Title:	Topology-based feature extraction and tracking in time-varying spatial muti-field data
DNr:	NAISS 2023/22-66
Project Type:	NAISS Small Compute
Principal Investigator:	Talha Bin Masood <talha.bin.masood@liu.se>
Affiliation:	Linköpings universitet
Duration:	2023-01-16 – 2024-02-01
Classification:	10201
Homepage:	https://scivis.github.io/research/climate/
Keywords:

Abstract

Extracting meaningful features is a crucial step in understanding complex spatial data generated either through scientific simulations (e.g. climate simulations) or directly measured through instruments (e.g. satellite observations). Furthermore, if the data includes the time dimension, the tracking of extracted features over time becomes very important. Questions like when a geometric feature is born, when and how it splits or merges with other features, etc. are of extreme interest. Topology-based techniques have proven to be very useful in providing efficient and robust answers to these questions. We have recently developed novel cyclone identification and tracking methods using topological data analysis techniques for climate simulation data [See Ref 1]. However, these algorithms have been tested on relatively small time-series with low spatial resolution. We would like to design and test [See Ref 2 for our efforts on benchmark data generation] more scalable feature extraction and tracking methods which are suitable for parallel execution. The idea is to divide the spatial data into smaller chunks and use parallel processing on larger number of compute nodes in order to scale tracking algorithms to larger data sizes. References: 1. Nilsson et al. "Exploring Cyclone Evolution with Hierarchical Features" https://ieeexplore.ieee.org/abstract/document/9975824 2. Nilsson et al. "Towards Benchmark Data Generation for Feature Tracking in Scalar Fields" https://ieeexplore.ieee.org/abstract/document/9975817 Other relevant classification codes for this project: 10204: Human Computer Interaction 10508: Meteorology and Atmospheric Sciences