1Collaborative Research Student, Korea Maritime and Ocean University OST 2Principal Research Scientist, Marine Bigdata Center, Korea Institute of Ocean Science and Technology
Corresponding author:
Hong-Yeon Cho ,Tel: +82-51-664-3786, Email: hycho@kiost.ac.kr
Received: October 1, 2025; Revised: October 20, 2025. Accepted: October 22, 2025.
Share :
ABSTRACT
Statistical tests of the independence for extreme data under varying block sizes are carried out. The data used in this study were 30 minutes-interval hindcast wave data at the KMA and KHOA buoy installation sites in the Ulleungdo-Dokdo waters. In addition to independence tests for extreme data extracted under various block sizes, we also performed correlation analyses between median wave height and maximum wave heights, because the mean wave height exhibit distinct seasonal variation patterns. Statistical testing revealed that the AM technique satisfies all independence conditions for extreme data. Furthermore, when the number of blocks per year was increased, independence was preserved up to approximately three to four (i.e., 3~4 blocks per year), whereas a larger number of blocks led to a violation of the independence assumption. Correlation tests between the maximum and median of the block sizes also yielded similar results. These analyses suggest that extracting extreme data under block sizes of two to four (seasonal time scales) can cope with the small data set limitations of the AM technique to some extent and contribute to a reduction in the estimation error range as the number of data increases.
PDF Links
Full text via DOI
Download Citation 
