I distinctly remember a respected poster from the Mid-Atlantic region doing a deep dive for dca snowfall for like 30years and pulled in I think at least seasonal average NAO/ENSO values if not monthly for a good portion of his study period to look at correlations. I wish I could find his post because I think he left links to his numerical data/calculations and writeup which was quite good. The thing that always stuck with me about trying to posit correlations with snow with indice values is that some are much more stable on daily, weekly, monthly readings then others (i.e ENSO vs MJO). Most of these indices were invented in the 1970s or later, with limited ability to use distant pass observations to calculate any values outside a best guess for a yearly or seasonal value. The estimates can vary wildly and older studies are super hard to get their data, hell even newer ones can be a bitch to get it coming from an outside academia amateur hobbyist position. Anyways, I distinctly remember how little seasonal and even monthly nao readings statistically impacted snowfall at DCA. I think maybe if one came up with a large multivariate dynamic system that looked at daily readings of all teleconnections to calculate parameters akin to distance, velocity, acceleration, force, momentum for each indice. While taking account for all the variables effect on each other on different time scales to adjust for inherent stability of each indice. Knowing the daily MJO phase is great, but how deep is it in the phase, how fast (absolute & net) in all plot dimensions is it traveling, same with any acceleration over different time periods, as well as distance. Now how does that daily pattern effect NAO (if at all) on different delayed time scales. If you think of a 7 days or 10 days (I e) of MJO readings as a unique song, then with calculated parameters each day as notes, with each parameters calculated parameters daily (as each notes flavors) over time then you slowly built up groupings via graph theory that essentially group all those songs into genres that have their own unique effects on the other indices depending on time scales. A large regional location like northeast, Midwest, ect maybe able to equate their exposure to genres taking into account order and time effects to a slight moderate correlation with regionwide seasonal snowfall. But even on a statewide level it comes down to randomness with a larger structural environment far more than any of us want to admit Sent from my SM-G970U using Tapatalk