KC and the sunshine (band) – permanently installed cameras to measure almond canopy size and better estimate irrigation requirements
Mark Skewes1,2,3, Dane Thomas1,3, Nigel Fleming1,2,3, Kavitha Shanmugam1, Vinod Phogat1,2,3, Darren Graetz1, Paul Petrie1,2,3 and Tim Pitt1,2,3
1South Australian Research and Development Institute (SARDI), Adelaide, SA, Australia
2College of Science and Engineering, Flinders University, Bedford Park, SA, Australia
3School of Agriculture, Food and Wine, University of Ade
INTRODUCTION
Inefficient irrigation scheduling may result in either under or over irrigation, in turn resulting in poor tree growth, lower yields and waste of water and other resources such as fertilisers. Estimation of irrigation requirements using evapotranspiration and crop coefficients is widely adopted in the almond industry.
However, the industry is undergoing significant change, particularly through the introduction of new varieties and rootstocks, and increased tree densities, resulting in altered tree growth habits and uncertainty around tree irrigation requirements.
In addition, there is an opportunity to link tree canopy monitoring with irrigation advisory software to assist almond growers to manage irrigation more effectively.
This project demonstrated the ability of under-canopy camera systems to estimate seasonal variation in canopy size and crop coefficient in a range of almond production systems (including different varieties, growth habits/management systems and tree densities).
Figure 1. Example RGB imagery from under-canopy timelapse camera (a) and the binary (blue channel) output (b) that informed daily calculations of leaf area index.
METHODOLOGY
Almond trees at the ACE orchard comprising multiple ages, densities, genotypes and training systems were selected for assessment of canopy size using permanently installed, high resolution, timelapse cameras (GERBER Trail Camera, 4G MMS).
Cameras were installed in a security box approximately 1m south of the trunk and in-line with the planted row, with the lens pointed vertically up through the canopy toward the sky. Cameras were programmed to collect one image per day, late in the afternoon to avoid the occurrence of sun-flair.
Under-canopy images were processed by splitting the original RGB images into red, green and blue channels. The blue channel was converted into a binary image (sky vs canopy) using the thresholding algorithm developed by Otsu (1979). The image was then divided into 25 sections (as a 5 by 5 grid, Figure 1) and the proportion of canopy in each grid was used to calculate the leaf area index of the image using the method described by Macfarlane et al. (2007).
Leaf area index was converted to density coefficient (Kd), from which basal crop coefficient (Kcb) was derived, using the methodology of Allen and Pereira (2009).
Figure 2. Timelapse K cb of Nonpareil, Vela and Shasta trees at low tree density (308 tree/ha) from November 2022 to July 2023, demonstrating differences due to variety.
RESULTS
Shasta and Vela are two new almond varieties with strongly contrasting tree habits. Shasta grows upright trees which open with the weight of crop due to tip bearing, whilst Vela trees are quite spreading. Trees of Nonpareil, the current industry standard variety, are relatively upright and compact. Figure 2 displays Kcb calculated from under-canopy camera images for Nonpareil, Shasta and Vela trees planted at 6.5 x 5m (308 trees/ha). The comparison indicates that Kcb measured using timelapse cameras was highest in Nonpareil and Shasta, while Vela gave lower values. The decline in Kcb at the end of the growing season also occurred earlier in Vela and later in Shasta, when compared to Nonpareil, despite there being no difference in management, and Shasta being the earliest variety harvested and Vela the latest.
Scan the QR Code to view a video related to this work.
CONCLUSION
This project demonstrated that an under-canopy camera system is able to monitor canopy growth in orchards to support improved irrigation scheduling. Under-canopy cameras were positioned approximately 1m from the trunk, suggesting that the commercial use of such sensors within an orchard environment would not be disrupted by orchard machinery and standard management practices. Permanently installed undercanopy timelapse imagery systems demonstrated their ability to track changes in canopy size and water requirement, particularly in the latter half of the season.
REFERENCES
ALLEN, R. G. & PEREIRA, L. S. 2009. Estimating crop coefficients from fraction of ground cover and height. Irrigation Science, 28, 17-34.
MACFARLANE, C., HOFFMAN, M., EAMUS, D., KERP, N., HIGGINSON, S., MCMURTRIE, R. & ADAMS, M. 2007. Estimation of leaf area index in eucalypt forest using digital photography. Agricultural and Forest Meteorology, 143, 176-188.
OTSU, N. 1979. A Threshold Selection Method from Gray-Level Histograms. IEEE Transactions on Systems, Man, and Cybernetics, 9, 62-66.
