Last fall, I watched a course from Yale University called Atmosphere, Ocean and Environmental Change via iTunes U. In one class Professor Ronald Smith showed a graph of forecast accuracy or skill over time. I think it may have been the chart below. Notice that during the time period shown, forecasting skill has continuously increased. The shorter term, 36-hour forecasting skill may hit a plateau, but the longer term, 72-hour skill is on a strong trajectory. Most of this increase comes from increased performance of the computer models. I think everything related to them has improved. Computer speed and capacity have improved over time. The input data quality has improved. The computer programmers have refined the forecast models.
When I saw this graph last fall it made me start to question if skill has really improved for Flagstaff. While I didn’t run an actual study to compare reality to the National Weather Service forecasts for Flagstaff, it felt like they were more accurate than back when I started blogging about the weather on AOL’s old, and now gone blog site.
I think it was roughly a decade ago when I wrote my first blogs on the old AOL site. Back then, it seemed like the National Weather Service couldn’t hit a Flagstaff forecast to save their life. Storms wouldn’t be in the forecast, like the April Fool’s day storm of 1999. As my wife and I bought a used Ford Bronco at Babbitt Ford on the evening of March 31st, it started to snow. It snowed continuously until April 4th and brought our temperatures down to single digits. Those dates included Easter weekend. The Easter weekend forecast had been for a repeat of the weekend before, sunny and warm. How could a multi-day super storm have been completely missed by the National Weather Service?/p>
This single event had a big impact on me. It is one of the early reasons for my interest in watching the weather. I felt like no one else was watching it for Northern Arizona. In fact, someone else was watching the weather. Around that time, I had the privilege to be on a ski lift with one of the forecasters from the NWS out in Bellemont. I asked him about the accuracy. He told me that they were often more accurate in their local office, but that the folks at “headquarters” would often trump their forecasts. That was all I needed, I was off and blogging.
I started to notice on their forecast discussions that if one guy, named Peterson, was doing the writing, the skill was much higher and the language was much easier to understand. I believe he was local and occasionally allowed to get a word in edgewise.
In the time since I watched the Yale course, I have noticed that it seems like the forecast skill has increased. Also, the forecasters at the National Weather Service in Flagstaff now to have more say in what is in the forecast for our area. One example of this is their regularly posted Weather Story images that are in the banner section of the Flagstaff weather.gov site. Here is one from last winter.
So, my hat goes off to NOAA and the National Weather Service for improving their forecasting skill and engaging the local office to improve their forecasts.