I wasn't sure exactly where to put this one (also fits into the "March for Science" thread a bit), but it has been bugging me for a week:
http://hosted.ap.org/dynamic/stories/U/US_SCI_WINTER_WEATHER_FORECAST?SITE=AP&SECTION=HOME&TEMPLATE=DEFAULT&CTIME=2017-03-14-17-17-34
For the un-initiated, a "Nor'easter" is a storm system unique to the northeastern US, where a cold front comes in from the north west and collides with warm, moist air coming up the coast. The collision of the air masses produces a severe and rapidly intensifying storm. In the summer they rival hurricanes and in the winter, they produce massive blizzards along the Washington-Boston corridor.
Because they involve a cold and a warm air mass, there is a potentially wide variation in impacts across the storm from east to west. In the east, you might get all rain and in the west it is all snow. There will be a gradient of each, with the center generally producing the most snow, along a swath 10-50 miles wide and up to several hundred miled long.
Last week's nor'easter was late for a snowstorm, which produced a forecasting problem. Early indications were that it would be a classinc winter nor'easter, almost entirely snow, and cutting straight through the population centers from Philly to Boston. But hours before the snow started (Monday morning), the models started showing the warm air from the east would win and produce mostly rain along the coasts and a snow/sleet mix further inland, only producing all snow much further inland. These models were correct. The National Weather Service held a meeting on Monday afternoon and decided against updating the forecasts, "out of extreme caution" (quote) and "...they didn't want to confuse the public." (AP paraphrase).
Wait, what? A coherent message is more important than the quest for accuracy?
So along the east coast, we went to bed last Monday night expecting to wake up to a foot+ of snow and actually finding totals less than half of the low-end of the forecast (NYC predicted: 18-24", actual: 7"). Scientists can claim somewhat of a win in that the mass of precipitation was actually accurate, it was just denser than predicted, but that difference matters a lot in how you respond to the storm. Particularly when eastern Delaware and NJ saw mostly rain instead of a foot of snow! You can't un-cancel school if it doesn't snow. As for me, I did notice something was off when I woke up, but I was sicklazy and stayed home from work on Tuesday though many of my colleagues ended up going in.
This is from a division of the same government agency responsible for collecting, interpreting and disseminating climate data. Which begs the question: is the climate data/warming predictions we get filtered with the same bias?