† Corresponding author. E-mail:
Project supported by the National Natural Science Foundation of China (Grant Nos. 61527805 and 41776181).
We used discrete dipole approximation (DDA) to examine the scattering and absorption characteristics of spherical ice crystal particles. On this basis, we studied the scattering characteristics of spherical ice crystal particles at different frequencies and non-spherical ice crystal particles with different shapes, aspect ratios, and spatial orientations. The results indicate that the DDA and Mie methods yield almost the same results for spherical ice crystal particles, illustrating the superior calculation accuracy of the DDA method. Compared with the millimeter wave band, the terahertz band particles have richer scattering characteristics and can detect ice crystal particles more easily. Different frequencies, shapes, aspect ratios, and spatial orientations have specific effects on the scattering and absorption characteristics of ice crystal particles. The results provide an important theoretical basis for the design of terahertz cloud radars and related cirrus detection methods.
Cirrus clouds are generally distributed from the upper troposphere to the lower stratosphere, covering an average of approximately 20%–30% of the Earth’s surface at any given time.[1] By reflecting solar short-wave radiation and absorbing long-wave radiation from the atmosphere and the ground, cirrus clouds play an important role in regulating the radiation budget of the terrestrial gas system. Cirrus clouds are ice clouds, and they are mainly composed of ice crystal particles of various shapes and sizes. When calculating the scattering characteristics of particles, the Mie scattering theory is only applicable to isotropic homogeneous spherical particles. Therefore, the selection of a suitable algorithm to accurately calculate the scattering characteristics of non-spherical ice particles has become a major challenge in the remote sensing of cirrus clouds.
Satellite remote sensing provides an essential means for atmospheric and cloud detection. Passive microwave remote sensing mainly uses millimeter frequency bands that can strongly penetrate cirrus clouds and are primarily used for the remote sensing of temperature, humidity, and precipitation. Typical ice cloud particles, which can be probed by electromagnetic radiation from microwaves to infrared waves, have diameters ranging from 50 μm–400 μm.[2] On the one hand, microwave (0.0003 THz–0.3 THz) satellite remote sensing can penetrate cirrus clouds, but information on clouds containing only large-grained particles is difficult to obtain with microwaves. On the other hand, the near-infrared band (100 THz–400 THz) has a short wavelength, which can obtain information on the upper surfaces of cirrus clouds, but it is difficult to probe the vertical structure of cirrus clouds. The terahertz band has important applications in many research fields.[3–6] The ice cloud particle sizes are within the range of 30 μm–3000 μm wavelengths of the terahertz frequency band (0.1 THz–10 THz).[7] Therefore, the terahertz band is theoretically the best one for simultaneously detecting the horizontal and vertical structure of cirrus clouds.
Many scholars have developed different numerical algorithms that describe the scattering characteristics of non-spherical ice crystal particles. These are the T-matrix method (TMM),[8] finite-difference time-domain (FDTD),[9] and discrete dipole approximation.[10] FDTD is suitable for small particles with complex and heterogeneous shapes. TMM is applicable to all kinds of uniformly symmetric particles, but it has difficulty with numerical stability and convergence.[11] The the discrete dipole approximation (DDA) method has a simple physical concept and is applicable to particles of any shape. Consequently, it has been widely used to calculate the scattering characteristics of non-spherical ice crystal particles. For example, Wu et al.[12] calculated the attenuation efficiency of several non-spherical ice crystal particles by using the DDA method for millimeter wave cloud radar. They concluded that in actual ice clouds, considering hexagonal ice crystals and elliptical ice crystals as spherical particles of the same volume underestimates their attenuation and backscatter. Ruan et al.[13] calculated the radiation characteristics of non-spherical particles according to DDA method, which are different from those calculated for spherical particles of the same volume. Such differences are affected by the shape, scale parameters, and complex refractive indices of the particles. Wang et al.[14] used Lorenz–Mie theory and the TMM method to calculate the terahertz scattering characteristics of spherical water droplets and non-spherical ice crystal particles, respectively. The average attenuation of the non-spherical particles was 28 dB/km, which was slightly less than that of the spherical particles at all frequencies. This research shows that the terahertz scattering characteristics of non-spherical particles are of great significance. Despite this, many studies employ only millimeter wave cloud radars, and use of the terahertz frequency band is still very limited. However, the terahertz frequency band is a vital research field for both pure research and its applications.[15–18]
In order to provide a theoretical basis for the detection of non-spherical ice crystals in cirrus clouds, the use of the DDA method to study the scattering characteristics of terahertz non-spherical ice crystal particles is crucial. Therefore, the scattering characteristics of spherical ice crystal particles and water droplets at 94 GHz, 220 GHz, and 340 GHz were calculated using the Mie scattering results as the standard values for the accuracy test. The effects of different frequency bands on spherical ice crystal particles and the effects of particle shape, aspect ratio and orientation on the scattering characteristics of non-spherical ice crystal particles were compared and analyzed.
