[H-TT15] Geographic Information Systems and Cartography

[HTT15-06] Enhancing Golf Course Detection in Vietnam: NDVI Analysis and Feature Recognition with Sentinel-2 Data

Kim-Anh Nguyen^2,3, *Yuei-An Liou^1,3 (1.National Central University, 2.Vietnam Academy of Science and Technology, 3.Taiwan Group on Earth Observations)

Keywords:NDVI (Normalized Difference Vegetation Index), Sentinel-2, geospatial data, Golf Course Detection, Remote Sensing

This study examines the impact of Vietnam’s tropical-subtropical climate on NDVI-based golf course detection and evaluates feature recognition techniques using Sentinel-2 data. With temperatures ranging from 15°C to 38°C, Vietnam’s vegetation exhibits distinct NDVI responses compared to temperate regions like Yamaguchi, Japan. The resilience of native flora and turfgrass reduces the need for intensive maintenance, contrasting with cooler climates. Additionally, the extensive rice paddies in Hanoi introduce NDVI fluctuations, posing challenges for golf course identification. Empirical field validation is essential to refine NDVI methodologies for improved accuracy.
Feature recognition using Sentinel-2 imagery proves highly effective, achieving 98.41% accuracy and a Kappa coefficient of 0.9665. The distinct spectral properties of turf grass and identifiable bunker features enhance classification precision. However, Landsat data presents challenges due to spectral limitations, leading to potential misclassifications between rice fields and turf grass. Additionally, its lower resolution complicates bunker detection, reducing reliability. These findings highlight the advantages of high-resolution satellite data and the need for advanced classification techniques to optimize golf course detection across diverse environments.