Automatic detection of overshooting tops and their properties using Neural Networks

Result code in IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00020699%3A_____%2F24%3AN0000052" target="_blank" >RIV/00020699:_____/24:N0000052 - isvavai.cz</a>

Alternative codes found

RIV/00020699:_____/24:N0000057

Result on the web

<a href="https://program-eumetsat2024.kuoni-congress.info/presentation/automatic-detection-of-overshooting-tops-and-their-properties-using-neural-networks" target="_blank" >https://program-eumetsat2024.kuoni-congress.info/presentation/automatic-detection-of-overshooting-tops-and-their-properties-using-neural-networks</a>

DOI - Digital Object Identifier

—

Result language

angličtina

Original language name

Automatic detection of overshooting tops and their properties using Neural Networks

Original language description

Overshooting tops (OT) are one of the phenomena at the cloud top of convective storms that can indicate their severity [1][2]. Because of satellite location, satellite data is the primary tool for their study as they have an unobscured view on the cloud top features. In this work, we focus on OT detection from satellite data and measurement of their height as an important parameter related to the severity of the associated storm. Research on the OT height determination has been previously carried out, for example, by Kristopher Bedka [3] and Ján Kaňák [4], whose method and database our work employs. The detection algorithm is based on convolutional neural networks. These networks are used for the detection of OTs in the satellite images as well as for the estimation of the OT height. The models are trained and tested on the OT database of Ján Kaňák [4], with approx. 10 thousand cases of OT manually detected from HRV images and height measured from their shadow. While the OT detection techniques are typically based on the distinct cold features visible in the thermal IR 10.8 channel, the presence of the ground truth information on the length of the OT shadow provided by the training database allows us to strengthen also the role of visible channels. We compare the importance of individual channels for the OT detection and evaluate the benefits of having the extra information on OT shadow length during training. If good quality FCI data become available during the 2024 convective season, we will also asses the transferability of the model trained on SEVIRI data to the new FCI instrument. [1] Setvák M., Lindsey D. T., Novák P., and Wang P. K. (2010). Satellite-observed cold-ring-shaped features atop deep convective clouds. Atmospheric Research. 1-2(97), p. 80-96. DOI: https://doi.org/10.1016/j.atmosres.2010.03.009 [2] Chernokulsky A., Shikhov A., Yarinich Y. and Sprygin A. An Empirical Relationship among Characteristics of Severe Convective Storms, Their Cloud-Top Properties and Environmental Parameters in Northern Eurasia. Atmosphere. 2023; 14(1):174. https://doi.org/10.3390/atmos14010174 [3] Bedka K., Brunner J., Dworak R., Feltz W., Otkin J., and Greenwald T. (2010). Objective Satellite-Based Detection of Overshooting Tops Using Infrared Window Channel Brightness Temperature Gradients. Journal of Applied Meteorology and Climatology 49, 2, 181-202, DOI: https://doi.org/10.1175/2009JAMC2286.1 [4] Kaňák J., Bedka K. M, and Sokol A. (2012). Mature convective storms and their overshooting tops over Central Europe: Overshooting top height analysis for summers 2009-2011. Conference: 2012 EUMETSAT Meteorological Satellite Conference, Session 7. Available from https://www.eumetsat.int/media/8672

Czech name

—

Czech description

—

Type

O - Miscellaneous

CEP classification

—

OECD FORD branch

10509 - Meteorology and atmospheric sciences

Project

—

Continuities

I - Institucionalni podpora na dlouhodoby koncepcni rozvoj vyzkumne organizace

Publication year

2024

Confidentiality

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů