A TOOLKIT FOR THREE-DIMENSIONAL RECONSTRUCTION AND VISUALIZATION OF WEATHER RADAR IMAGES


Tezin Türü: Yüksek Lisans

Tezin Yürütüldüğü Kurum: Orta Doğu Teknik Üniversitesi, Fen Bilimleri Enstitüsü, Jeodezi ve Coğrafi Bilgi Teknolojileri, Türkiye

Tezin Onay Tarihi: 2019

Tezin Dili: İngilizce

Öğrenci: MUSTAFA AHMET PESEN

Danışman: Uğur Murat Leloğlu

Özet:

Weather radars are among key instruments that feed numerical weather prediction models. They form two-dimensional (2D) images at various elevation angles with radii of tens of kilometers. These images are actually sections from three-dimensional (3D) time-dependent volumetric data. In modern radars, each cell can be assigned a type of hydrometeor after processing. However, the visual interpretation of these images is not easy due to the complicated viewing geometry. Nevertheless, computer graphics can make weather radar images easily observable and analyzable. In this thesis, a method and a toolkit for 3D reconstruction and visualization of weather radar images are presented. The toolkit enables users to process and visualize NEXRAD Level III hydrometeor classification images in the form of closed-volumes and collected images are converted into volumetric display of precipitation types in following steps: Firstly, continuous and smooth density estimates are obtained for each precipitation type in these 2D images and the type with the maximum density is assigned to the corresponding cell if it is above a threshold. Secondly, the resulting image is segmented using connected components analysis. Then, the boundaries of each segment are extracted and transferred to the 3D space by using metadata. Subsequently, the 3D polygons from different levels but belonging to the same object are matched. Finally, a 3D wireframe structure is created from the matched 3D polygons and displayed sequentially to create 3D visual animation using computer graphics algorithms. The toolkit can help weather radar operators to better grasp the complicated weather events in 3D.