Statistical Analysis Techniques Applied to North Atlantic Temperature-Salinity Data
The Rational Oceanographic Data center (NODC) has been archiving temperature-salinity data since the turn of the century. In this data archive, and in many others, significant trends and features are hidden within the large quantity of data. This thesis presents a method for the analysis of large da...
| Main Author: | |
|---|---|
| Format: | Text |
| Language: | unknown |
| Published: |
DigitalCommons@URI
1980
|
| Subjects: | |
| Online Access: | https://digitalcommons.uri.edu/theses/2270 https://doi.org/10.23860/thesis-heinmiller-paul-1980 https://digitalcommons.uri.edu/context/theses/article/3243/viewcontent/heinmiller_paul_1980_ocr.pdf |
| Summary: | The Rational Oceanographic Data center (NODC) has been archiving temperature-salinity data since the turn of the century. In this data archive, and in many others, significant trends and features are hidden within the large quantity of data. This thesis presents a method for the analysis of large data bases, enabling the researcher to isolate unique behavior for detailed inspection. In this application, Empirical Orthonormal Function (EOF) Analysis and Cluster Analysis are applied in succession to the HOVC temperature-salinity data base in the North Atlantic. The first three total ocean EOF’s provide for a fit of 98% of the variance. Five clusters are identified, without the use of spatial or temporal variables, using only the first three total ocean EOF’s to locate areas of extreme behavior. Clusters produced by the unbiased computer algorithm coincide with classical regions relating to Mediterranean outflow, Labrador downwelling, and Gulf Stream meanders and recirculation. The EOF’s of each cluster are then calculated and used for further investigation of the physical properties of the identified regions. |
|---|