Even the originator of chaotic contributions to climate and the Butterfly Effect, Edward Lorenz, had concerns about claiming all aspects of climate were chaotic. This included of course tides :
but also QBO & ENSO (El Nino) :
A recent article even read more into the Lorenz quote statement:
> "After seminal researches of E.N. Lorenz, almost all meteorologists agree that weather variations are chaotic, i.e., they are unstable to small disturbances, and so unpredictable for more or less distant future. Moreover, the same opinion is widely accepted among climatologists concerning climatic variations (Ghil 1985 ); however, Lorenz himself said (Lorenz 2006 ): *"I do not know: whether are climatic variations chaotic or nonchaotic."* The main aim in this paper is to demonstrate that short-term climatic variations (the periods from 2 years to about one decade), even if they look to be very complex(strange in terms of mathematics), are nonchaotic. Instead of chaos, a mutual order exists in these variations. The existence of such an order admits to predict the variations with no predictability limit in principle." --I. V. Serykh and D. M. Sonechkin, “Nonchaotic and globally synchronized short-term climatic variations and their origin,” Theor Appl Climatol, pp. 1–18, Jan. 2019.
I always had an optimistic goal of making climate variability as easy to predict as the tides, to where we could train the past time-series and use that as a predictor similarly to tidal analysis. This seems like an obvious approach to try, but I can see how researchers would miss the necessary ansatz required to make it work.