Manual for LAI-Assessments

Manual for LAI-assessments

5.1.2. Planar photography (Ukonmaanaho, L. & Heikkinen, J.)

Forest canopy cover is an important ecological indicator, that can be used for example to characterize forest microclimate and light environment or torecognize habitants suited for several plant and animal species (e.g. Jennigs et al. 1999, Korhonen & Heikkinen 2009). Canopy cover is also an important ancillary variable in the estimation of LAI using empirical orphysically based vegetation reflectance models (Jasinski 1990, Spanner et al. 1990, Nilson and Peterson 1991, Kuusk and Nilson 2000). An estimate of the canopy cover can be obtained conveniently using digital camera and image analyses techniques.

Canopy cover is defined as the proportion of the forest floor covered by the vertical projection of the tree crowns, further canopy cover is distinguished from canopy closure, which is defined as the proportion of sky hemisphere obscured by vegetation when viewed from a single point (Jennings et al. 1999). The difference between these concepts is clear; if canopy is measured with instruments than have an angle of view (AOV) (i.e measure larger area than just a vertical point), like cameras or spherical densiometers, the results are estimates of canopy closure (Kuusipalo 1985, Cook et al. 1995, Korhonen et al. 2006). In other words canopy closure is just a percentage figure describing the fraction of non-visible sky within a certain angle, whereas canopy cover describes the fraction of ground area covered by crowns.

Digital camera photography with standard digital cameras provides a reliable method for estimating canopy cover in mature forests (Korhonen & Heikkinen 2009). Canopy closure can be estimated from the entire image and canopy cover from a more limited AOV. Planar photos, which have taken using AOV 30° to 60°, are not directly usable for LAI measurements, but could be used in regression models which predict LAI.

5.1.2.1. Location of measurements and sampling

5.1.2.1.1. Assessmentdesign

A grid net of at least 10x10m resolution covering an area of 0.25ha should be used. This is minimum size of a Level II plot, excluding the edges of the area, as defined in the ICP Forests manual. A denser grid net than 10x10m may be used.

-It is recommended to use those subplots where litterfall, throughfall and soil water is collected.

-If a measurement point is situated less than 1 m from an obstacle (e.g tree or bolder) the measuring point should bemoved so that it has a distance of least 1 m from anyobstacle. Each point must be marked permanently.

-Themeasurement height should be1.5 m.This avoids disturbances by lower shrubs or installations like litterfall or deposition samplers.

-The location of each measurement point has to be documented. This will be done by its relative coordinates. The origin (0/0) is the lower left measurement point (this is in general the point situated most south west; point D1 in sketch above), the adjustment of the system is north to south (A to D) and east to west (1 to 4). If another metric coordinate system is already established to the plot, its respective coordinates may be submitted instead.

-It is recommended to have more than 16 measurement points. Additional points could be for example above litterfall traps.

Coordinates are submitted to the data centre using form XX2009.LAC.

5.1.2.1.2. Assessmentequipment

1) Standard digital camera –preferably camera with Auto Exposure Lock-feature (AEL).

2) Image processing programme, e.g. MATLAB 7.1 numerical computing environment and programming language (MathWorks Inc. 2008) and image processing toolbox extension.Heikkinen and Korhonen have written a script, which automatically analyses canopy images (Heikkinen and Korhonen 2009 –

5.1.2.1.3. Data collection, transport and storage

Photographing:

Planar photographs can be taken using standard digital cameras. Digital cameras have considerably higher spatial resolution than traditional AOV instruments (densitometer, moosehorn).

-The images should take pointing the camera in a near-vertical, skyward direction at breast height (1.5 m) - use of a tripod to balance the camera into an exact vertical direction is optional. According to Korhonen & Heikkinen (2009) the plot level error caused by variation in camera angles is considered negligible.

-Place camera at least 1 m distance from the nearest tree bole to restrict the stem area visible in the image.Preferably there should be clear sky in the middlepoint of the photo.

-Use camera’s maximum AOV (i.e shortest possible focal length) – which in normal camera is between 30-50° .All images should be taken with the widest possible AOV without zooming.

• AOV can be calculated:

AOV= 2 tan-1(d/2f)

AOV = the angle of view for the camera

D= the width of the charge-coupled device (CCD) array in the selected direction

f = the focal length of the camera

-Images can be taken in varying weather conditions, with the exception of rain, as raindrops in the images disturb analysis. Sunny weather is not an obstacle as long as the sun does not appears directly in the images.

-Before taking photos, point the camera to the sky and press AEL-button (if the camera has that feature) to freeze exposure according to the light conditions of the sky. This decreases the risk of over-exposed photos, which make automatic image analysis more difficult.

-Use exposure time and other settings which have automatically set by camera

• Few words about camera settings if set manually. To freeze movements of the camera and of the branches and leaves, it is important to use short shutter speeds. This is especially true, if canopy photos are taken handholding the camera. Guideline for camera settings: Shutter speed<1/200s, ISO<800, Aperture>f/5.6, Whitebalance=cloudy/daylight. However, in most of the cases automatic settings give satisfactory results.

