Connectivity: What are the relationships between our indices?

The ocean affects our climate.

Figure 1: The average location

Figure 1 was borrowed from NOAA’s website. It shows how the seasonal, wintertime, El Nino from previous years can affect larger scale climate patterns in the United States.

That statement is something you learn from day 1 in any oceanographic or weather studies class. While we know that other natural and human-caused factors also affect climate (i.e. greenhouse gases, volcanic eruptions), our ocean has a “pulse” of natural, cyclic variability that influences our weather.

Many scientists have noticed oscillations (cyclical patterns) in sea surface temperature that change between positive and negative phases on decadal and multidecadal time scales. For example, the Atlantic Multidecadal Oscillation had “cool” phases from 1900-1920 and 1960-1980 and a “warm” phase from 1930-1950 (Knight et al., 2006).  This warm phase, for example, has been noted as a period of decreased rainfall and streamflow in parts of the United States. These variations in the sea surface temperature and sea level pressure (in this case, the Atlantic), can influence the larger scale climatic processes, such as rainfall (Nye et al., 2014).

Table 2: The indices we present in this post!

Table 2: The indices we present in this post! The bold text indicates what will be highlighted below.

As another example, Figure 1 illustrates how the wintertime El Nino can influence general weather patterns in the United States. What is important to note here is that the natural variation of tropical Pacific sea surface temperatures can affect the weather we experience on the east coast!

So, it makes perfect sense that we would want to look for any correlation between these larger scale climate teleconnections with our extreme climate indices and streamflow (Table 1). Positive and negative correlations between these extreme climate indices and indices such as the AMO and NAO will give us insights on how extreme climate events are connected to this natural climatic variability.

Annual Correlations

Figure 2:

Figure 2: A correlation matrix for our extreme climate indices! Which correlations make you excited?

Figure 2 is a correlation matrix for all annual-based extreme climate indices, select teleconnection indices, and 3 major tributaries into Chesapeake Bay (Table 1).

How to interpret this figure:

1) Gray stars mean the correlation was insignificant (p-value > 0.05), so we will ignore any correlation and assume there was no trend.

2) The size and color of the squares tell you the same thing: the strength and direction of the linear relation, also called the correlation coefficient (r). So the bigger the square, and closer to red or blue, the stronger the fit.

3) Speaking of red versus blue: The color indicates the direction of the linear relationship where warm colors are positive and cool colors are negative.

Below I will briefly acknowledge and discuss a few significant correlations. Feel free to add your favorite relationship in the comments!

Frost Days

Frost days is the annual count of days when the daily maximum temperature was below 0°C (32°F). Thus, you can think of this index as the number of days when frost could likely occur.

The raw time series and 21-year moving average for Frost Days using the HadEX2 time series.

Figure 3: The raw time series and 21-year moving average for Frost Days using the HadEX2 time series.

So, it is no surprise that Frost Days are negatively correlated to the Growing Season Length (r= -0.72) and the Warm Spell Duration Index (r= -0.44). This makes sense, more Frost Days in a year results in a shorter Growing Season Length and fewer Warm Spells. Our previous work has found that Frost Days, in general, appear to be decreasing in the near-shore Chesapeake Bay region (Figure 3)…suggesting that this trend may be associated with an extension of the Growing Season Length and a higher probability of having a warm spell!

Interesting, Frost Days was also significantly negatively correlated to the NAO (r= -0.26), EAO (r= -0.27), and AO (r= -0.42). In general, a negative, or “cool” NAO phase is associated with cooler weather and sometimes more snow. Could we then hypothesize that a negative NAO phase could imply we would have more Frost Days? This may help explain the variability in our data!

Summer Days

Figure 4

Figure 4: The raw time series and 21-year moving average of the Summer Days index using the HadEX2.

The Summer Days index is the annual count of days when the daily maximum temperature is greater than 25°C (77°F). Our previous analysis found that this Summer Days index appeared to follow a cyclical pattern and had no apparent rate of change in the historical record. But that does not mean it will not have any correlations to our other climate indices or teleconnections!

Figure 2, part 2: The same correlation matrix, just with the correlation coefficient (r) as a percent.

Figure 2, part 2: The same correlation matrix, just with the correlation coefficient (r) as a percent.

