Season Spotter Major News

Sorry that it’s been so quiet on the blog this summer. We’ve been busy analyzing all the data you have helped produce and we’ve written and published the first Season Spotter paper!

paper-snipIn the paper, which was published in the science journal Remote Sensing, we describe the project and the data produced, and we show that the data is of good quality and useful. You can read the abstract (scientific summary), the whole paper, or a plain language summary I wrote about the paper. And you can find yourself on our list of contributors.

We now know which type data from Season Spotter is good quality and we have created the post-processing software to turn your classifications into that data. We have also learned what doesn’t work so well (e.g. identifying grass seedheads) and what has been less than ideal (e.g. not taking advantage of the fact the images are in sequences).

So we’ve got two major next steps:

  1. We’re going to revamp the classification interface. When we launched in July 2015, the Zooniverse Project Builder was still pretty simple. Now it’s more sophisticated and I think we can make many of the classification tasks much more efficient by asking questions about multiple images at a time, instead of just one or two at a time. To create the new classification interface, I’d love to have your feedback. I’ve put together a project called Season Spotter Sandbox, where we can try out different ways of doing classifications. Tell me what you like and what you don’t like in its attached Talk forum.
  2. We’re going to identify the science question(s) we’d like to address next with Season Spotter data. I’m personally leaning towards tree-circling classifications, so we can figure out how to connect different types of phenology data at different scales. In other words, we have data from ground observers, from PhenoCams, and from satellites, but it’s not always clear how to use them together. If we could calculate individual tree phenology from the PhenoCam images, we could connect the first two. But there are other possibilities. If you have questions you think we should ask with the Season Spotter data, please leave a comment below.

Thank you again for all your classifications. I’m looking forward to the next season of Season Spotter.

Posted in Project report, Research | Tagged , , | Leave a comment

It’s finally springtime in the Northeast

Last time I posted, I commented on the fact that winter temperatures in New Hampshire were almost 7°C warmer in 2016 than 2015, and that “spring 2016 is just around the corner.”

It turns out that spring has been a long time coming. While it isn’t a particularly late year overall, it does seem that it is a particularly slow year. Take a look at the “greenness” data we derive from imagery from our PhenoCam overlooking the Boston Common. What you will notice is the much more gradual rise in greenness compared to previous years. In 2016, greenness began to trend upward in mid-March, and isn’t going to peak until late May.  By comparison, in 2015, greeness began to trend upwards beginning in late April, and it reached its peak by mid-May.

The pictures below compare April 19, 2015 (left) with March 19, 2016 (right) – they look pretty similar, despite the 2016 picture being a full month earlier.


And these pictures compare May 17, 2015 (left) with May 17, 2016 (right) – you can see that this year, we’re lagging behind last year on the same date by just a little.


Temperature is the main factor driving these differences; 2016 got off to a quick start in early April because March was much warmer  in 2016 (monthly mean temperature of 5.8°C) than 2015 (monthly mean temperature of 1.0°C). But this year we had a lot of cool weather in April, which continued on into May, and that’s really slowed down the rate of development: for example, the mean temperature for the first two weeks of May was 12°C this year, compared to 16°C last year.

That said, the last few days have finally started to feel like spring!

Interested in reading more? Check out the article I wrote last year for Arnoldia, the magazine of Harvard’s Arnold Arboretum, about using the Boston Common PhenoCam to track the phenology of urban trees.

Posted in Camera images, Science | Tagged , | 1 Comment

Friday favorites: Grassland at dawn


Dawn at the Sevilleta Long Term Ecological Research Site in New Mexico highlights a patchwork of grasses and herbaceous plants.

Posted in Camera images | Tagged , , , | Leave a comment

Season Spotter Jornada

Sorry for the lack of posts last week. I have been intensely working on all the data you all have generated in Season Spotter and putting together a scientific paper. Thanks so much for helping out to get those fall images classified in time for the paper. I’ve been analyzing them yesterday and today and the data look awesome. More about autumn in a future post.

Meanwhile, we have another small task that would be great to complete soon. At Jornada (pronounced HOR-na-da) Experimental Range in New Mexico, there’s a dry grassland dotted with mesquite shrubs. A collaborator there has on-the-ground field data of when various grasses and the mesquite flower. We’d love to compare Season Spotter data with the field data to get a sense of how accurate the Season Spotter data is. Can we see flowers only when there are many of them? Or do we capture the whole flowering period?

