Extracting Tsunami and Short-Period Motions from HF Doppler Radar Data through Empirical Orthogonal Function Analysis

High-frequency Doppler radar (HFDR) maps surface currents over an area, providing information on current patterns otherwise impossible without a full array of other current meters. Empirical orthogonal function (EOF) analysis, which rotates data into a set of orthonormal modes with maximum variance in the fewest modes, was used to explore surface current data; using subsets of the spatial and temporal domain allowed specific types of motion to be targeted. Two years of HFDR data bandpass filtered from 8 to 120 min from south of Oahu, Hawaii, were projected onto EOF modes calculated over shallow Penguin Bank during the 2011 Tohoku tsunami to form indices representing the activity of those modes. Analysis of the time series of these indices yields information about these modes being active due to other forcings (e.g., meteotsunamis, storm surges). EOF analysis was also done using data from the northern Adriatic Sea (Italy) and Mindoro Strait (Philippines) selecting times and locations to isolate the response to specific forcings such as bora events and typhoons, based on high-resolution atmospheric models (COAMPS) and scatterometer data. In addition to indices based on scalar-projections as described above, maps of correlations between time series data and EOF time series were used to highlight expression of EOFs outside their spatial calculation domains.