North Atlantic Oscillation (NAO)
I define the NAO as the first (varimax rotated) EOF of standardized daily 500 hPa height anomalies in an extended North Atlantic domain (20N-70N, 135W-45E) from 1979 through 2008. This pattern, shown below, is then projected onto recent and forecast data to create the index. The index is standardized by the mean and standard deviation of the base index (again, 1979 through 2008), which is just the first PC. Note that my definition is similar to, but not the same as, CPC's definition. My index correlates with theirs at roughly r = 0.76. Arctic Oscillation (AO)I define the AO as the first (unrotated) EOF of standardized daily 700 hPa height anomalies poleward of 20N from 1979 through 2008. This pattern, shown below, is then projected onto recent and forecast data to create the index. The index is standardized by the mean and standard deviation of the base index (again, 1979 through 2008), which is just the first PC. Note that my definition is similar to, but not the same as, CPC's definition. My index correlates with theirs at roughly r = 0.75. Pacific/North American PatternI define the PNA as the fourth (varimax rotated) EOF of standardized daily 500 hPa height anomalies poleward of 20N from 1979 through 2008. This pattern, shown below, is then projected onto recent and forecast data to create the index. The index is standardized by the mean and standard deviation of the base index (again, 1979 through 2008), which is just the rotated PC. Note that my definition is similar to, but not the same as, CPC's definition. My index correlates with theirs at roughly r = 0.65. Tropical ProjectionsTropical projections use a modified version of Paul Roundy's algorithm (explained here). The most notable change I've made is introducing a time period difference of 10 days between the upper end of the Kelvin waves (20 days) and the lower end of the MJO (30 days). This forces a separation between the two modes. Recent research suggests that this may not be a physically accurate separation, but for the purposes of tropical forecasting, I believe the separation helps to better understand the tropics. Horizontal MapsThese maps are very similar to Paul Roundy's OLR Diagnostics. I use the CFSv2 upward longwave radiation flux instead of OLR to produce these maps.
This is modeled after the Wheeler and Hendon (2004) RMM phase space and is most similar to a similar approach that I've written about in MacRitchie and Roundy (2012). This approach uses the CFS MJO projections, as used in other figures on this website, to track the MJO. This results in a smoother signal than the WH04 index with less interference from other modes. |