Author: Leslie Goldman

© 2013 National Ecological Observatory Network, Inc. All rights reserved. The National Ecological Observatory Network is a project sponsored by the National Science Foundation and managed under cooperative agreement by NEON, Inc. This material is based upon work supported by the National Science Foundation under the following grants: EF-1029808, EF-1138160, EF-1150319 and DBI-0752017. Any opinions, findings, and conclusions or recommendations

expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.

Phenology is the study of the timing in plant and animal life

cycle events, which are mostly related to climate and

weather (Schwartz 2003)1. The study of plant phenology in

relation to global warming and increases in greenhouse

gases can give information on the fluctuations of plant

phenology due to climate change.

NEON will collect data for 30 years at 60 terrestrial sites in

different ecosystems, and this requires automatized and

standardized methods to gather reliable data. For this

project, an automated algorithm was designed to quantify

phenology in different ecosystems.

Introduction Algorithm Flowchart Modifications for NEON Determining Phenology Changes in the Long Term

The PhenoCam Network is a cooperative network that

archives and distributes imagery and derived data products

from digital Stardot cameras located at different research

sites around the world and North America. This network

archives image time series of vegetation that can be

analyzed with the PhenoCam Software Tools

(http://phenocam.sr.unh.edu/). NEON will use PhenoCam

network protocols and install Stardot cameras in 60

terrestrial sites.

Setup and Requirements

The PhenoCam Network

Excess of Green (EG) vs. DOY: Results of the Algorithm Applied to Different Ecosystems

Contact Information:

sp.cortes1026@uniandes.edu.co

www.neoninc.org

Figure 1. The algorithm quantifies shifts in green and red from

digital images . *

Read image

The algorithm followed the PhenoCam Network tools by:

• Applying the suggested dark threshold (15%)

• Calculating the 90th quantile and averaging consecutive

images.

• Plotting the 90th quantile vs. day of the year.

• Calculating the day of the year with the image name.

The algorithm was modified to be automated and standardized:

• Region of interest is fixed and a grid is used for having

detailed phenology information per subarea.

• Calculation options are fixed: dark threshold and number of

consecutive images to average

• 6 Phenology metrics were calculated: green chromatic

coordinate (G/[G+R+B]), red chromatic coordinate

(R/[G+R+B]), normalized difference ratio ([G-R]/[G+R]),

simple band ratio (G/R), excess of green (2G-R-B), and

excess of red (2R-G-B).

• Algorithm calculates the DOY when the maximum, minimum

and inflection points occur.

Phenology metrics plots will be calculated for every year

during the 30 years of NEON’s data gathering. By having a

standardized algorithm, plots will be comparable between

different years and sites, so changes in phenology can be

measured.

Figure 2. Depiction of camera setup. *

Internship Project: Interpreting Canopy Phenology using an Automatic Image Analysis

Algorithm to Process Phenocam Images

Stephanie Cortés (Senior, Universidad de los Andes) Co-Author: Kevin Sacca (Junior, Rochester Institute of Technology) Mentors: Michael SanClements, Sarah Elmendorf (NEON)

Begin End

Time series images from PhenoCam

Dataset

Crop vegetation area

Split cropped area in color channels: R,

G,and B

Perform dark threshold of 15%

Calculate 6 phenology

metrics per pixel

15° below horizontal line

Image Composition:

80% Canopy, 20% sky

Camera connected to the

PhenoCam network

Images taken every 30

minutes (down to dusk)

with a resolution of 1.3 MP

Create a grid on the cropped area

Calculate 90th quantile of

phenometrics per subarea on grid

Average the 90th quantile for each

3 consecutive images

Plot 90th quantile vs. DOY

Curve fitting to a 6 degree

polynomial

Calculate minimum and

maximum points in the graph

Calculate inflection points

in the graph

Graphs and readout per

subarea of the grid with curve fit

and minimum, maximum, and

inflection points

Calculate day of the year (DOY) with the

image name

Figure 3. Possible climate change impacts on plant phenology.

Climate change could cause an earlier spring and a later autumn.

Bartlett Forest: Mixed Forest Wind River Forest: Coniferous Forest

Burns Sagebrush: Dryland Harvard Forest: Temperate Forest

0 100 200 300 4000

50

100

150

DOY

EG

0 100 200 300 4000

10

20

30

40

50

DOY

EG

0 50 100 150 200 250 300 350 4000

20

40

60

80

100

DOY

EG

0 50 100 150 200 250 300 350 40020

40

60

80

100

120

DOY

EG

The algorithm is standardized and automated to different ecosystems:

The figure shows the result plot with the 6 degree polynomial fit for four of the ecosystems tested. The results show a clear shift in the EG phenology metric (EG=2G-B-R) during spring and

summer, even though all of these ecosystems have different vegetation. These clear shifts show that the algorithm does not need changes depending on the ecosystem and that it can be

standardized and automated for the 60 terrestrial NEON sites.

Bartlett Forest during summer. * Wind River Forest during summer. *

Harvard Forest during summer. * Dryland during summer. *

--- 2011

--- 2041

Excess of Green vs. DOY

a b c d e

a. Winter

b. Spring

c. Summer

d. Early fall

e. Late Fall

1 Schwartz MD (2003) Phenology: an integrative environmental science. Kluwer, Dordrecht. * Images taken from the PhenoCam Network (http://phenocam.sr.unh.edu/).

Last image?

Figure 4. Algorithm output for 4 different ecosystems.

True

False