A few weeks ago, I wrote a post briefly touching on our study area. More specifically, I discussed how and why we divided Chesapeake Bay into a North and South region.
Now that I have presented this work at a few different places, I thought a follow-up was needed. More specifically, what data is represented in Figure 1?
What do those symbols mean?
The size of the Picture in Fig. 1
Yup, you read that correctly! Our first data set is a globally gridded data product over a 3.75° by 2.5° gridpoint that represents the entire near-shore Chesapeake Bay (longitude: -75 to -77. 5°, latitude 40 to 37.5°).
This data set is called the HadEX2 and incorporates hundreds to thousands of meteorological observations throughout that “square”. This gives us great data coverage of our study area, but is an area-weighted average, so the signal is likely dampened. For example, this data set includes weather data from up in the mountains. Anyone who has ever hiked a tall mountain can tell you it is chillier and windier at the top!
The take home: The HadEX2 offers a lot of weather time series data that has been quality controlled for us, but since it represents a really large area, the magnitude of trends and patterns are likely smaller; any trends and directionality will be the same, but the slope will likely be gentler.
This is confusing, so let me give an example using Frost Days, how many days annually had the daily minimum temperature go below freezing. All time series showed a decrease in Frost Days, but the magnitude was different between the HadEX2 (-0.35 days per decade) and the North (-3.3 days per decade) and South (-2.8 days per decade). The cooler air at the higher elevations likely lowered the intensity of this trend, but we still see the same pattern (a decrease!).
Yellow and Blue Diamonds
Each diamond symbol, both yellow and blue, represent an individual weather station. These weather stations are from the Global Historical Climatology Network as part of their daily weather monitoring.
We are using 18 of these individual stations (Table 1) which met our quality standards. We only used GHCN-Daily stations that 1) where still actively collecting data as of December 2014, 2) has been collecting data since at least 1960, 3) had few multi-year data hiatuses, and 4) was situated within 115 km of Chesapeake Bay within Maryland and Virginia. These 18 are long and continuous weather records of the area around Chesapeake Bay.
We took care into demonstrating that each of these weather stations were correlated to each other, justifying our aggregation into a North and South region. You can see in Table 2 that the daily temperature of each weather station in the North Chesapeake (yellow diamonds) had a statistically significant linear fit with the weather station at Jug Bay of at least R2=0.82.
Precipitation is a bit more difficult to understand since it can be a very local process. Have you ever experienced a torrential rainstorm, but your friend across town is outside playing in a light sprinkle? So, we correlated the individual stations to the HadEX2 (Figure 3). And now we can see a great fit! Think about it this way: our weather stations captured the rain, but station may have a slightly different amount of rain. To understand regional trends, we aggregated the weather stations together to undercover trends….and it turns out the HadEX2 and grouped stations are both pretty similar!
These represent the seven Chesapeake Bay National Estuarine Research Reserves of Maryland and Virginia. These sites have water quality and meteorological data collected at 15-minute frequency. While the GHCN-Daily and HadEX2 give us the regional patterns, trends, and variability, the NERRS data allows us to zoom-into weather and water parameters at a fine detail. Remember this post? It was made possible only by having 15 minute data!
In this project, we combine a Chesapeake-Bay wide data set (HadEX2), a near-shore data aggregate (GHCN-Daily), and high frequency place-based data (CBNERRS) to understand how extreme climate change and variability has manifested in the shallow water environments we care so much about! Two data sets is always between that one…so using more than 20 must be even better!
Next week, we will be looking at relationships in our indices with large scale climatic teleconnections like El Nino!