Posted by: norstarnewengland | May 9, 2016

SciFri Sample Everything – Picture Post

In addition to sampling water for two watershed associations, I also participate in a project called Picture Post, which samples data in several ways.

ExampFixtture

One of the picture post fixtures on Blue HIll

PicturePost is a citizen scientist project where people go to specific locations where a fixture specifically for this project is used to create panoramic image sets of the area.  These images are used primarily to view and track changes in foliage over time.

For this challenge, I used pictures to monitor the foliage on Great Blue Hill just south of Boston from the end of March to early May.  I used pictures in the south direction, which looks over a variety of deciduous trees, including norway maple, ash, hickory, and fruit trees.

How sampling is used – observing all the leaves on every tree in the region every moment in time is just impossible.  Taking pictures periodically reduces the volume of data, but may still be enough to note changes.  This challenge investigation whether weekly monitoring results in changes that can be noted not only during the challenge time period, but also compared to a previous year.

The first results are the pictures used, in chronological order:

See the link for Picture Post to see the full-sized images.

On the second week, April 3, it snowed.  On the other weeks, buds are developing; however, the changes are not easy to detect, visually, from a distance.  But how can these visual changes be reported more objectively?  One way is to take a clip of a portion of each picture where there are trees – another way the data is sampled.  The following is one example:

PBHT20160430S10195SSP001KTCCA2

Clipping an area helps limit the changes to only those related to vegetation – you don’t want to capture changes in the sky, for instance.

Next, a graph is generated that shows the brightness of each red, green and blue color in this clipped area – called a histogram (every pixel has a sensor for red, green, and blue, which, when combined, represent any color in an image).  From the histogram,  the mean value for each color is determined and a graph is created that shows the change in this average over time.  The idea is that in the spring more green will appear in the image.  Thus, the graph should show an increase in the curve.  The following is the graph created.

GPH_SciFriSampEv_BHT5SS_RGBIndex_Spring2016

The results are inconclusive.  The snow storm on April 3 caused the image to be darker around the trees, which resulted in the red, green, and blue values to be less different from one another.  Also, in the last couple weeks, the weather turned colder and the sky was mostly cloudy.  This slowed the foliage progression rate. The following image shows the progress in 2015.  This showed that, around 5/19, the green value finally topped the red value.

GPH_SciFriSampEv_BHT5SS_RGBIndex_Spring2015

Thus, if this challenge were just one or two weeks longer, I would expect that the green values to be greater than the red, signaling that the foliage is out!

Taking picture from Blue Hill Observatory Tower, and me with Boston in the background.

Links:

Blue Hill Observatory (Sponsor of three picture post locations on Blue Hill):  http://www.bluehill.org

Picture Post Project:  http://picturepost.unh.edu

Picture Post Project, Blue Hill Tower:  http://picturepost.unh.edu/analyze/post.jsp?postId=179

 

 

 

 

Advertisements

Categories

%d bloggers like this: