Are we good at predicting the weather where we need to be?

The ability to forecast future weather is very valuable to society, and as someone who does not follow meteorology very closely it seems like we are quite good at it, I trust the projected weather for the next few days almost absolutely and cant remember any time when it was very wrong. Because of my perception of strong accuracy, it was surprising for me to learn that there is still alot of value in projecting small deviations in weather from forecasts. While doing background for this article, I read a story about weather prediction experts receiving million dollar offers to join trading firms. When you realize how valuable the weather can be in predicting economic outcomes like gas demand or the yields of agriculture, this starts to make sense. However in reading about the benefits to firms in accurately forecasting the weather I began to question to what extent these benefits extended to governments, particularly agriculture or car heavy regions. More importantly it turned out, research has shown that accurately predicting the weather has a high impact on fatality rates, with even small errors leading to excess hundreds or even thousands of excess deaths, as people expose themselves to hotter or colder temperatures then they anticipated. Because of this I wonder if older/more vulnerable to weather regions will have stronger forecasts all else considered then regions with younger/healthier populations.

First I had to try to figure out a way to fairly judge the accuracy of each state, which I did by first comparing the error on a given day to the average error of days with its weather condition(ie compare sunny days error to the average error on a sunny day), then I found key predictive variables such as the cities average annual precipitation and the variance of its temperatures etc, and factored them in to the predicted error. While this model is far from perfect im sure, I hope it does a better job of fairly comparing the performances of each city then the raw data. While finding some of the most the predictive variables, I was surprised to learn that on average areas that receive more rain each year are much better at predicting the weather, which I was not expecting.

Without a dataset of the average age by city/health condition, I resolved to make a chart of the top 10 best and worst cities and kind of eyeball to see if I could see an association with age.

Unfortunately for efficiency, there does not seem like there exists a large relationship between the age of a cities population and its relative accuracy in predicting the weather. Also there is several states with 2 of the best/worst cities at predicting weather but only 1 state(Wyoming) with one of each, which likely suggests my adjustments have not done a good enough job at accounting for the increased difficulty of predicting certain areas(like Wyoming or Montana).

On the bright side, the worst city at predicting the weather in my model(Fairbanks,AK) has a median age of only 29 years old, 10 below the national average. Unfortunately all of the best cities(except for Buffalo,WY which has a median age of 48), are also below the median age, indicating their excellence in forecasting isn’t having the maximum possible impact.

I also made a map grouping the cities by state, which shows a little more promise. The older northeast seems to be broadly abve average, while the younger south seems to struggle, aswell as the young state of Alaska.Unforunatly New Hampshire and West Virginia, 2 of the oldest states, have below average accuracy, while also being the worst in their immediate region. Additionally the relativly young midwest is widely above average, as is the ridiculously young state of Utah.

NWS funding is allocated by the federal government, and while obviously it should definitely take into account other variables, like population, I think its important to also allocate more funding to areas with the most vulnerable populations to deviations in weather.