One of my favorite parts of science is the investigative aspect.
That is, being presented with a question, then using data to come to an answer.
Since it is Friday, and I have been working on a return period analyses, I indulged myself and tackled a question that I myself was curious to investigate. “How many rainy days do we get each summer and has that number changed?”
Background and Motivation
To many living in the Chesapeake Bay region, summertime is filled with memories of going to the beach, maybe going to a park, toasting marshmallows by a campfire, laying in the sun by a pool…. most of the activities I can think of involve the outdoors!
But we all know what happens on a rainy day: we get confined indoors.
The fluctuation between rainy and sunny days can affect local tourism and influence critters living in shallow water environments. For example, variability in precipitation can alter the flux of nutrients and organic matter from land into the coastal environment and also changes the available photosynthetically active radiation needed for photosynthesis, among others.
My motivation for this question was actually from an economic contemplation, but it is certainly applicable to ecological impacts on vital habitats in the near-shore Chesapeake Bay system, such as the marsh!
Ready for my reasoning: “I hope it doesn’t rain during my beach trip next weekend!” This led me to wonder, how often do we have rainy days in summer, and has that historically changed?
First, I must define what I called a rainy day. Since I am using the ETCCDI extreme climate indices, it only made sense that I use their definition!
The R10mm index defines a wet day as a day in which at least 10 millimeters of liquid precipitation fell (that’s about 0.40 inches of rain!). As a note, the R10mm index is reported as the annual count of wet days.
Next, I choose a weather station very close to Jug Bay, Maryland (Figure 1). The Upper Marlboro NCDC-Daily weather station has daily precipitation data available since 1956, giving us a pretty long record for this investigation!
For the analysis, I calculated the 2-year, 5-year, 10-year, 25-year, and 50-year return period for wet days at Upper Marlboro. This will give us an idea of what a “common” amount of annual wets days (2-year) and a rarer amount of wet days (50-years) looks like.
Since I did all the calculations myself rather than use an R package, I had the ability to gather some more information that one would normally get using these extreme climate indices. Thus, I would able to determine when the 50-year return periods occurred and, importantly for my question, what percentage of these wet days occurred in the summer.
For a more straightforward calculation, I defined summer and all the days in June, July, and August!
The Results: What’s the answer to my question?
The first thing you may notice about the annual count of wet day return periods at Upper Marlboro (proxy to Jug Bay), is the spread. About 15 days, or half a month, separates a 2-year from a 50-year return period. Is this a big spread or a little spread….what do you think?
But what does this data actually mean to you? In a very (very) general sense, we can think of the 2-year return period as the “average” amount of wet days. (The actual mean in the data series is 34.8 days). That is, in any given year, we can expect between 34 or 35 wet days…. that’s about 10% of the year!
However, as climate scientists, we know that what we expect and what we get are two very different things! That 50-year return period tells us that on a given year, we have a 1/50 chance (2%) of having a year with about 50 wet days.
You may ask, especially if you look at Figure 2, in which years did we have a 50-year wet day event? Well, at the Upper Marlboro station, that occurred in 2003 and 2009. It is interesting to think that the two years with the most wet days have occurred within the last 11 years (2014 as the end year) in a 58 year record.
Now, what percentage of the summer is considered a wet day? (This is simply the ratio of wet days in June, July, and August divided by the total annual amount). It turns out, in the Jug Bay region, an average of 28±0.08% of summer days are wet by our definition. The range is from 12 to 52%.
Now, of course, this does not mean that the entire day was cloudy and rainy. Since our definition is simply that at least 10 mm of precipitation fell, this could be a gloomy day with a consistent sprinkle or a quick one hour downpour. Regardless, it gives us insight on how many wet days we may get in the summer!
Now for my final question, has this percentage of summertime wet days changed at all over this time series? It was pretty clear to me looking at Figure 3 that the answer would be no, but I checked it for significance as well….yup, at least in this location, the percentage of wet summer days has not appeared to change!
Lastly, how about the variance of wet days? I took a moving decadal variance over the summertime percentage of wet days and found no apparent changes in variance as well. In fact, this variance looks to have a fairly cyclic pattern.
Things to keep in mind: this is just one station with a (meteorologically speaking) short time series! So I am not drawing any conclusions, just providing insight!
While this may seem like a “detour” in our on-going analysis, it gave me some good R code to use as a foundation for future analyses. I have spent part of this week calculating the regional return periods of extreme indices, and we be repeating that analysis on some tidal gauge and streamflow data!
My ultimate hope is to relate the return periods to marsh flooding. But at face value, data such as that from Table 1, are ecologically and economically important values that do not exist for the Chesapeake Bay Region.
As a side note, I could do an economic evaluation for what this result means to the local communities in Maryland and Virginia…but that is way out of the scope for this project!