As alluded to in previous posts, I have been accessing weather data with the goal of calculating the 27 climate extreme indices created by the World Meteorological Organization Commission for Climatology Expert Team on Climate Change Detection and Indices (what a name!).

For today’s post, I decided to show one of the most obvious climate extreme indices: **The total annual precipitation**. This index is abbreviated as PRCPTOT, or the total precipitation that falls each year in millimeters.

We can think of PRCPTOT as an intensity index, or an annual value that describes how much precipitation an area got over 1 year. Since it is an annual value, we cannot differentiate if the precipitation fell evenly throughout the year, or came as a few big events (that’s what the other indices are for!) But what it can tell us is just how “wet” or “dry” a year was based on the average annual precipitation over that time period!

In **Figure 1**, you can see that the average annual rainfall over the Chesapeake Bay region is 1107 mm (that’s 43.6 inches!). Loosely speaking, we can say that years which fall above the loess or annual mean are “wetter than average” years, and below the line are “drier than average” years.

One thing you should also notice, is that there is no linear trend. In other words, the amount of annual precipitation over this time range is not statistically increasing or decreasing. Climate change is not always a gradual linear increase….it is more complicated!

So, we need to look at the PRCPTOT another way to assess changes that could be ecologically important. For example, is the standard deviation changing?

Let me take a minute to explain **Figure 2**. I used the Rollapply function from the zoo package (R-project) to calculate the rolling 10 year standard deviation.

Standard deviation (SD) is square root of the variance and gives a measurement of how spread out the data is. The higher the standard deviation, the more variable the time series is.

In the Rollapply, I calculated the SD for every 10 years, but in a rolling window. This means that the first SD was calculated from 1901-1911, the next SD was calculated from 1902-1912, then 1903-1913, and so on up until 2010.

So what does **Figure 2** show? It shows that the decadal variability of the total annual precipitation is increasing. This infers that the probability of getting a wet year or a dry year is increasing.

Let’s put this in context: Ever notice how one year might be the rainiest you’ve ever seen then the next year is full of clear blue skies? The Hadex2 time series does not indicate that Chesapeake Bay is getting wetter and wetter, rather it is saying that the variability between extreme high or low annual precipitation increasing. (So we might expect more of those wet years followed by drier years….but more work is needed before we can make this conclusion!)

I used the Hadex2 data set to demonstrate this trend since it is 110 years. The Hadex2 is a gridded product that includes many different weather station data but it is an area-weighted average over a 3.75° X 2.5° grid (will miss large localized precipitation events). We will be using multiple data sets to best character the weather trends at the CBNERRS sites, so tune in next time for more on this!