Coincidentally, Zhang et al. first presented the results of their Lili forecast analysis at a NOAA-sponsored GPS-RO workshop on 24 Aug 2005, six days before Katrina struck the US coast. This plot shows the error of the forecast path, determined by comparing against the actual path, for 36 hours after the occultation was acquired. At 36 hours the forecast path error was reduced from 219 km to 128 km with the addition of the GPS profile.