You might be right, but I don't think so. If you look at the screenshot of the analysis tool in the blog post, you can see that even through all of the noise there is a *very* clear curve for Hurricane Sandy. Although, we do know that the data is very noisy and we are working with some professors who are researching this very topic (calibration of phone sensors for weather data collection). I think we will have useful data for short-term prediction.
But if not, that's okay too. Science!