Andrew Richardson, 18 August 2017
A phenocam is a digital camera that is used to track vegetation phenology, in terms of seasonal changes in the greenness of the canopy. This is done by recording time-lapse images of a fixed scene, over the course of a year, and then using simple image analysis techniques to extract quantitative color information from each image. Canopy greenness indices then provide information about the amount of foliage present, and its color.
One advantage of this approach is its ability to serve as a bridge between direct observations and satellite data, and thus to facilitate scaling from organisms to landscapes.
The method is fully described in the paper by Sonnentag et al. (2012):
Sonnentag, O., K. Hufkens, C. Teshera-Sterne, A.M. Young, M. Friedl, B.H. Braswell, T. Milliman, J. O’Keefe, and A.D. Richardson. 2012. Digital repeat photography for phenological research in forest ecosystems. Agricultural and Forest Meteorology, 152: 159-177, doi: 10.1016/j.agrformet.2011.09.009
PhenoCam (phenocam.sr.unh.edu) is a cooperative network that archives and distributes imagery and derived data products from digital cameras deployed at research sites across North America and around the world. On our Gallery page, we display the most recent image from each site. From there you can navigate to the individual site pages, browse images, and view canopy greenness data extracted from the image time series. You can also download raw image timeseries and processed data products.
The project is led by PI Andrew Richardson ( email@example.com ) of Northern Arizona University. Tom Milliman ( firstname.lastname@example.org ) at the University of New Hampshire is the data manager. Bijan Seyednasrollah ( email@example.com ) at the University of Northern Arizona is the data science post-doc.Funding from the National Science Foundation’s Macrosystems Biology (award EF- 1065074 and EF-1702697) program has enabled the expansion of the network and improvements in our online data visualization and delivery tools. The science questions motivating the project are as follows:
We are actively seeking new PhenoCam collaborators. Membership in the network is free and completely open. Our basic philosophy is as follows: if you contribute camera imagery, we will archive it, process it, and make imagery and data products available to a global community through our project web page.If you want to contribute camera imagery:
The paper by Sonnentag et al. (2012) demonstrated that a wide range of different digital cameras could be used for phenological monitoring. You can spend less than $100 or more than $3000.
Our network has largely been built using the StarDot NetCam SC (1.3 IR and 5.0 IR models), which has a proven record of reliability in conditions from hot (Arizona) to humid (Maine) to cold (Alaska). This is the camera we recommend you use, and the only camera for which we can provide technical support and assistance.
StarDot cameras are easy to use and configure, and have excellent image quality. Plugged in to electric power and an Ethernet socket, the cameras function as stand-alone devices—they do not need to be connected to another computer. The camera runs its own Linux-based operating system, featuring a web server and FTP client, and can be easily customized and scripted. For example, we have written scripts that take advantage of the camera’s IR sensitivity, so that we can obtain sequential visible and visible+IR imagery, from which we can compute NDVI-style indices.
For pricing information, etc., we recommend you contact Anthony Watts at StarDot (www.stardot.com). The standard PhenoCam camera, a 5 megapixel StarDot NetCam SC with IR capability, is available to PhenoCam collaborators (bundled with power supply, cables, housing and mounting bracket) directly from StarDot for $950. Note that this is substantially below list price, and you should identify yourself as an institutional researcher, and request the ‘standard PhenoCam 5MP bundle’ when emailing Anthony. You can also purchase these individual components separately from retailers like B&H Photo or Amazon.com. For additional details, see our PhenoCam Install Instructions).
We are not in any way associated with StarDot and we do not receive any commission for referrals. We just think they make a very high quality product. But, please tell them we sent you, and Anthony will get you set up with a package just like we use ourselves.
There are a few reasons why we have never used cameras with PTZ capabilities for the PhenoCam network. The main reason is that in our experience, the highest-quality data come from cameras with a fixed field of view: even slight shifts in the field of view can greatly increase the noise in the phenological signal. Also, many of our cameras are deployed in remote field settings that receive only periodic visits from a technician. We therefore prefer to avoid the risk of a burned-out PTZ motor, which would compromise data integrity.
