The average temperature of the earth has been rising steadily, and climate scientists are ver confident in their prediction that the trend will continue for the foreseeable future. While global conditions can be forecast pretty well, relating global changes to local conditions is much harder. One of the simplest climate predictions is that rising temperatures will lead to fewer days of winter-like weather. Here is an animation of changes in the duration of winter across North America:
Positive numbers indicate longer than average winters. So, winters are definitely getting shorter over North America, and the trend is very consistent since the late 1970s. The 30s and 40s tended to have shorter winters, while those in the late 60s and early 70s were longer. Maine is much noisier, but tends to follow this overall pattern. The period of shorter winters in the 1940s is more extreme in Maine, and only in the last couple of years have we exceeded those values. I would love to hear some recollections of the late 1990s: were winters really ~25 days longer than the last few years? If you're not a fan of cold weather, this looks like a pretty good trend: fewer days in the puffy jacket and fewer days running the old oil burner. However, it's not all sunshine and roses (which of course will bloom earlier). In Maine, mild winters allow the dreaded deer tick to flourish, making gardening an extreme sport.
In the movie, red indicates a shorter winter at that location, relative to the average duration between 1871-2010. I used a very human-centric definition of winter. Winter was declared to start on the day where temperature falls below 5°C and stays below this level for five consecutive days. Winter ended when temperature exceeded 10°C for five days. More detail on the data is below.
Not surprisingly, there is a lot of blue (longer than average winters) in the beginning and lots of red (shorter winters) at the end. In between, you see red and blue blobs come and go. Even early in the time series, there are some areas with shorter winters, and even at the end there are some areas with longer winters. In any one location, winter duration fluctuates, but the general trend is towards shorter winters. For example, I plotted the winter durations for Maine (thin blue line) against the average duration for the whole region (thick black line).
Positive numbers indicate longer than average winters. So, winters are definitely getting shorter over North America, and the trend is very consistent since the late 1970s. The 30s and 40s tended to have shorter winters, while those in the late 60s and early 70s were longer. Maine is much noisier, but tends to follow this overall pattern. The period of shorter winters in the 1940s is more extreme in Maine, and only in the last couple of years have we exceeded those values. I would love to hear some recollections of the late 1990s: were winters really ~25 days longer than the last few years? If you're not a fan of cold weather, this looks like a pretty good trend: fewer days in the puffy jacket and fewer days running the old oil burner. However, it's not all sunshine and roses (which of course will bloom earlier). In Maine, mild winters allow the dreaded deer tick to flourish, making gardening an extreme sport.Data Processing
The data for the animation and the figure were taken from the NOAA Earth System Research Lab's 20th C Reanalylsis (V2). I downloaded the daily maximum temperatures and extracted North America. Starting from midsummer in each year, I went through the next 365 days of data looking for a period where the temperature was 5°C or lower for 5 consecutive days. The first period I found was declared to be the start of the following winter. For example, if I started searching in June 1973, and found December 15 as the first winter day,my algorithm would say that the start of winter in 1974 was -15 days. I then searched until I found five days of temperatures above 10°C and declared that to be the end of winter. The duration was then the difference between the ending and starting dates. If a location never fell below 5°, there was no
winter at that location. Similarly, if the location was always colder than 5°, then the duration was 365 days.
I now have a map for each year with the winter durations. To reduce some of the local variability, I computed a five year running mean at each location (1973 is now the mean of 1973-1977). I then took the average duration at each location in the map and then subtracted the observed duration from the average, producing a map of anomalies. In the movie, I interpolated on to a finer grid, to make the images less blocky. Between each year, I inserted five images that were blendings of this year and the next. This allows the movie to change smoothly.
