CT> of the HLCs with different crystal particle orientations are quite similar and demonstrate a different trend over individual lidar stations. We only observed higher values of <h_CT> for cirrus clouds over Évora, which consists of preferentially oriented ice crystals in the horizontal plane. Overall, and particularly over Limassol, the mean values of the HLC parameters considered in this study with different cloud particle orientations, obtained over various ground-based lidar stations, are quite similar.

The threshold values (ρ ≥ 0.15 and ε ≥ 0.4) found in this work are consistent with the results of their determination over Western Siberia, as reported in [26]. Therefore, they can be used to estimate the horizontal dimensions of the specular reflecting layers in the HLCs and calculate the mean values of their parameters both on regional and global scales from long-term satellite data. Figure 4 shows a fragment of the original MODIS image in full size, with the location of the lidar station in Bucharest marked with a red triangle. Figure 4b shows the result of selecting the presumed location of specularly reflecting layers (highlighted in green) among all the HLCs (highlighted in blue). According to the results of laser probing in Figure 4c,d, the specular reflecting layer was located over the entire height of the HLC observation from 11,500 to 15,000 m. The results of retrieving the properties of the cloud field in Figure 4, consisting of preferentially oriented ice crystals, showed that its area is 11,600 km², <h_CT> = 11,800 m, <ρ> = 0.23, and <ε> = 0.71. Note the homogeneity of the specular reflective layer in Figure 4b, which can have a significant effect on the radiative transfer in the region under study at the considered time point.

5. Conclusions

This study aimed to identify the characteristic features of specular reflective layers in HLCs by analyzing data from EARLINET ground-based lidar stations and MODIS satellite imagery. Successfully solving this problem would provide ample opportunities to study cirrus clouds, which are composed of ice crystals oriented horizontally, in terms of assessing their impact on radiative transfer in the Earth’s atmosphere and predicting future climate change. The results of this study can be considered as a step towards improving the understanding of cloud feedback and its radiative effects. However, before using the threshold values ρ ≥ 0.15 and ε ≥ 0.4 found in this study for operational processing, further research is required. Nonetheless, the results obtained can be used as an auxiliary source of information about the possible sizes of specular-reflecting layers, at least over Europe, in combination with the results of ground-based laser sounding. It is worth noting that the estimates of ρ and ε obtained over Europe are in good agreement with the results given in [26] for Western Siberia. Additionally, the properties of cirrus clouds with different orientations of crystal particles, obtained from MODIS data, are quite similar over different regions of Europe (Figure 1 and Table 1). Furthermore, it was found that specular reflecting layers can be present in both thin HLCs (τ < 1) and in denser ones (τ > 5). In this case, the effective radius of ice crystals oriented horizontally varies from 5 to 60 µm in the region under consideration, and the water path of such cirrus clouds is less than 100 g/m² (Figure 3). The threshold values of ρ and ε found in this study can be used to determine the horizontal dimensions of the specular reflecting layers, which are capable of reflecting more than 15% of solar radiation and retaining more than 40% of the outgoing long-wave radiation, in the HLCs over vast areas (Figure 4). A promising area of work is the analysis of the spatial properties of cirrus clouds, consisting of preferentially oriented ice crystals, over various regions of the planet and a comparison of the results obtained on this basis.