The PhenoCam Installation Tool (PIT)f is a set of scripts that will automatically configure your camera according to the standard PhenoCam protocol. Manual configuration of the camera is not recommended.
There are a few things you should do. First, please use the PhenoCam Installation Tool (PIT) to configure your camera. This will ensure that your camera settings conform to our network standards. Second, the picture from your camera should be dominated by the vegetation of interest, so as to minimize the effects of changes in clouds and lighting. Third, it is critical that your camera be securely attached to its mount. Even one or two small shifts in the camera field of view can make it very difficult to extract a consistent (and hence useful) time series of “vegetation greenness” for the whole year. Fourth, try to minimize network outages. A day or two of missing data, here or there, is not a problem—but a week of missing data, at a critical point in spring or fall, can make it very hard to determine the dates of key phenological transitions.
No. We have run a number of sites using solar electric power systems. For sizing your battery bank, know that the StarDot camera draws about 1/3 A at 12.6 V. Northern Arizona Wind and Sun (http://www.windsun.com/) have a great web page with lots of useful information. Fire Mountain Solar ( http://www.firemountainsolar.com) is also a excellent source for solar power equipment and questions.
The StarDot camera uses FTP to send images, on an automatic schedule, over the Internet to our PhenoCam server.
What if you don’t have Internet? There are a few options. We have successfully used commercial-grade 802.11 wireless networking equipment to establish communications with tower sites located up to a couple of miles from field stations with a wired network connection. Note that line of site communication is important, as trees and buildings in the way will greatly diminish signal strength.
Also, we have a number of sites where cell phone service is reliable enough to warrant the use of cell phone modems. We have used CradlePoint ( http://www.cradlepoint.com) modems and can vouch for their reliability.
A third solution is to store the images locally. This can be done, for example, by networking your StarDot with a Campbell CR1000 data logger. The data logger pulls images from the camera and stores them on a flash card. The flash card can be swapped out during periodic site visits and, on return to the lab, images manually uploaded via FTP to the PhenoCam server. Additional details can be found on our Tools page.
New cameras are tagged as ‘hidden’ and won’t appear on our Gallery page. But, even for hidden cameras, you can view the most recently uploaded image by following this link (replace ‘site_name’ with the camera name we agreed on in initial discussions):
Long cable runs can increase the probability lightning damage. In every case of lightning damage we have experienced, StarDot has been able to simply replace the network card in the camera; this repair costs about $75.
If your site experiences common or intense lightning storms, you can put a surge protector in line between the camera and your router. For maximum protection, install it at the camera end of your Ethernet cable.
We have been using the APC ProtectNet 100BT/10BT/TR Ethernet Protector, which is available on Amazon for about $25. For instructions on integrating a surge protector, see our Surge Protector Instructions.
A reference panel can be used to monitor day-to-day variation in lighting conditions, due to solar elevation, clouds, aerosols, and other atmospheric effects. A reference panel can also be used to track the long-term stability of the imaging sensor.
Do not use a white panel as your reference. True white is fully saturated in each color channel, and has an RGB triplet of (255,255,255). You will not be able to obtain any information from a saturated panel. The best option for a reference panel is a material with a grey matte finish that will not attract dirt or change color over time.
Spectralon is the best material for a reference panel because it is a diffuse reflector. This means that the apparent color of the panel won’t be affected by illumination and viewing geometry (e.g. BRDF effects).
However, Spectralon is expensive. This is why we haven’t used it at a lot of sites.
Labsphere sells Spectralon targets in a variety of designs, including different levels of grey. See their web page (http://www.labsphere.com/) for more information.
We have also used reference panels made from plastic squares spray-painted with matte grey primer. While not ideal, such a panel may be better than nothing.
We are still searching for the ideal reference panel design.
