Using Digital Photography and Spectral Reflectance Indices

1
Using Digital Photography and Spectral
Reflectance Indices to Quantify Foliar Pigment
Concentrations in Temperate Forest Tree
Species. Applying MVH/DEW image analysis
software to Authentic Research! Michael T.
Gagnon Complex Systems Research Center,
University of New Hampshire, Durham, NH, USA.
2
Making Forest Measurements with Digital
Photography
Results and Techniques Published in the 2004-05
Forest Watch Data Book
3
Other Hyperspectral Chlorophyll Indices
4
Digital Imagery Analysis with MVH Image
Red vs. Green Green
True Color
Red vs. Blue Red
Photo of a white pine 1st year (2007) needle
stack with gray card in the picture background
Blue vs. Green Green
5
Un Enhanced RGB Photo Processing
Reference
Sample area
6
Color Enhancements Red vs. Blue Red
(normalized) With whole leaf polygon sampling
7
Color Enhanced Images (Red vs. Green Green)
(Red vs. Blue Red) (Green vs. Blue
Green)

• Provides a comparison of two color intensities.
• Displayed color is the greater of the two
  intensities.
• Equal intensities are displayed in black.
• The difference of the color intensities is
  divided by the sum of the color intensities.
• Minimizes effects of shadows uneven lighting
  within the image.
• Example Red vs. Green Green
• RGB of an image (40, 80,60) respectively. The
  difference of Red vs. Green is 40 and the sum
  of the two intensities is 120. The normalized
  value is 40 /120 or .33 or 33. The displayed
  color for that pixel will be a dark green with
  RGB values of (0,33,0)

8
Comparing Spectral Indices, Camera Data
Chlorophyll Extractions Of 2007 Forest Watch
White Pine Foliar Samples
Extracted total chlorophyll (in-vitro) serves as
the anchor to which (in vivo) chlorophyll
estimation techniques can be tested for accuracy
and compared.
22A
22B
22D
Figure 22 Results of One-Way ANOVA, and
Students T-tests for mean differences in
chlorophyll concentration for seven Forest Watch
schools (22A), compared to mean differences for 4
common spectral indices (22B,22C,22D,22E) and a
digital camera approach (22E).
22C
22F
22E
9
Comparing Spectral Indices, Camera Data
Chlorophyll Extractions of 2007 Forest Watch
White Pine Foliar Samples
Hyperspectral Indices REIP Max 1st
derivative between 680-750nm Chl_NDI
(R750-R705)/(R750R705) RE3/RE2
(R734-736)/(R715-726) FD715/FD705
(AvgFD710-720/avgFD700-710) Camera Approach Red
vs. Blue Red (Red
Blue)/(Red-Blue)
10
Studying phenology and foliar senescence in
American beech, Red Maple, and Yellow Birch
Three different American beech (Fagus
grandifolia) foliar samples collected on
September 26th 2008 at the Bartlett Experimental
Forest, NH.
1
2
3
RGB true color
Red vs. Green
Red vs. Blue
Blue vs. Green
11
1
2
3
12
Hyperspectral indices as chlorophyll predictors
r2 0.83 RMSE 0.036 p lt0.0001
Figure 6
13
Digital imagery from affordable cameras as a
chlorophyll predictor
TRUE COLOR (RED)
ENHANCED COLOR (Red vs. Blue Red)
Digital imagery can produce results comparable to
or even better than some hyperspectral indices
derived from field spectrometers.
14
The Relationship between hyperspectral indices
and enhanced digital imagery is very strong!
Preliminary Results show strong evidence that
inexpensive techniques for estimation of
chlorophyll content using DEW software and
affordable digital cameras is possible,
Relationships observed between total extracted
chlorophyll content and camera RGB color data is
comparable to hyperspectral laboratory techniques
and the two are strongly related. For senescing
vegetation of red maple, American beech and
yellow birch, the (Red vs. Blue Red) approach
to image processing appears to be a very
effective estimator of pigment content and merits
further exploration.
r2 0.72 RMSE 4.39 p lt0.0001
Figure 12
Red vs. Blue( RED)
15
More to Come! Phenology study!
7/22
sugar maple, Acer saccharum
7/29
9/01
9/11
9/21
9/30
9/21
9/30
10/09
black oak, Quercus velutina
10/19