Satellite images of bright meteor (bolide) east of Kamchatka

By Martin Setvak (CHMI ) and Jiří Borovička (Astronomical Institute of the Czech Academy of Sciences), HansPeter Roesli (Switzerland), Jochen Kerkmann (EUMETSAT)

The very bright bolide, which exploded on 18 December 2018 at 23:48:20 UTC above the Bering Sea, east of Kamchatka, gained a lot of attention due its total released energy, which placed the event right after the famous Chelyabinsk bolide, as the second strongest case in the last decade.

The event was well covered by several meteorological satellites and their instruments. Below we focus on a series of images, captured by polar-orbiting satellites shortly after the bolide explosion.

The True Color RGB image (Figure 1), captured by the MODIS instrument of the Terra satellite, shows the area of bolide explosion about 6.5 minutes after the event, at 23:54:51 UTC.

As the area of the bolide was almost at the centre of the MODIS data swath, these images represent a nadir view of the bolide trail (the 'blob' at the bottom part of the image) and its shadow (darker elongated area at the upper part of the image).

Moreover, the almost perfect nadir-view of the bolide's fragmentation plume hides details of its vertical structure, in contrast to the extreme oblique view from Himawari at 140.7°E above the equator (see rising plume of the bolide in Figure 14, below the blue arrow on the right-hand panel in particular).

Terra, MODIS imagery

The following images show the same area in the individual MODIS bands. While band 1 (0.64µm, red) (Figure 2) and band 2 (0.86µm, NIR) (Figure 3) images are shown in their full, 250m resolution (ground pixel size), the other two images, band 3 (0.47µm, blue) (Figure 4) and band 4 (0.55µm, green) (Figure 5), both originally 500m bands, have been remapped to the same scale and map projection as bands 1 and 2 (Transverse Mercator, 57N, 172.5E, with bilinear convolution), thus appear slightly blurred as compared to the band 1. The 1km band 20 (3.8µm) (Figure 6) and band 31 (11µm) (Figure 7) were also remapped to the same projection, using the nearest neighbour resampling method.

In the True Color RGB143 image (Figure 1), the shadow of the trail appears dark brown, most likely due to the fact that the longer is the wavelength, the more easily it passes through the dust left behind in the bolide trail. Thus, the red light gets least attenuated during its passage through the trail, while green and, namely, blue are scattered more, making the shadow darker in these bands. This effect can also be nicely seen in the black and white images of the individual bands — the trail shadow is more pronounced (darker and longer) in band 3 (blue) and band 4 (green) as compared to the band 1 (red) or band 2 (0.86µm, NIR).

At an even longer wavelength of the band 20 (3.8µm, composed of mixture of reflected and emitted radiance) the shadow is much weaker and shorter, corresponding to the most dense parts of the trail. Naturally, there is no shadow in band 31, as this is already a pure thermal emission band.

The next series of images focuses on details of the bolide trail itself. To avoid remapping artifacts, all the following detailed zoom images are left in the original satellite swath projection. The zoom factor is 4x for the QKM (250m) data, 8x for the HKM (500m) data, and 16x for the 1KM (1000m) data. Resampling of the images was done using either the Nearest Neighbor (NN) method (all the images where the original pixel structure can be seen), or using 'Optimized Bicubic' (OBC) method of ENVI software, in which all the processing was done (all the other images with smoother and more blurred appearance).

The area of the zoomed images is shown in RGB143 mosaic 1 (Figure 6), highlighted by the red box. The bottom left zoom image is shown in NN version, with the same original RGB enhancement as the larger image above.

Colour balance of these RGB images was done in such a way that the brightest, regular clouds nearby appear white (with RGB components approaching 255-255-255 values, with no colour tint), and the brightness function for all these bands starts at reflectance equal to zero. The zoom image at bottom right is the OBC version of the same RGB143 image, after additional contrast enhancement.

Mosaic 2 (Figure 7) shows the same contrast-enhanced RGB143 images as above, in both NN and OBC versions. Right column shows these images with overlays, highlighting two major parts of the bolide train. Part #1 is closer to the ground, where most of the bolide body disintegration took a place, composed of thick brown dust. Part #2 appears to be located higher, above part #1, with lower concentrations of the material left behind the bolide, and though it appears brighter than part #1, it also has a light brown tint. The most narrow, east portion of part #2 is located almost exactly above the centre of part #1, possibly representing the part of bolide train just before fragmentation. The red box, located almost exactly at the centre of part #1, represents the location of the 'warmest' pixel from band 20. The arrows indicate solar illumination direction of the area (note that in this swath projection the north is not exactly upwards).

It can be seen that with increasing wavelength part #2 gets more and more transparent, disappearing from band 6 onwards. This confirms that it (part #2) is composed of very thin, transparent material. Bands 20 and 25 are in a spectral region, where both, reflected and emitted components contribute to the total measured radiance.

In principle, for very high temperatures, this can also be true for bands 6 and 7; however, taking into account the 6.5 minutes delay of the Terra overpass after the bolide explosion, during which the area cooled down significantly, this option is, in this case, rather unlikely. In band 20, the highest total radiance is detected from a pixel almost exactly in the centre of area 1, which can mean either high reflectivity of very small particles, or higher temperature. Should it be related to reflectivity, similar distribution of higher radiance values would have to also be observed in bands 6 and 7, which is not quite the case here. Therefore, the higher total radiance in the center of part #1 in band 20 is (besides the reflected component) most likely related to still somewhat higher temperatures in core of part #1. From band 28 on, the centre of part #1 already appears cold.

NOAA-20 VIIRS imagery

About an hour later, another polar orbiting satellite, covered the area — the first of the new generation of NOAA/NASA JPSS satellites, NOAA-20 (Figure 10). Even at this time, the shadow of the trail plume is still well visible in the True Color RGB of the VIIRS bands 5, 4 and 2, while the material of the plume itself still seems to be detected in this microphysical RGB (indicated on the image). This case was also observed at nadir of the satellite.

Suomi-NPP VIIRS imagery

Fifty minutes later, another VIIRS-equipped satellite, Suomi-NPP, passed above the region. The trail shadow is still detectable on the True Color RGB (Figure 11) and band M1 image (Figure 12), while the trail material is no longer detectable, even in the microphysics RGB (not shown in the page).

In this case the bolide area was at eastern part of the data swath (satellite ground track was west of the bolide area), thus any material possibly left in the bolide trail would be parallax-shifted eastward.

Himawari-8 imagery

The trail of the meteorite and its shadow on lower clouds were also observed by the Japanese Himawari-8 satellite, although at a very high viewing angle, see True Colour RGB (Figure 13) and the High resolution VIS0.64 image (Figure 14), which caused a big parallax displacement compared to the nadir MODIS image.

Note: Figures 13 and 14 are shown in native, satellite projection, to avoid distortion of the trail from re-projection (for example, see reprojected VIS0.6 image from 23:50 UTC )

As in the MODIS image, the trail appears with a red colour in the True Colour RGB and the shadow is dark grey. The cause of this, i.e. the high backscatter of the trail material in the VIS0.6 channel and the low backscatter in the VIS0.4 channel, has already been discussed above.