Block Island Tern and Weather Data Analysis

Created by Tyler Sym and Rachel Tham

Welcome!



The first offshore wind turbines in the USA have been in operation since early 2017, off the coast of Block Island, Rhode Island. The purpose of this app is to evaluate the activity patterns of seabirds around Block Island. We analyzed the activity of Common and Roseate Terns (the latter of which is federally endangered), relative to weather and time of day covariates. This information is important in assessments of potential collision risk.

Please be patient for each tab to load properly!


Photo by Tyler Sym


Photo by Tyler Sym

Our objective was to display a large amount of data collected about nanotagged terns and weather in the Block Island area and make it easily navigable for users. There are several tabs located at the top of this page that let you view the data in different ways. The first three 'Tern' tabs plot different variables over time. The data for these tabs are aggregated by day, meaning that the weather data are averaged across 24 hours and the tern detections are summed across 24 hours to create each single plot value. This is helpful for viewing trends in the data over a long period of time. The next tab is similar, but aggregates the data by hour. This is helpful for zooming in on very specific portions of the graph, where the data are more detailed. It also allows you to plot a new hourly dependent variable, tidal height. The ‘Covariate Plots’ tab plots different variables against each other, eliminating the scale of time. Plotting variables against each other helps us explore correlations between them. The data used in the aforementioned tabs was collected with a stationary detection tower located on the southeast tip of Block Island. Finally, the 'Field Work' tab shows abundance data that were collected in July 2017, after the construction of the wind farm was completed and the turbines became fully functional. The data in this tab were collected manually; the research team used binoculars to spot birds and determine their behavior and proximity to the boat, and also used a handheld yagi antenna to detect tagged birds around the boat.


Photo by Rachel Tham

The data on this tab and the following 4 tabs were collected with a stationary detection tower on the southeast tip of Block Island. These data are aggregated by day. To isolate a variable, double click on the variable name in the legend. To add or remove variables, click once on the variable name in the legend. To zoom in on a particular part of the graph, click and drag a box around it. To view a specific time range, click on the graph and drag to the right or left.


Summary Statistics for Variables (if best-fit-lines were drawn through the data):

          

Data Analysis Examples


The screenshot of the graph below was taken from this 'All Terns' tab. It shows visibility on the right y-axis and tern detections on the left. Most bird detections occurred during clear skies (e.g. 10 miles). Some bird detections occurred at lower visibilities, but it was rare to detect any birds on days when the average visibility was below 5 miles. This conclusion is also supported by a plot that is shown in the 'Data Analysis Examples' section of our 'Covariate Plots' tab.























These data are aggregated by day. To isolate a variable, double click on the variable name in the legend. To add or remove variables, click once on the variable name in the legend. To zoom in on a particular part of the graph, click and drag a box around it. To view a specific time range, click on the graph and drag to the right or left.


Summary Statistics for Variables (if best-fit-lines were drawn through the data):

          

Data Analysis Examples


From the above graph, we extracted the plot below to show the relationship between the daily number of Roseate Tern detections and the average wind speed. Fewer Roseate Terns (n=16) were detected than Common Terns (n=30) and a maximum of 10 Roseate Terns was detected on July 3rd, 2016. The Roseate Tern detections consisted of 79.3% females and 20.7% males – these males were detected between June 12th and June 25th, 2016. Only one Roseate Tern was detected after July – it was a female, detected on August 5th. From June 19th to August 8th, 2016, as the wind speed increased (e.g. greater than 10 miles per hour), the number of tern detections increased, especially from June 20th to August 5th.


























These data are aggregated by day. To isolate a variable, double click on the variable name in the legend. To add or remove variables, click once on the variable name in the legend. To zoom in on a particular part of the graph, click and drag a box around it. To view a specific time range, click on the graph and drag to the right or left.


Summary Statistics for Variables (if best-fit-lines were drawn through the data):

          

Data Analysis Examples


From the above graph, we extracted the plot below to show the relationship between the daily number of Common Tern detections and the average wind speed. More Common Terns (n=30) were detected than Roseate Terns (n=16) and a maximum of 18 Common Terns was detected on July 3rd, 2016. The Common Tern detections consisted of 37.7% females and 62.3% males – only two female Common terns were detected after August 7th. From June 19th to August 8th, 2016, as the wind speed increased (e.g. greater than 10 miles per hour), the number of tern detections increased, especially from June 20th to July 31st.


























These data are aggregated by hour. To isolate a variable, double click on the variable name in the legend. To add or remove variables, click once on the variable name in the legend. To zoom in on a particular part of the graph, click and drag a box around it. To view a specific time range, click on the graph and drag to the right or left.





Data Analysis Examples


Included below is a screenshot taken from this 'Hourly Data' tab. The graph is zoomed in between June 29th and July 4th (the time during which bird detections were highest) and the “All Terns” and 'Tidal Height' variables are plotted. The first thing to notice is that the bird detections appear to occur in specific aggregations each day. Most of the birds were detected in the early morning, starting at 5:00AM. It is interesting to note how the timing of these detection clusters correlates with the rising tide. This suggests that the terns feed in the Block Island area as the tide increases. This is consistent with evidence in the literature that the positive effect of low tides on food availability increases foraging activity in terns (Safina and Burger: 1988, 160). The rising waters increase the accessibility of forage fish to terns feeding over shoals.






















The next screenshot shows data between June 27th and July 13th. Plotted on the left axis is all tern detections and on the right axis is inches of precipitation. Notice the negative correlation between these two variables. During periods of rainfall, tern detections are rare. In fact, across the entire season, only once did a tern detection fall within an hour when rainfall was recorded. That one case occurred on the afternoon of July 14th, when 1 male tern was detected during an hour in which 0.02 inches of rainfall were recorded.






















Temperature


6080000606876
41000048
5580000556779
60100000607692
29.730.300029.730.1
417000412
03600000144288
06200002856
0400004

Summary statistics for generated best-fit-line. All birds included.

              
Summary statistics for chosen x-variable.

              




Please specify the bird sex and species that you would like to view:



Summary statistics for generated best-fit-line. Filtered for sex and species.

              
Summary statistics for filtered sex and species.

              


Data Analysis Examples


The plot below was created in this 'Covariates Plot' tab. The x-axis shows the wind speed measured in miles per hour and the y-axis shows bird detections. It is important to note that these data are aggregated by day, meaning that the wind speed values are daily averages and the bird detection values are daily summations. In this plot, we can see from the best-fit-line (which explains almost 11% of the variance) that bird detections are related to wind speeds in a roughly parabolic way. The p-value of this line is <0.001, which indicates that the relationship between the two covariates is significant. The most birds were detected on days when the average wind speed was between 9 and 13 miles per hour.





























The following plot once again shows the correlation between between decreased visibility and decreased tern detections. In 91 total hours when the visibility was less than 5 miles, 6 terns were detected. In 1,352 total hours when the visibility was greater than 5 miles, those same 6 terns were recorded, in addition to 40 others.















Click on a data point to view tern detection or sighting information. To pan the map, click and drag mouse. To zoom in, use the left buttons or double click.


These data were collected manually while performing two independent boat surveys on July 5th and July 6th, 2017. The surveys were conducted after the construction of the Block Island wind farm was completed, with the intention of assessing how the tubines are impacting bird behavior. Researchers used binoculars to spot birds from the boat and used a handheld yagi antenna to detect tagged birds around the boat. The sightings and detections are plotted on the map below.

Sources: Google Maps and Rhode Island Energy Gov: http://www.energy.ri.gov/Pictures/Figure%208.36%20RE%20Zone.JPG





Shiny App coded by Rachel Tham and Tyler Sym


We would like to thank Dr. Holly Goyert, Dr. Pamela Loring, Mr. Kevin Rogers, Professor Curt Griffin, and Professor Paul Sievert for their guidance.


Copyright 2017
Summer 2017 University of Massachusetts-Amherst Wind Energy Research Experience for Undergraduates (REU)
This study was funded in part by the U.S. Department of the Interior, Bureau of Ocean Energy Management through Interagency Agreement M13PG00012 with the U.S. Department of the Interior, Fish and Wildlife Service. The findings and conclusions in this article are those of the authors and do not necessarily represent the views of the U.S. Fish and Wildlife Service.