The DDA method is a discrete expression of the integral form of electromagnetic scattering formulas,[19] that is suitable for studying the scattering and absorption characteristics of particles with arbitrary shapes. Its main principle is to assume that a particle can be discretized into N small cubes, and the scattering characteristics of each small cube can be represented by a dipole. Thus, a finite number of discrete, interacting dipole arrays can be used to approximate particles of arbitrary shape. Therefore, the study of the scattering characteristics of actual particles can be transformed into the study of the scattering characteristics of these dipoles.
According to previous studies,[20–23] the DDA method is sufficiently accurate when the following condition is met:
The relationship between the actual volume V of the target and the number of dipoles N is
Due to limitations in the computing performance of modern computers, the number of dipoles cannot exceed 106. For this paper, the main input parameters are frequency (wavelength), particle equivalent radius, particle shape, and particle orientation. The main simulation outputs of the DDA algorithm are the extinction efficiency Qext (the ratio of the extinction cross section), the scattering efficiency Qsca (the scattering cross section), the absorption efficiency Qabs (the absorption cross section), and
Because measuring the scattering characteristics of ice crystal particles is difficult, it is assumed that the simulated results of Mie scattering are accurate for spherical particles. Therefore, in this paper, Mie scattering and the DDA method are used to simulate spherical ice crystal particles with the same parameters. In order to verify the accuracy of the DDA algorithm, calculations were carried out for three frequency bands: 94 GHz, 220 GHz, and 340 GHz. The complex refractive indices of the non-spherical ice crystal particles for the three frequency bands were m = 1.782 + 0.003 i, m = 1.782 + 0.0049 i, and m = 1.781 + 0.007 i, respectively.[25] The range of effective radii of the spherical ice crystal particles simulated in this paper was 50 μm–1550 μm.
Based on the DDA and Mie scattering methods, DDSCAT7.3 and MATLAB software were used to calculate the scattering and absorption characteristics of spherical ice crystal particles at 94 GHz, 220 GHz, and 340 GHz. The results are shown in Fig.
Figure
In order to compare the two methods more intuitively, we calculated the relative errors between the results generated by the DDA and Mie scattering methods for the scattering of spherical ice crystal particles, as shown in Fig.
Initially, the ice crystal particles were assumed to be spherical or ellipsoidal, but most of the simulated ice crystal particles were shaped as hexagonal columns, hexagonal plates, and bullet flowers. Among these, the most common non-spherical particles were hexagonal in shape. Using the aspect ratio, the hexagonal ice crystal particles were divided into the following categories:[26] long column (L/2a = 4), thick column (L/2a = 2), block column (L/2a = 1), thick plate (L/2a = 1/6), and thin plate (L/2a = 1/20), where L is the height of the hexagonal prism (the distance between the two hexagons at either end of the particle), a is the length of each side of the hexagon, and 2a is the distance between the vertical angles of the hexagon. Therefore, the effects of different frequency bands, sizes, aspect ratios, and orientations on the scattering characteristics of hexagonal ice crystal particles could be studied.
Figure
Spherical ice crystal particles and water droplets were selected as representatives in the study of scattering and absorption characteristics at 94 GHz, 220 GHz, and 340 GHz. As shown in Fig.
Figure
To sum up, analyzing the scattering and absorption characteristics of water droplets and ice crystal particles in the 94-GHz, 220-GHz and 340-GHz bands is effective for quantitative remote sensing of ice crystal particles in the terahertz band.
As shown in Fig.
In summary, the scattering and absorption characteristics of ice crystals are influenced by particles with a wide range of shapes. This suggests that it is unreasonable to assume that particles are either spherical or ellipsoidal and that the exclusion of block column particles introduces additional error.
Hexagonal ice crystal particles with aspect ratios of L/2a = 1/20, L/2a = 1/6, L/2a = 1, L/2a = 2, and L/2a = 4 were selected to study the effects of different aspect ratios on particle scattering characteristics in the 340-GHz frequency band, as shown in Fig.