-Turn off the camera flash to increase contrast between sky and canopy pixels

-Use maximum resolution andhighest JPG-image quality

-Check camera settings and make sure photos are sharp and not severely under-or over-exposed.

-You can check camera properties later on from image properties.

-All images should be saved.

5.1.2.2. Measurements

Image processing

All image processing can be done usingMATLAB 7.1 numerical computing environment and programming language (MathWorks Inc. 2008) with an image processing toolbox extension or a similar type of programme. Heikkinen and Korhonen (Heikkinen and Korhonen 2009 – have written a custom script.Using this script it is possible to automatically analyze a large set of canopy images in onedirectory. Note, the script works only in MATLAB.

The script reads in the images in thedirectory, extracts the blue RGB component, and thresholds each image using the algorithm proposed by Nobis and Hunziker (2005). This algorithm is based on edge detection.In this implementation the image is thresholded for each threshold between 60 and 220 (the 8-bit grayscale value range inside of which the final thresholdis selected) and the mean brightness difference between the pixels on the crown and sky sides of the edges is calculated. The threshold that yields the largest mean brightness difference is selected for the final thresholding. The painting of the tree crowns visible in the binary images below was done with morphological dilation and erosion operations (Gonzales and Woods 2002, p. 523-527, or Wikipedia). See the example below.

1 2. 3. 4.

1. Original RGB image

2. Images were thresholded according to the method proposed by Nobis and Hunziker (2005). Basically, the idea is to find the value of the blue channel that gives the greatest contrast between the canopy and the sky.

3. Canopy closure (thresholded image)

4. Traditional canopy cover .

In some cases automatic thresholding can fail, e.g. due to reflection from the leaves on sunny days or to the presence of patchy clouds in the sky. In these cases the threshold value should be determined visually.

- the ratio of the crown pixels to all pixels in the image, which results in an estimate of the percentage of canopy closure

- If canopy cover is the variable of interest, the crowns visible in the binary images should still be processed to fill within-crown gaps.

- Manual threshold value – very subjective, different people choose a different threshold value!

- The average cover of images represents the canopy cover of the plot.

5.1.2.2.1. Variables measured and reporting units

Plot

Sampling point

Tree species

Date of observation

Time of observation

Brand and model of the camera

Exposure time - seconds

Resolution - pixels

Lens (focal length

Aperture (f/value)

Angle of view (=AOV), x°

Short description of light condition

5.1.2.1.2. Quality assurance and Quality control

Quality control is assured by providing photos for each measuremnt point. Every photo should be numbered and named XXPPPPNNNNDDDDDDTTTTTTSS.jpg

Country (X), plot number (P), measurement point (N), date of image (D – DayMonthYear e.g 150810), time of image (T – hhmmss), sequence number (S)

Table 5.1.2: Variables to be reported

Literature:

Cook, J.G., Stutzman, T.W., Bowers, C.W. Brenner, K.A. & Irwin, L.L. 1995. Spherical densiometers produce biased estimates of forest canopy cover. Wildlife Society bulletin 23(4):711-717.

Heikkinen, J. & Korhonen, L. 2009. MATLAB codes fro canopy image analyses. Available on line at

Jasinski, M. F. 1990. Sensitivity of the normalized difference vegetation index tosubpixel canopy cover, soil albedo, and pixel scale. Remote sensing environment 32: 169-187.

Jennigs, S.B., Brown, N.D. & Sheil, D. 1999. Asseessing forest canopies and understorey illumination: canopy closure, canopy cover and other measures. Forestry 72(1): 59-74.

Korhonen, L. & Heikkinen; J. 2009. Automated analysis of in situ canopy images for the estimation of forest canopy cover. Forest Science 55 (4): 323-334.

Korhonen, L., Korhonen, K.T., Rautiainen, M. & Stenberg P. 2006.Estimation of forest canopy cover: a comparison of filed measurement techniques. Silva Fennica 40(4): 577-588.

Kuusipalo, J. 1985. On the use of tree stand parameters in estimating light conditions below the canopy. Silva Fennica 19(2): 185-196.

Kuusk, A. & Nilson, T. 2000. A directional multispectral forest reflectance model. Remote sensing of environment 72: 244-252.

MathWorks Inc. 2008. Matlab- the language of technical computing. Avaialble on line at

Nilson, T. Peterson, U. 1991. A forest canopy reflecting model and a test case. Remote sensing of Environment 37:131-142.

Nobis M, & HunzikerU. 2005.Automatic thresholding for hemispherical canopy-photographs based on edge detection. Agricultural and Forest Meteorology 128, 243-250.

Spanner, M.A., Pierce, L.L., Peterson, D.L. & running, S.W. 1990. Remote sensing of tmeperate conifeorus leaf area index. The influence of canopy closure, understorey vegetation, and background reflectance. Internatioal journal of remote sensning 11(1):95-111.