Some of you may recall my post last winter about a simplistic
scoring system for chase seasons I developed. My motivation was eventually to use these scores as a basis for analog forecasting. Recently, I've begun to explore this by calculating correlation coefficients between climate/teleconnection indices and the scores I've calculated. Thus far, I've done so for these indices: AO, NAO, ONI, PDO, PNA, QBO, WPO. For each index, I'm looking at correlations for mean index values at various lead times and durations.
Because the sample size is N~60 (complete Storm Data, and thus my scores, are only available back to 1955), correlation coefficients of |rho| > 0.21 are considered statistically significant. At lead times of several months (i.e., climate index values during the wintertime), two indices appear to have statistically significant correlations to the quality of the subsequent Plains chase season: the Arctic Oscillation (AO), and the Pacific Decadal Oscillation (PDO). Specifically, a positive correlation exists for the AO, and a negative correlation exists for the PDO.
In very simplistic terms, the
Arctic Oscillation represents the surface pressure gradient between the Arctic and the sub-Arctic north Atlantic (i.e., between the northeast US and western Europe, equatorward of 50°N). A positive AO indicates a relatively strong gradient, with low pressure anomalies over the Arctic and high pressure anomalies to the south; this is associated with relatively strong westerlies over high latitudes, which helps to limit Arctic air intrusions into the midlatitudes (at least on average over the northern Hemisphere). A negative AO indicates the opposite, with weakened westerlies around the Arctic Circle and a greater tendency for Arctic air to spill southward.
The
Pacific Decadal Oscillation is based entirely on SST anomalies, and can be thought of similarly to ENSO, but calculated for the northern Pacific (specifically, along and near the North American coast from around Oregon up to the Gulf of Alaska). A +PDO is associated with warm SST anomalies, while a -PDO is associated with cold anomalies.
This winter, the mean value for the AO has been positive, despite the frequent cold intrusions into the eastern half of the CONUS during Jan-Feb. This can be seen as a "good sign" for the chase season, as rho(PDO_Dec-Jan, Score_Mar-Jun) = +0.33. The correlation for the AO actually peaks at this lead time, becoming less significant as we move into Feb-Mar-Apr (which makes sense, as the AO is primarily a wintertime mode).
On the other hand, the PDO is literally off-the-charts positive this winter: SST anomalies along the northern Pacific coast have been the warmest on record since 1900 for Dec-Jan. This can be seen as a "bad sign" for chase season. One difference between the AO and PDO correlations, however, is that the negative PDO correlation continues to increase in magnitude as we move into spring, peaking for the April-June period (rho = -0.30). Thus, if the +PDO can somehow fall off rapidly over the next 60 days, this may not be as much of a factor. That seems unlikely, though, as warm SST anomalies over the far eastern Pacific have only increased between
late January and
now.
Looking a little more closely at sub-regions of the Plains, the coefficients would imply that a +AO is more "beneficial" to the northern Plains than southern Plains, whereas a +PDO is more "harmful" to the southern Plains than northern Plains. However, I'm even less comfortable making claims about smaller spatiotemporal windows, as the scores tend to be "sparser" (i.e., for a small region over a short period of the year, there are lots of quiet years and only a few active years). Even for the Plains-wide coefficients I discussed above, the limited sample size and modest values (|rho| ~ 0.3) means any practical inferences should be made cautiously. I figured I'd share just to break up the winter doldroms a bit, though. I'll be sure to post again if I have any new findings of significance - and as always, comments/questions/complaints are welcome.