Figure
If the aspect ratio is < 1, it is called a hexagonal plate; if it is > 1, it is called a hexagonal column. The above analysis suggests that the scattering characteristics of hexagonal column ice crystal particles show multiple peaks, which are more abundant than those of particles in the shape of a hexagonal plate. In addition, the total absorption value of the hexagonal plate ice crystal particles is smaller than that of the hexagonal column particles. Therefore, ice crystal particles with different aspect ratios have significantly different scattering and absorption characteristics.
Spatial orientation refers to the position of the particles in space. For non-spherical particles, the spatial orientation of the particles also has significant influence on their scattering characteristics. Laboratory fame (LF) and target fame (TF)[28] were used to determine the spatial orientation of particles. As shown in Fig.
The orientation of the target in space is determined by the polarization angles θ, Φ, and β, where θ is the angle between a1 and the x axis, Φ is the angle at which a1 rotates about the x axis, and β is the angle at which a2 rotates about a1. In this paper, thick column particles with an aspect ratio of 2 are selected as representatives of all particles. In addition, the following four simple orientations of thick column particles are used: horizontal orientation, oblique orientation 45°, oblique orientation 60°, and vertical orientation. The polarization angles Φ and β corresponding to the four spatial orientations are all identical (i.e., 0), and the polarization angle θ was fixed at values of 0°, 45°, 60°, and 90°.
Figure
Figure
In summary, the variation in spatial orientation of the thick column particles causes significant differences in their scattering and absorption characteristics. This indicates that the wide range of spatial orientations of ice crystal particles introduces uncertainty in cirrus cloud particle detection. However, the results presented in this section indicate that the information obtained by the terahertz frequency band (at 340 GHz) can probe the orientation of cirrus ice crystal particles. Therefore, more research on terahertz cloud radars should be conducted in the future in order to fully exploit this advantage.
In this paper, the DDA algorithm was used to study the scattering and absorption characteristics of spherical and non-spherical ice crystal particles with equivalent radii in the range of 50 μm–1550 μm using the terahertz frequency band. The effects of different frequencies, shapes, aspect ratios, and spatial orientations on the scattering and absorption characteristics of cirrus ice crystal particles were analyzed. This provides an important theoretical basis for the design of terahertz cloud radars and the inversion of cirrus parameters. The following are the major results of this paper.
(i) The calculation experiments show that the scattering characteristics of spherical ice crystal particles in DDA and Mie at 94 GHz, 220 GHz, and 340 GHz are almost identical, and the relative error of scattering calculation is < 1.8%. The calculated absorption characteristics of DDA and Mie scattering in 94-GHz, 220-GHz, and 340-GHz bands are highly consistent within the equivalent radius of 1100 μm, and the calculated relative error is < 3.5%. Therefore, the DDA algorithm is suitable for the calculation of scattering and absorption characteristics of non-spherical ice crystal particles.
(ii) The scattering and absorption characteristics of spherical ice crystal particles show different variations in different frequency bands. The millimeter wave band is suitable for monitoring larger particles, but the terahertz band (340 GHz) is more suitable for monitoring the majority of cirrus ice crystal particles because it has a scattering peak at the equivalent radius of 350 μm, which is exactly the size of most ice crystal particles. The scattering of ice crystal particles at 340 GHz is more obvious than that of water droplets, indicating that the terahertz frequency band (340 GHz) is more suitable for remote sensing monitoring of ice crystal particles than the commonly used millimeter wave frequency band (94 GHz).
(iii) A comparison of the scattering characteristics of spherical, ellipsoidal, and hexagonal column particles reveals that the widely used calculation approximation of assuming particles are spherical or ellipsoidal is unreasonable and produces errors when applied to terahertz remote sensing.
(iv) Ice crystal particles with different aspect ratios and different spatial orientations have different scattering and absorption characteristics. In general, the scattering characteristics of hexagonal column particles with horizontal orientation show much more variation than those of hexagonal plate particles, indicating the need for further experimental research.
[1] | |
[2] | |
[3] | |
[4] | |
[5] | |
[6] | |
[7] | |
[8] | |
[9] | |
[10] | |
[11] | |
[12] | |
[13] | |
[14] | |
[15] | |
[16] | |
[17] | |
[18] | |
[19] | |
[20] | |
[21] | |
[22] | |
[23] | |
[24] | |
[25] | |
[26] | |
[27] | |
[28] |