Note on the compatibility of ICOS, NEON, and TERN sampling designs, different camera setups for effective plant area index estimation with digital hemispherical photography

Variable	Camera setup
Variable	I	II	III
Camera model	Canon EOS 5D	Canon EOS 600D	Canon EOS 2000D
Fisheye Lens	Sigma 8 mm F3.5 EX DG	Sigma 4.5 mm F2.8 EX	Sigma 4.5 mm F2.8 EX
Camera height	1.3 m	1.3 m	1.3 m
Camera orientation	magnetic north	magnetic north	magnetic north
Initial ISO	640	400	100
Initial exposure compensation	0	0	−1
Initial shutter speed	1/32	1/60	Aperture priority
Image aquisition	single	single	auto-bracketing
F-stop	f/8	f/8	f/8
Light metering	Spot	Evaluative	Evaluative
Sensor size, pixels	12.8×10⁶	18.0×10⁶	24.1×10⁶
Raw image radius^*	1386	1450	1730

* Image radius on senor (pixels for 90 deg. zenith angle).

Comparison of the average gap fraction of the test stand with different camera setups (SI, SII, SIII) using ICOS, NEON and TERN sampling schemes.
*Joonis 2. ICOS, NEON ja TERN võrgustikel kolme erineva kaameraga* (SI, SII, SIII) *tehtud mõõtmis-test arvutatud keskmised võrastiku läbipaistvused sõltuvalt vaatesuuna seniitnurgast.*

Table 2.

Estimations of effective plant area index PAI_e for each sampling scheme taken with each setup and their corresponding standard errors (SE).

Tabel 2. Mõõtmisvõrgustike (ICOS, NEON, TERN) ja kaamerate (Setup I, II ja III) kaupa võrastiku läbipaistvusest arvutatud efektiivne taimkatteindeks PAI_e ja selle standardviga (SE).

Camera	Sampling network
	ICOS		NEON		TERN
	PAI_e	SE	PAI_e	SE	PAI_e	SE
Setup I	3.11	0.06	3.38	0.05	3.11	0.06
Setup II	2.99	0.06	3.22	0.05	2.94	0.05
Setup III	2.90	0.05	3.21	0.04	3.08	0.06

Effective plant area index PAI_e calculated from the data of three camera setups. RSE is the linear regression residual error. Overexposed measurements with Setup III are marked with red crossed circles and excluded from the regression.
*Joonis 3. Mõõtmispunktides saadud efektiivse taimkatteindeksi* PAI_e *võrdlus erinevate kaamerate* (Setup I, II ja III) *kaupa. Kaamera III ülevalgustatud mõõtmised on tähistatud punaste ringristidega ja regressioonimudelist välja jäetud. RSE on mudeli jääkviga.*

PAI_e values from three camera measurements for the ICOS scheme and the locations of the sample images shown in Figure 7.
*Joonis 4. Efektiivse taimkatteindeksi PAI_e väärtused mõõtmiste jadas ICOS võrgustiku punktides. Ristiga on näidatud joonisel (8) toodud näidispildi mõõtmispunkt.*

PAI_e values from three camera measurements for the NEON scheme.
*Joonis 5. Efektiivse taimkatteindeksi PAI_e väärtused mõõtmiste jadas NEON võrgustiku punktides.*

PAI_e values from three camera measurements for the TERN scheme.
*Joonis 6. Efektiivse taimkatteindeksi PAI_e väärtused mõõtmiste jadas TERN võrgustiku punktides.*

Zoomed-in central parts of the blue channel of the photos taken from ICOS point A8 with setup I (A), setup II (B), and setup III (C).
*Joonis 7. Suurendus kaameratega (I, II ja III) ICOS võrgustiku punktis A8 tehtud ülesvõtte keskosast. Kaamera III pildil on näha tugeva ülevalgustatuse ilminguid.*

Comparison of differences in average gap fractions in the ICOS scheme calculated from measurements with setup I and setup II and processed by the camera operators OpI or OpII. The influence of the image analyst is small compared to the camera factor.
Joonis 8. ICOS mõõtmisvõrgu andmetel saadud võrastiku keskmise läbipaistvuse erinevus kaamerate I ja II andmetel. Sama kaamera piltide töötlemisel on operaatori mõju (OpI, OpII) väike võrreldes kaamera faktoriga.

Table 3.

Linear relationships between effective plant area index between three camera setups by sampling schemes and using only correct (C) i.e. non-overexposed images of setup III or all (A). RSE is model residual error and N is the number of observations.

Tabel 3. Lineaarsed regressioonmudelid kaamerate ning mõõtmisskeemide kaupa. Kaamera III mudelid on antud kõikide (A) või ainult korrektsete mõõtmiste ehk ülevalgustamata piltide andmete (C) järgi. RSE on mudeli jääkviga ning N on vaatluste arv.

Camera setup		Sampling		Linear regression S_y = a + bS_x
S_x	S_y	Network	Subset	R²	a	b	RSE	N
I	II	ICOS	A	0.97	0.21	0.89	0.06	36
I	II	NEON	A	0.90	0.05	0.94	0.10	36
I	II	TERN	A	0.92	0.21	0.88	0.09	36
I	III	ICOS	A	0.28	1.55	0.45	0.27	36
I	III	ICOS	C	0.92	0.10	0.96	0.10	21
I	III	NEON	A	0.23	1.90	0.40	0.23	36
I	III	NEON	C	0.84	−0.52	1.15	0.12	23
I	III	TERN	A	0.71	0.55	0.81	0.18	36
I	III	TERN	C	0.84	0.10	0.97	0.13	24
II	III	ICOS	A	0.29	1.44	0.50	0.27	36
II	III	ICOS	C	0.93	0.02	1.02	0.09	21
II	III	NEON	A	0.17	2.15	0.35	0.24	36
II	III	NEON	C	0.85	−0.73	1.28	0.12	23
II	III	TERN	A	0.77	0.36	0.92	0.16	36
II	III	TERN	C	0.87	0.16	1.01	0.12	24

References

Authors

Metrics

Articles in this issue

DOI: https://doi.org/10.2478/fsmu-2023-0010 | Journal eISSN: 1736-8723 | Journal ISSN: 1406-9954

Journal RSS Feed

Language: English, Estonian

Page range: 21 - 36

Published on: Apr 13, 2024

Published by: Estonian University of Life Sciences

In partnership with: Paradigm Publishing Services

Keywords:

ICOS,

NEON,

TERN,

digital hemispherical photography,

Related subjects:

© 2024 Mihkel Kaha, Mait Lang, Shaohui Zhang, Jan Pisek, published by Estonian University of Life Sciences
This work is licensed under the Creative Commons Attribution 4.0 License.

Volume 79 (2023): Issue 1 (December 2023)

Note on the compatibility of ICOS, NEON, and TERN sampling designs, different camera setups for effective plant area index estimation with digital hemispherical photography

Figures & Tables

Figure 1.

Table 1.

Figure 2.

Table 2.

Figure 3.

Figure 4.

Figure 5.

Figure 6.

Figure 7.

Figure 8.

Table 3.

Paradigm

My account