Research Activity 1 Updates

July 2020

Hello! My name is Ty Dickinson, and I am a Ph.D. student working on the PRES2iP team. I am originally from Port Huron, Michigan, but I have been living in Oklahoma for several years now, I received my B.S. in Meteorology from OU in 2018. 

My first research priority was defining criteria to identify extreme precipitation events on the subseasonal-to-seasonal (S2S) timescale. This timescale is defined as 2 weeks to 3 months. The key thing to keep in mind is that our criteria will distinguish between events that saw a large amount of rainfall over only 2 or 3 days and events that were truly extreme over a period of 14 days. 

There are a few criteria we used to identify extreme precipitation events. First, we defined the amount of precipitation for an event to be considered extreme, accounting for changes in the climate. Throughout the Lower 48 states, heavy rainfall events are becoming heavier with time. Using this information, I have defined a value for “extreme” at each location over 14 days for each 14-day window throughout the year (e.g., January 1 – January 14, January 2 – January 15, etc.). This threshold captures the  extreme values, even if those values are changing over time.  

Each point also had to meet a second criterion relation to event timing in order to be considered extreme. I calculated an average daily precipitation amount specific to the 14-day period at each location. For a point to meet this criterion, at least half of the days (7) had to have experienced precipitation at or exceeding the average amount. In summary, a precipitation event at a specific location in the U.S. is flagged as extreme if its 14-day precipitation total exceeded the 95thpercentile and if 7 of the 14 days recorded above normal daily precipitation.

Because the first step only identified extreme values at points in the U.S., I developed a method to group the points into a region of extreme precipitation. To accomplish this, I used a statistical technique to draw polygons around regions with extreme precipitation events. Figure 1 displays an example event from the database I have created.

Figure 1: Example 14-day extreme event depicting the Ohio River flood of January 1937. Shading depicts (a) the recorded total precipitation (in millimeters, mm) between January 12 and January 25, (b) the 95th percentile as calculated from a quantile regression model (in mm), and (c) the point locations that were flagged as extreme. The red contour in (d) outlines the region defined as extreme.

I am currently taking all the 14-day events I found and am grouping them into clusters to see if there are naturally occurring regions or tracks that extreme events tend to follow. From here, I will be working with others in the group to examine the large-scale drivers of the events.