In recent years, we have moved towards near-real time processing of PhenoCam imagery on our project server. For each camera in the network, we have defined one or more regions of interest, typically targeting the dominant vegetation in the camera’s field of view. Each night we process the new imagery uploaded that day, and we extract color information for each region of interest. The data files for each region of interest are then updated, along with the online graphs displaying time series of canopy greenness. While these data are considered preliminary, all registered PhenoCam users can download data files and use them for their own research and teaching, subject to the PhenoCam fair use policy, found here. If these data sets meet your needs, then there is no need to analyze the imagery yourself.
Some users may want to process the imagery themselves. The image processing steps are described in our first paper, by Richardson et al. (2007). The basic steps are as follows:
Richardson, A.D., J.P. Jenkins, B.H. Braswell, D.Y. Hollinger, S.V. Ollinger, and M.-L. Smith. 2007. Use of digital webcam images to track spring green-up in a deciduous broadleaf forest. Oecologia, 152: 323-334.
In practice, there are a lot of different ways to accomplish these steps. We offer a variety of solutions on our Tools page).
The easiest method is to use our stand-alone image processing program, which features a graphical user interface (GUI). The PhenoCam GUI application is available as a pre-compiled MATLAB program (no license required). It is available for both Windows (XP, and Windows7) and Mac OSX platforms. To run the application you need to first install the MATLAB runtime compiler (bundled in the download). Included in the download are instructions for running the application.
Alternatively, you might use the phenopix R package, developed by our collaborators in Italy and documented in the paper by Filippa et al (2014):
Filippa, G., E. Cremonese, M. Migliavacca, M. Galvagno, M. Forkel, L. Wingate, E. Tomelleri, U. Morra di Cella, and A.D. Richardson. 2016. Phenopix: An R package for image-based vegetation phenology. Agricultural and Forest Meteorology, 220: 141-150. DOI: 10.1016/j.agrformet.2016.01.006
Finally, the code we use for image processing, including extraction of colour information, and generation of “all-image”, and “summary” time series data product files, is available at https://github.com/tmilliman/python-vegindex/, while an R package for time series processing, including interpolation and uncertainty characterization, as well as outlier detection and transition date extraction, is available at https://khufkens.github.io/phenocamr/.
We make all of our imagery freely available (to registered users) for download by the community on our data page. If you join the network and share your imagery with us, you are agreeing to this data policy.
We have made a fully processed, curated data set publicly available through the ORNL DAAC. This data set is documented in the following publication:
Richardson, A.D., K. Hufkens, T. Milliman, D.M. Aubrecht, M. Chen, J.M. Gray, M.R. Johnston, T.F. Keenan, S.T. Klosterman, M. Kosmala, E.K. Melaas, M.A. Friedl, and S. Frolking. 2018. Tracking vegetation phenology across diverse North American biomes using PhenoCam imagery. Scientific Data 5, Article number: 180028. doi:10.1038/sdata.2018.28
Provisional data sets, containing the most recently processed imagery, can also be dowloaded from the ROI pages for each site.
Our fair use policy, and recommended data citations, can be found here.
For an overview of sensor-based monitoring of phenology, see:
Richardson, A.D., S. Klosterman, and M. Toomey. 2013. Chapter 22: Near-surface sensor-derived phenology. In: M.D. Schwartz (Ed.). Phenology: An Integrative Environmental Science (2nd Edition). Springer, New York, pp. 413-430. doi: 10.1007/978-94-007-6925-0_22.
The following paper is the standard reference for PhenoCam, describing the camera configuration, image processing, and dataset structure:
Richardson, A.D., K. Hufkens, T. Milliman, D.M. Aubrecht, M. Chen, J.M. Gray, M.R. Johnston, T.F. Keenan, S.T. Klosterman, M. Kosmala, E.K. Melaas, M.A. Friedl, and S. Frolking. Tracking vegetation phenology across diverse North American biomes using PhenoCam imagery. Scientific Data, in review.
Click here here for a full list of PhenoCam-related publications. Please send an email to PI Andrew Richardson to request PDF copies of specific papers.