Sequential Self-tuning Clustering for Automatic Delimitation of Coastal Upwelling on SST Images

• Susana Nascimento ^{[1] [1]} ; Sayed Mateen ^[1] ; Paulo Relvas ^[2]
  1. [1] Universidade Nova de Lisboa

     Universidade Nova de Lisboa

     Socorro, Portugal

     
  2. [2] Universidade do Algarve

     Universidade do Algarve

     Faro (Sé), Portugal

     
• Localización: Intelligent Data Engineering and Automated Learning – IDEAL 2020. 21st International Conference: Guimarães, Portugal; November 4–6, 2020. Proceedings / Cesar Analide (ed. lit.), Paulo Novais (ed. lit.), David Camacho Fernández (ed. lit.), Hujun Yin (ed. lit.), Vol. 2, 2020 (Part II), ISBN 978-3-030-62365-4, págs. 434-443
• Idioma: inglés
• Enlaces
  • Texto completo
• Resumen

  • Upwelling is of major environmental and economic importance for coastal regions. Sea Surface Temperature (SST) satellite imagery provide an expedited method of monitoring its variability.This work proposes a one-by-one extracting version of a spatial clustering algorithm with self-tuning thresholding derived from anomalous clustering, able to precisely delineate coastal upwelling from SST images. The stop condition is defined based on properties of the phenomenon and allows to model the appropriate number of upwelling regions.The algorithm, Sequential Self-Tuning Seed Expanding Cluster (SSTSEC), shows to outperform the homologous sequential version of Seeded Region Growing (SRG) on the automatic delimitation of coastal upwelling from a collection of 207 SST images comprising two distinct upwelling systems: from the Portuguese coast and from Canary upwelling system. Four popular internal clustering validity indices were combined to measure the quality of the results.