Applicability of data mining algorithms in the identification of beach features/patterns on high-resolution satellite data

• Type: Article in International Scientific Journal
• Date: 2015

Authors

Ana C Teodoro (Author) FCUP

Journal

Title: Journal of Applied Remote Sensing

Vol. 9

Final page: 095095

ISSN: 1931-3195

Publisher: SPIE

Indexing

ISI Web of Knowledge - 11 Citations

Scopus - 12 Citations

Scientific classification

FOS: Engineering and technology > Environmental engineering

Other information

Authenticus ID: P-00A-62T

DOI: 10.1117/1.jrs.9.095095

Abstract (EN): The available beach classification algorithms and sediment budget models are mainly based on in situ parameters, usually unavailable for several coastal areas. A morphological analysis using remotely sensed data is a valid alternative. This study focuses on the application of data mining techniques, particularly decision trees (DTs) and artificial neural networks (ANNs) to an IKONOS-2 image in order to identify beach features/patterns in a stretch of the northwest coast of Portugal. Based on knowledge of the coastal features, five classes were defined. In the identification of beach features/patterns, the ANN algorithm presented an overall accuracy of 98.6% and a kappa coefficient of 0.97. The best DTs algorithm (with pruning) presents an overall accuracy of 98.2% and a kappa coefficient of 0.97. The results obtained through the ANN and DTs were in agreement. However, the ANN presented a classification more sensitive to rip currents. The use of ANNs and DTs for beach classification from remotely sensed data resulted in an increased classification accuracy when compared with traditional classification methods. The association of remotely sensed high-spatial resolution data and data mining algorithms is an effective methodology with which to identify beach features/patterns. (C) 2015 Society of Photo-Optical Instrumentation Engineers (SPIE)

Language: English

Type (Professor's evaluation): Scientific

No. of pages: 19