A New Snow Index for Improved Predictions of Winter Weather in the Northern Hemisphere Mid-latitudes

Type: Presentation

Venue: AGU Fall Meeting 2012


Judah L. Cohen; Justin Jones; Swen F. Brands (2012) A New Snow Index for Improved Predictions of Winter Weather in the Northern Hemisphere Mid-latitudes. (Invited). AGU Fall Meeting 2012, San Francisco, CA

Seasonal climate prediction remains a difficult challenge. During Northern Hemisphere (NH) winter the large-scale teleconnection pattern of the Arctic Oscillation (AO) explains the largest fraction of temperature variance of any other known climate mode. However the Arctic Oscillation is considered to be a result of intrinsic atmospheric dynamics or chaotic behavior and therefore is unpredictable. Earlier studies established that fall Eurasian mean snow cover extent is significantly correlated with the winter AO. However we have found that the daily change in snow cover extent is a much better predictor of the winter AO, which we refer to as the Snow Advance Index (SAI). With the development of the SAI we can now explain approximately 75% of the variance of the winter AO. The high correlation between the SAI and the winter AO demonstrates that the AO is mostly predictable, which can be exploited for skillful seasonal climate predictions.

An immediate benefit of the development of this new index is improved seasonal climate predictions. The ability to predict the winter AO is considered the single most important advance in achieving successful winter forecasts. We created cross-validated hindcasts of winter land surface temperatures using the SAI as a predictor in the AER seasonal forecast model and compared those hindcasts with hindcasts using the observed winter AO and ENSO. Skill or accuracy of the AER model compares favorably to that of the observed winter AO and ENSO especially for the Eastern US and large portions of Northern Eurasia. And considering that the index is known four months prior to the winter AO, yet matches the skill of the winter AO so closely, demonstrates great potential for improved real-time winter forecasts. If time permits we will also show that snow cover can help explain decadal variations in the AO and subsequently decadal winter temperature trends.