Summer Days were positively correlated to the AMO index (r= +0.29) and EAO index (r= +0.29), indicating that a “warm” phase AMO could indicate more warm summer days. The warm AMO phase is often associated with extended droughts, such as the 1930’s Dust Bowl (Nigam et al., 2011). We can see from Figure 2 that Summer Days was positively correlated to Consecutive Dry Days (r= +0.28), but there was no significant correlation, on an annual basis, between Consecutive Dry Days and the AMO idex.

The set of “connections” that I found most exciting were the negative correlations between Summer Days and the three rivers selected for this study; Summer Days were negatively correlated to the Susquehanna (r= -0.15), Potomac (r= -0.23), and James (r= -0.24) Rivers.

What this suggests to me: warmer temperatures in the summer season is associated with more evaporation. This process has been hypothesized to decrease summertime streamflow into the Chesapeake Bay (Najjar et al., 2010). This initial finding appears to agree with that hypothesis: the more warm days we have, the more potential for enhanced evapotranspiration, thus a weaker streamflow.

Wet Days

Figure 5:

Figure 5: The raw time series and 21-year moving average of the R10mm index using the hadEX2.

The R10mm index is the annual count of days when at least 10 mm (0.4 inches) of precipitation fell. Our previous work (Figure 5) has suggested that this index has historically increased.

While there were no significant correlations with the teleconnection indexes, there were some pretty strong correlations observed. This R10mm index was highly correlated to the total annual precipitation (r= +0.95) as well as the annual exceedance of the 95th percentile of precipitation (r= +0.65).

This index was also highly correlated with streamflow, which makes sense: more precipitation means more water which means a stronger streamflow! This connection helps us to understand that changes in precipitation will affect the streamflow entering Chesapeake Bay.


The R99p index from the HadEX2 time series with some noted major precipitation events.

Figure 6: The R99p index from the HadEX2 time series with some noted major precipitation events.

The R99p index is the annual exceedance of the 99th percentile of precipitation. Percentiles are often used to describe how well you did on a test; if you scored within the 99th percentile, then you did better than 99% of those test takers (Great job!). For precipitation, exceeding the 99th percentile means we got a lot of precipitation!From our historical calculation of the R99p index, many of the “extra wet” years can be traced to huge hurricanes, blizzards, and tropical storms (Figure 6)

This index was positive correlated to the AMO (r= +38),  suggesting that a “warm” AMO phase could bring some “extra” rain in the form of major events.


What other neat correlations do you see?

Are there any that make you scratch your head and have to ask “is that a coincidental correlation?”


Works Cited

Knight, J. R., C. K. Folland, and A. A. Scaife (2006), Climate impacts of the Atlantic Multidecadal Oscillation, Geophys. Res. Lett., 33, L17706, doi:10.1029/2006GL026242.

Najjar, Raymond G., Christopher R. Pyke, Mary Beth Adams, Denise Breitburg, Carl Hershner, Michael Kemp, Robert Howarth et al. “Potential climate-change impacts on the Chesapeake Bay.” Estuarine, Coastal and Shelf Science 86, no. 1 (2010): 1-20.

Nigam, Sumant, Bin Guan, and Alfredo Ruiz‐Barradas. “Key role of the Atlantic Multidecadal Oscillation in 20th century drought and wet periods over the Great Plains.” Geophysical Research Letters 38, no. 16 (2011).

Nye, Janet A., Matthew R. Baker, Richard Bell, Andrew Kenny, K. Halimeda Kilbourne, Kevin D. Friedland, Edward Martino, Megan M. Stachura, Kyle S. Van Houtan, and Robert Wood. “Ecosystem effects of the atlantic multidecadal oscillation.” Journal of Marine Systems 133 (2014): 103-116.

Kari Pohl

About Kari Pohl

I am a post-doctoral researcher at NOAA and the University of Maryland (Center for Environmental Science at Horn Point Laboratory). My work investigates how climate variability and extremes affect the diverse ecosystems in Chesapeake Bay. I received a Ph.D. in oceanography from the University of Rhode Island (2014) and received a B.S. in Environmental Science and a B.A. in Chemistry from Roger Williams University (2009). When I am not busy being a scientist, my hobbies include running, watching (and often yelling at) the Boston Bruins, and taking photos of my cat.
This entry was posted in Data, Precipitation, Streamflow and tagged , , . Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *