1. INTRODUCTION

The glory is one of the most beautiful natural meteorological phenomena. Appearing as colored concentric rings surrounding the shadow of the observer, it is caused by backscattering of sunlight from clouds. This phenomenon has been the subject of intense interest and study for several decades through laboratory experiments, field observations [1–3], and numerical simulations [4–8]. For theoretical considerations and various hypotheses about the formation of glories, readers are referred to [9–16].

Despite the above advancements in the understanding of the glory phenomenon, in-situ observations of clouds capable of producing glories are rarely available. Recently, observations of glories have been reported from the space shuttle [17] and from a satellite [18]. Apart from the glory, the first quantitative remote sensing measurement of the cloudbow, another atmospheric phenomenon, from space was provided by POLDER instruments, suggesting that clouds capable of producing cloudbows have narrow cloud droplet distributions [19]. To the best of our knowledge, in-situ measurement of clouds producing glories have not previously been reported. Here, we present properties of in-situ measurement of a thin layer cloud where a glory was observed from the surface of the same cloud (from the instrumented aircraft). The observation was made in October 2014 over Mahabaleswar, India, during the cloud aerosol interaction and precipitation enhancement experiment (CAIPEEX) [20]. We used measurements of the cloud droplet size distribution to understand the microphysical properties of these clouds. Using these measurements as an input to the Mie scattering theory, we simulated the glory and compared the simulation with the observed concentric rings.

This paper is arranged as follows. Section 2 describes the data analysis and methodology as well as the instruments. Section 3 presents our results. Finally, Section 4 summarizes the major findings of this study.

2. DATA ANALYSIS AND METHODOLOGY

Aircraft observations of a thin layer cloud producing the glory phenomenon were made on 28 October 2014 (73.66 °E, 18.08 °N). The cloud properties were measured by a cloud droplet probe (CDP) at 2 Hz resolutions ($\approx 45\mathrm{m}$ of spatial resolution at $90{\mathrm{ms}}^{-1}$ of true air speed). The CDP, manufactured by Droplet Measurement Technologies, is a spectrometer that works on the principle of forward scattering. The CDP uses a laser, and the forward scattered light from the illuminated particle is used to size the droplet (http://www.dropletmeasurement.com/cloud-droplet-probe-cdp-2). It measures cloud droplet size distribution (DSD) in the diameter range from 2.5 to 50 μm in 30 unequal range bins. The total number concentration ($N_c$, ${\mathrm{cm}}^{-3}$), mean diameter ($D_m$, μm) liquid water content (LWC, ${\mathrm{gm}}^{-3}$), and spectral width ($\sigma$, μm) can be calculated from the DSDs [21]. The altitude of the cloud measurement and temperature are obtained from the aircraft integrated meteorological measurement system (AIMMS) probe and Rosemount Temperature probe, respectively. We also used concentrations of supermicron aerosol ($1\le D\le 3\mathrm{\mu m}$) to understand the haze particles present in the atmospheric boundary layer. Aerosol data are obtained from onboard passive cavity aerosol spectrometer probe (PCASP) that provide aerosol concentrations in the diameter range of 0.1–3 μm.

The measured DSDs are used to simulate glory phenomenon by the Mie scattering theory. We used the MiePlot computer program (available at http://www.philiplaven.com/mieplot.htm) to simulate the glory using the measured DSD and compared it to the glory rings photographed.

3. RESULTS

During a research sortie on 28 October 2014 near Mahabaleswar, India, we noticed a glory phenomenon from the surface of a thin layer cloud. The glory phenomenon was observed in the afternoon hours at around 11:00 UTC. The photograph of the layer cloud is shown in Fig. 1. The cloud layer was above several convective clouds, but it was detached from the cloud systems below [see Fig. 1(a)]. Figures 1(b)–1(d) show that the layer cloud was thin and transparent as the ground was visible. It may also be noted that a thick haze layer between 1.8 and 2.5 km was observed. Mean concentration of supermicron size particles ($1\le D\le 3\mathrm{\mu m}$) between these two altitudes was found to be $1.20{\mathrm{cm}}^{-3}$. In addition, the aircraft was flying over the dark green mountainous regions of the Western Ghats. The mountainous regions and the presence of boundary layer haze thus provided a dark background to observe the glory. The colored rings of the glory were formed in the transparent sections, which were located at the edges of the layer cloud. Three horizontal cloud transects were made with the help of a CDP. From the vertical positions of the aircraft the geometrical thickness of the cloud was estimated to be of nearly 100 m. Using the mean cloud droplet effective radius ($r_e$, μm) and LWC, we calculated the cloud optical depth (COD) [22]. It is assumed that the average microphysical properties of the layer cloud are uniform vertically. Using mean $r_e\approx 5.74\mathrm{\mu m}$ and $\mathrm{LWC}\approx 0.03{\mathrm{gm}}^{-3}$ values obtained from three cloud transects, COD is found to be $\approx 0.78$. This COD value suggests the presence of a very thin layer cloud.

 

Fig. 1. Photographs of the thin dark layer cloud where the glory phenomena were observed.

Download Full Size | PDF

We consider one cloud pass to discuss microphysical properties of this glory-forming cloud.

Figure 2(a) shows the altitude and temperature of the cloud pass. The observations were made at an altitude of 4.8 km above mean sea level. The temperature range was from 2.33°C to 2.37°C, indicating that the cloud droplets are in the liquid phase. Note that the glory was observed in the first few hundred meters of the thin layer cloud. Figure 2(b) shows the properties of cloud droplet spectra, e.g., $N_c$, $D_m$, LWC, and $\sigma$ in which the glory phenomenon was observed. Total droplet number concentrations ($N_c$, ${\mathrm{cm}}^{-3}$) of the cloud transect are shown in Fig. 2(b). It suggests that low cloud droplet number concentrations ($<45{\mathrm{cm}}^{-3}$) are present in these clouds. The mean diameters ($D_m$, μm) of the cloud droplets are also shown in Fig. 2(b). At the edges of the cloud, $D_m$ values are close to 13 μm while in the later part of the cloud transect their values are decreased nearly to 11 μm and remained constant thereafter. Decreased $D_m$ values may indicate evaporation of cloud droplets. The spectral width ($\sigma$, μm) of the DSDs is more or less constant throughout the cloud transect with their values close to 1. Small values of $\sigma$ indicate that the DSDs are narrow, implying that the cloud droplets are tightly clustered around a droplet diameter, even though the sampling frequency is only 2 Hz. As expected, these clouds contain very small amounts of liquid water content, i.e., $\mathrm{LWC}<0.05{\mathrm{gm}}^{-3}$.

 

Fig. 2. (a) Altitude and temperature of the aircraft observation of clouds on 28 October 2014. Glory phenomenon was observed formed by these thin layer clouds. (b) Variations of total droplet concentrations ($N_c$, ${\mathrm{cm}}^{-3}$), mean diameter ($D_m$, μm), liquid water content (LWC, ${\mathrm{gm}}^{-3}$), and spectral width ($\sigma$) of the droplet spectra. DSDs used to simulate the glory phenomenon were of the first few seconds of cloud pass. (Shown by an ellipse.)

Download Full Size | PDF

To understand the DSD properties capable of producing a glory, an instantaneous DSD measured at 2 Hz resolution is shown in Fig. 3. We selected this droplet spectrum from the first few hundred meters of cloud pass. In addition, to distinguish its properties from other types of clouds, we also plotted typical DSDs observed in convective and stratus clouds, also measured at 2 Hz resolution. It clearly shows that the DSD that can produce the glory phenomenon is narrower than the DSDs of convective or stratus clouds. Note that this distinction is true for data at 2 Hz resolution in the present scenario. A peculiar distinction of the glory cloud from other DSDs is the well-defined $N(D)$ peak at around 12 to 13 μm. The width of the DSD is also extremely narrow. This implies that the glory DSDs are well ordered at certain droplet diameter ranges; e.g., here it is 12–13 μm.

 

Fig. 3. Typical cloud droplet size distribution of glory-producing cloud (shown as glory cloud), convective cloud, and stratus cloud as observed from aircraft observation at 2 Hz resolution. DSD parameters, e.g., $N_c$, $D_m$, $\sigma$, and LWC are provided for each distribution.

Download Full Size | PDF

We used the various DSDs shown in Fig. 3 as inputs to the MiePlot program to simulate the backscattering of sunlight as shown in Figs. 4(a)–4(c). The inner red ring in Fig. 4(a) has a radius of about 1.5° that corresponds to scattering from a droplet of about a 32 μm diameter. The rings of the glories in Figs. 4(b) and 4(c) are significantly larger, implying that they are caused by smaller droplets. All three simulations in Fig. 4 show glories, but it is important to recognize that glories were observed only on the cloud corresponding to the DSD used for Fig. 4(c). The reason for this discrepancy is that the simulations consider only single scattering. In practice, the colored rings of glories on clouds are generally seen against an almost white background due to multiple scattering from the cloud droplets. Further work is necessary to model the effects of multiple scattering for the cloud parameters measured in this study.

 

Fig. 4. Simulations of single scattering of sunlight using the measured DSDs shown in Fig. 2 for (a) the convective cloud, (b) the stratus cloud, and (c) the glory-producing cloud.

Download Full Size | PDF

By superimposing the simulation from Fig. 4(c) on a photograph of the observed glory as in Fig. 5, we can see that the simulation provides a good fit with the observed glory in terms of the sequence of colors and the size of the colored rings. This confirms the validity of the Mie theory simulations.

 

Fig. 5. Glory phenomenon observed on 28 October 2014 near Mahabaleswar, India. Contrast of the colored rings are enhanced. The simulated rings are superimposed on the observed glory rings. The Mie scattering simulation is based on the measured droplet size distribution shown in Fig. 3 for the glory-producing cloud.

Download Full Size | PDF

Our results do not support Nevzorov’s hypothesis [15] that glories are caused by scattering by particles of amorphous water. First, our temperature measurements indicate that the water droplets causing the glory were in the liquid phase at $T=2.33–2.37°\mathrm{C}$, whereas Nevzorov assumed droplets at subzero temperatures. Second, our measurements indicate mean droplet diameters of $D_m\approx 13\mathrm{\mu m}$, whereas Nevzorov postulated $D_m>20\mathrm{\mu m}$ or even larger. Third, the close match in Fig. 5 between the photograph and our simulations using the Mie theory was obtained assuming that the refractive index of water is about 1.33, whereas Nevzorov’s calculations assumed a refractive index of about 1.8.

4. CONCLUSIONS

While flying in a research aircraft equipped to measure the properties of cloud particles, a visual observation of a glory allowed us to investigate the cloud microphysical properties that generate glories. Our principal findings are:

This study provides the first direct measurement of glory-producing clouds by an instrumented aircraft.