We have Season Spotter Jornada images for most of 2013, 2014, and 2015, for a total of just under 1,000 images. This is peanuts compared to the spring and fall images, which had about 10,000 image pairs each. So hopefully we can zip through the Jornada images quickly. If you have a moment, click on over and classify a few. Thanks!


Posted in Field work, Project report, Research | Tagged , , , , | Leave a comment

Friday favorites: Cool spring

harvardbarn2This week has been abysmally cool and wet in Massachusetts. Buds are eager to burst into leaves as soon as we have a little warmth at Harvard Forest. Here the imminent green of new spring leaves vibrates against the orange carpet of last-year’s leaves.

Posted in Camera images | Tagged , , | Leave a comment

Results from tree outlining

In Season Spotter Image Marking, one of the tasks we ask you to do is to outline individual trees and tell us if they are broadleaf trees or needle-leaf ones.


It’s easier to see different trees at different times of year. For example, evergreens are easier to see in winter when nearby deciduous trees have lost their leaves. Likewise, deciduous trees whose leaves change color are easier to see in autumn. And different lighting conditions change the highlights and shadows around different trees, making some easier to see in one image and others easier to see in other images. So we’ve taken a handful of images from each site in each year, and we show these pictures to you.

Then we combine your classifications across images. Remember that we only ask you to outline three trees per image. But not everyone outlines the same three trees. In the end we get a good sampling of all the most easily delineated trees in the image.

Here is the view from the Arbutus Lake PhenoCam in Huntington Forest, New York:


And here’s what it looks like if we put everyone’s markings for this site on top of one another:


Here, white shapes are broadleaf tree markings and yellow shapes are needle-leaf ones. As humans, it’s pretty easy to pick out the major trees. We also see a lot of stray shapes. For example, sometimes people outline entire vegetated regions instead of single trees. This is probably because another question in Season Spotter Image Marking asks volunteers to do just that, and it gets a bit confusing.

What we really want is a single shape for each major tree. So I implemented a clustering algorithm that takes all these shapes and finds those that are most similar to one another. It groups all these shapes into clusters, and hopefully each cluster represents one tree. What’s tricky is that we have to define a threshold of just how similar the shapes must be to be in the same cluster. If we require that the shapes be super similar, then shapes around the same tree that are just slightly different don’t get grouped together. But if we are too lax about our clustering rules, then multiple nearby trees — especially small ones — get grouped together in a single cluster.

Once we have our clusters, we need to take all the shapes in that cluster and combine them into a single shape. I did this by taking a few of the smallest shapes in each cluster and taking their geometric union. This gives a conservative estimate of the extent of the tree crown. I did it this way because the next step is to run some automated algorithms on the resulting regions, and I wanted to make sure I wasn’t including any pixels that were outside the focal tree.

Here’s what the clusters and resulting “consensus shapes” look like:


The next thing to do is to run our automated algorithms on each outlined tree and compare that with the results when we run it on the whole landscape. Traditional phenology measures have been on-the-ground and done on a tree-by-tree or plant-by-plant basis. And satellite phenology measures integrate across broad areas. This analysis will help link those two types of phenology measurements, so that we can scale-up from the ground observations and understand the satellite observations more biologically.

Posted in Camera images, Project report, Research | Tagged , , , , | Leave a comment

What is Temporal Mismatching?

There are many reasons data that examine the timing of plant phenophases are important. One of these reasons is to provide data to support conservation efforts. Specifically, there are many species that depend on the timing of when plants flower, leaf, and produce fruit for their reproduction and survival.  As temperatures warm due to climate change, shifts in the timing of when plants flower and when pollinators emerge can result in what we call temporal mismatching.

Why is this important?  If a pollinator emerges before its host plant blooms or if a plant blooms before its pollinator emerges, this leaves the pollinator without food or a plant without a way to reproduce.  The interactions between these species are complex, so new research continues to inform how climate change might result in temporal mismatching between plants and pollinators but also other species.  The data you provide through Season Spotter supports this type of research.


Town of Washington Nature Trail

Virginia Master Naturalist volunteers install a native plant garden in the Town of Washington., Virginia. Photo: Marie Majarov

There are additional ways you can support the conservation of pollinators in your community by planting a native plant garden.  Native plants provide nectar, pollen, and seeds to other native species that depend on them. Some pollinators, like the monarch, are host dependent so the monarchs’ larvae need milkweed to survive.  Declines in milkweed populations in recent years have led to declines in the monarch population.  So, you can support conservation efforts of the monarch by planting milkweed in your own garden or planting other native plants that support other native pollinators.

Posted in Uncategorized | Leave a comment