It's true that three of the NMME models are indicating a stronger La Nina this autumn and winter - the disagreement among the NMME models is a bit striking considering the short lead times. However, I note that the NASA model has been biased cold for many months now, so I would discount that forecast. Also, the ocean analysis used to initialize the CFSv2 (GODAS) has been a bit colder than other initializations, and it may be due to missing TAO array data - the CPC Monthly Ocean Briefing has some discussion of this. The bottom line is that I am skeptical of the very cold forecasts, but I agree that these possibilities should not be dismissed outright.