Many of us would like to think that the models can look at the evidence more effectively than we can. They crunch the equations, using the starting data -- the same data you or I would examine by eye, plus other data that we might ignore or not have time to look at. I think Aaron's post said that El Nino development requires not just above-average warm water volume (WWV), but also some westerly wind events during the key times of the year, to set things in motion and keep the event going strong. And those westerly wind events are very hard to predict. So how would a "good" dynamical model know if these wind events will occur or not? Most of the model anomalous westerly winds are responses to SST increases near and east of the dateline. But how do we get the increases without the winds? It's the vicious cycle we call coupling. The random winds can play a large role in getting that coupling started. Some of the models may incorporate the westerly wind events using statistical abbreviations, which assume an average amount of them. This would be similar to how they estimate convective cells (thunderstorms) too small to be modeled directly, since most of them occur in between the model's grid points.

When you say that there isn't a whole lot of evidence other than the model forecasts, it sounds like you're working from an implicit assumption that the models are not necessarily trustworthy, and that we would like to be able to note some obvious signs ourselves in the absence of the model forecasts. Notice that I am not saying I disagree with you! But with this many dynamical models agreeing on a good likelihood of El Nino, it becomes hard to ignore them. On the other hand, around half of the statistical models are not calling for El Nino. Statistical models are more "transparent" than dynamical ones in that the relationships between their predictors and the ENSO-related SST are often understood and trusted by the people who develop those models. And the relationships are usually linear, which makes them simpler. We know from our studies that when faced with the spring ENSO predictability barrier (like right now), many dynamical models tend to be too bullish in their forecasts -- i.e., too confident and strong. It's possible many of the dynamical models calling for El Nino are coming on too strong. Even if we reduce their anomalies by, say, 40%, we still have many of them predicting at least a weak El Nino.

I and other ENSO forecasters have the same questions and doubts that you do. The only possible difference is that we may give the models a little more weight than you do, even if we can't follow their "reasoning" as easily as we can note obvious signs, like WWV, current anomalous low-level winds, etc. So I think your doubts are healthy ones. The way we deal with our doubts is that we assign lower probabilities to an El Nino event coming this summer/fall than the models do if taken literally. Our probabilities are just barely over 50%, while the models imply a greater probablity.