碩士 / 國立中山大學 / 海洋科學系研究所 / 103 / This study validates the Sea Surface Temperature (SST) reanalyses data from two ocean models around Taiwan (15 °N to 30 °N, 115 °E to 130 °E) for the year 2012. The two models are the US Naval Research Laboratory (NRL) Global Navy Coastal Ocean Model (GNCOM), and the HYbrid Coordinate Ocean Model (HYCOM). Both models had the most advance data assimilation techniques including satellite altimetry and SST data. The spatial resolutions of GNCOM and HYCOM are 12.5 and 8 km, respectively. The observational SST data include the satellite SST from the Group for High Resolution Sea Surface Temperature (GHRSST) and buoys maintained by Central Weather Bureau (CWB).
The parameters in statistical analysis include mean error (ME), standard deviation (S), root mean square errors (RMSE) and correlation coefficient (R). The first step was to validate SST data of GHRSST against CWB buoys. The RMSE between CWB buoys and GHRSST data are less than 2 ℃, and the correlation coefficients (R) are higher than 0.88. The GHRSST data, therefore, are considered meaningful for the study region around Taiwan. The next step is to validate SST data of GHRSST against GMCOM and HYCOM. The results show larger RMSE between models and GHRSST in two regions, the Taiwan Strait and the northeast of Taiwan especially in winter (January to March). Further analyses of the modeling data suggest that the large errors were attributed to two major reasons. Firstly, the southward China Coastal Current (CCC) in winter was not well resolved in the modeling, likely due to assimilated forcing. Secondly, the coarse bathymetric data at the northeast of Taiwan was unable to produce topographic upwelling which has been observed years ago.
Identifer | oai:union.ndltd.org:TW/103NSYS5276013 |
Date | January 2015 |
Creators | Hsin-shuo Peng, 彭信碩 |
Contributors | Yu-Huai Wang, 王玉懷 |
Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
Language | zh-TW |
Detected Language | English |
Type | 學位論文 ; thesis |
Format | 126 |
Page generated in 0.0019 seconds