Skip to main content
ARS Home » Research » Publications at this Location » Publication #312208
Title: Automated canopy estimator (ACE): Enhancing crop modelling and decision making in agriculture

Author
item COY, A. - University Of The West Indies
item RANKINE, D. - University Of The West Indies
item TAYLOR, M. - University Of The West Indies
item Nielsen, David

Submitted to: Meeting Abstract
Publication Type: Abstract Only
Publication Acceptance Date: 11/2/2014
Publication Date: 12/1/2014
Citation: Coy, A.D., Rankine, D.R., Taylor, M.A., Nielsen, D.C. 2014. Automated canopy estimator (ACE): Enhancing crop modelling and decision making in agriculture. Meeting Abstract. Available: http://www.climate-services.org/sites/default/files/ICCS4/Automated%20canopy%20estimator,%20Andre%20Coy.pdf

Interpretive Summary:

Technical Abstract: The Caribbean agriculture sector is dominated by small holdings, which are overly reliant on rainfall and highly dependent on manual means of optimization. The sector is therefore very vulnerable to the vagaries of climate variability and change, with rainfall variations being of particular concern. The routine use of crop modelling is not highly practiced, despite its obvious benefits, but largely due to the lack of tools sufficiently customized to meet local needs. The FAO AquaCrop model is a yield-response-to-water model that uses a small number of parameters and is ideally suited to the Caribbean. AquaCrop can also simulate yields under varying conditions of water availability and climate change. The model, unlike many others however, uses canopy cover (CC) instead of leaf area index (LAI) to track canopy development. One limitation of using CC in the model is that there are very few automated techniques that provide accurate estimates of CC. Existing software give inaccurate results in a number of situations, including varying illumination conditions, where canopies are close to open or closed and when vegetation reaches senescence. The Automated Canopy Estimator (ACE) developed in 2014, overcomes these challenges through a novel technique that directly measures canopy cover from inexpensive digital photographs. The tool outperforms seven other leading image segmentation approaches and advantageously operates in different lighting conditions and at different stages of crop development. ACE also has the capability to differentiate between green and senescent vegetation, for which there are no other simple approaches. The tool has yielded impressive results when used with open or near closed canopies for a number of crops including, corn, wheat, flax, and most recently, sweet potato. This paper will introduce ACE and show that the improvement in canopy estimation it provides allows for markedly improved crop model simulation with the AquaCrop Model, as demonstrated with the parameterization of rain-fed and irrigated sweet potato. The paper will also discuss additional applications of ACE, including but not limited to land use monitoring; detection of forest change cover and cloud cover estimation.