It may be potentially important issue since observed decadal SST variability in the wintertime over the North Pacific is concentrated in the subarctic and subtropical OFZ (Nakamura et al. 1997). It is known that SST in mid-1atitude North Pacific dropped significantly around 1976 and tended to be lower than the long-term average until 1988 (e.g., Nakamura et al. 1997). This‘cool’period was followed by‘warm’period until mid-1990s with higher than average SST. Thus, while 16 years-long time series is insufficient to detect decadal cycle, it still may by useful to document changes in OFZ associated with this specific climatic shift. This paper presents some preliminary indications of North Pacific subpolar and subtropical frontal zones (hereafter referred as SPFZ and STFZ, respectively) response to above-mentioned decadal climatic changes.
DATA
The SST observations used for this study are the nighttime weekly mean multichannel SST (MCSST) at 18km resolution. The MCSST for 1982-97 have been produced by the University of Miami Rosenstiel School of Marine and Atmospheric Science from the National Oceanic and Atmosphere Administration/National Environmental Satellite Data and Information Service (NOAA/NESDIS) global retrieval tapes of advanced very high resolution radiometer (AVHRR) measurements (Smith 1992). For each grid point, the SST product is computed as the average of all MCSST measurements available for 1 week. Data void areas are then filled using a Laplacian interpolation technique. This interpolated product proved to be useful to address intraseasonal to seasonal variability in OFZ (Kazmin and Rienecker 1996). However, it should be used with caution for decadal-scale variability study since the amount of interpolation applied to each weekly composite is crucial issue. Indeed, the more missed pixels we have the more interpolation should be applied to fill them in, which results in a smoother SST field. Consequently, smoother SST field produces weaker SST gradients. Thus, significant trend through the whole time series in the number of missed pixels to be filled in may create artificial trend in SST gradient magnitude. To check out the data we calculated the percentage of missed pixels (both global and within specific latitudinal boundaries) in each weekly composite and find out that it have been decreasing almost linearly. Typically, SST products for 1996-97 contain about two times more valid (actually measured) grid points compare to 1982-83 which means actual improvement of data quality. However, calculated from interpolated data global 5-year mean SST gradient map for 1991-95 shows almost 20% overall increase in SST gradient magnitude compare to the same map for 1984-88.
