Since this is the model output discussion, I've got a question to ask about model variability. I've noticed that the MJO is very hard to model - it does seem like one of the larger sources of model error. Now, given that the areas where the MJO is in an active phase are huge areas (e.g. maritime continent is a pretty big place), it has struck me that the models can get the position of the MJO badly wrong - like off by a quarter of the globe in 10 days. Or the strength forecast can be wrong - sometimes one would forecast the MJO decaying to virtually nothing, and yet another model would predict that several hundred square miles of convection would amplify - within 7-10 days. That does seem to me to be a serious error - to miss a huge area of enhanced convection or to misplace it by a quarter of the globe. Or is the error smaller than what I imagine it to be? So then - why is the MJO such a fickle, difficult to predict beast?