Detection of gravity waves in Meteosat imagery by grating cell operators

Abstract

An effort was made to objectively identify regions of parallel stripes in images recorded by the Spinning Enhanced Visible and InfraRed Imager instrument on board the Meteosat Second Generation geostationary satellites. Such a grating pattern can signal the presence of gravity waves which eventually may lead to atmospheric turbulence, a hazard to be dealt with in the daily routine work of aviation meteorologists. A pattern detection algorithm, based on Gabor filters plus subsequently applied grating cell operators, is implemented, adapted and tested. The method is shown to be capable of identifying grating patterns despite the unfavourable relation between the scale of the sought waves and the spatial resolution of the imagery. Some phenomena producing similar patterns (hence similar responses of the algorithm) have been identified: marine stratocumulus and regular arrangements of mountain ranges and valleys. If present in the image, they have the potential to considerably impact the analyses. The results for the 7.3 µm water vapour channel are particularly encouraging in this context as filtering the unwanted alarms is straightforward in this case.

Analysis of Trapped Small‐Scale Internal Gravity Waves Automatically Detected in Satellite Imagery

“…In general, for example, it is unclear whether a weaker signature is actually due to a smaller wave amplitude or rather the result of the satellite's resolution or viewing angle, or the large scale variability in temperature and moisture that affect the brightness temperature field. Since TSIW signatures are barely resolved in ABI imagery, certain wavelengths are simply better detected by ABI, as pointed out in Jann (2017, Figure 5). Regarding the ERA5 data, we expect it to be smoother than reality, possibly with local biases due to limitations in data assimilation.…”

“…In particular, radiosondes could provide a more realistic background state to help understand the observed signatures (Lane et al., 2003). Furthermore, the integration of such detection methods into nowcasting systems has already been discussed in Jann (2017).…”

“…Our TSIW trapping measure is only one possible approach to an automated analysis of TSIW trapping signatures. While we have carefully optimized this measure to capture TSIW signatures as well as possible, in the wavelength range of interest, there are of course other processes that can induce a locally wavy brightness temperature field, as already discussed in Jann (2017). Our approach has been to iteratively improve the measure with different cases, while manually sanity checking the results with cloud imprints in true‐color satellite imagery.…”

“…However, they have not applied this method for a climatological study. Jann (2017) has proposed a method that uses an extended wavelet analysis that tests given angles and wavelengths, which can be a computationally expensive task depending on the number of angles, wavelengths and test locations. In Hindley et al.…”