Smartphone based image analysis to assess vine water stress

Smartphones contain a variety of sensors that have the potential to monitor the surrounding environment and provide an aid to decision making across a range of industries, from medicine through to agriculture. Smartphones have several advantages over specialist monitoring systems, including ubiquity, price, user familiarity and the ease of implementing updates. They also contain sufficient computing power, so the analysis and support software can be contained within the phone.

A range of methods has been developed for the assessment of vine water status – however none currently meets the affordability, portability and ease of use requirements for wide scale adoption. A wide range of sensors could potentially be interfaced with smartphones to assess vine moisture status. The aim of this project work was to evaluate a range of smartphone based tools for measuring vine water status, leading to the development of the most promising tool into a smartphone application that can be easily used by vineyard managers.

Systems that were evaluated included:

• An infrared camera that is integrated into or connected directly to the smartphone and uses established techniques for the analysis of thermal imagery to assess water status;
• A portable Near Infrared spectrophotometer that interfaces with the phone and measures reflectance across relevant wavelengths for the calculation of water status indices;
• A 3D camera that is integrated into or connected to the phone via Wi-Fi and can use image analysis to assess the shape or orientation of the leaves;
• A microscope attached to the smartphone camera or as a separate portable unit that can be used to measure stomatal number and aperture and then calculate stomatal conductance.

A trial site with a range of irrigation deficit treatments applied to Chardonnay and Cabernet Sauvignon grapevines was established in the Riverland of South Australia. Water status measurements from the smartphone based sensors described above were benchmarked against conventional methods including mid-day stem water potential and stomatal conductance.

The thermal infrared camera system, measuring the Crop Water Stress Index (CWSI), was selected as the most accurate and robust option for development into an app. The app was tested during the 2017 growing season to demonstrate its accuracy and confirm that the most appropriate indices were being used. User acceptance testing was also completed by viticulturists to assess its utility with further improvements being made during the season based on feedback from this testing. Positive feedback on the utility of the app was received from the beta testing group. Over the 33 days of formal assessment the CWSI, as calculated by the app, had a strong relationship with the reference methods of mid-day stem water potential (R2 = 0.62) and stomatal conductance (R2 = 0.74).

There are a range of scenarios where CWSI could be used to inform irrigation scheduling, which could complement or replace existing soil moisture monitoring systems. Regular assessment of the CWSI in a vineyard will help develop an understanding of what values to expect from different blocks or varieties. If the target is to maximise vineyard yield then water stress needs to be avoided, while not applying excessive water. Checking the CWSI immediately prior to applying irrigation would confirm that no stress has occurred. Preliminary estimates suggest that a CWSI of less than 0.7 recorded from a shaded canopy on a hot day, or 0.5 on a cooler day would indicate vines are well irrigated. If the target is to optimise quality and minimise water use as part of a regulated deficit irrigation strategy, then the CWSI could also be used to inform irrigation decisions. If the vines are being maintained at a moderate water deficit (for example during the post flowering period) then irrigation could be withheld if the CWSI is below approximately 0.8 assuming very hot weather is not forecast. Tracking the CWSI over time may give a viticulturist more confidence to extend the period between irrigation applications. Soil moisture monitoring systems are normally point based at a limited number of sites across a vineyard; it is uncommon for all the blocks within a vineyard to be covered. The thermal camera is very portable and can be used to compare different parts of a block and across blocks. This system provides an easy opportunity to check sections of a block that may not be receiving enough water, and benchmark these against the section adjacent to the soil moisture probe.

If the app is developed and maintained so that it can be used by the wine industry, then the savings in water applications are likely to occur as growers will be able to easily assess their vineyards irrigation needs. Fruit quality will also potentially be improved through the better management of regulated water deficit, and by applying strategic irrigations to maintain canopy health and avoid defoliation.

This project was completed as a collaboration between the South Australian Research and Development Institute (SARDI) and the University of New South Wales (UNSW). The measurement of vine water status was completed by SARDI at the Loxton Research Centre, and the image analysis and app development were completed by UNSW in Sydney.

Objectives

To evaluate a range of potential sensor/analysis options to assess vine moisture stress and develop an app to allow the most promising system to be easily used by growers.

This system will provide vineyard managers with better information on the water status of their vineyards; which in turn will allow better irrigation decisions to be made to improve the use of irrigation water and produce higher quality fruit.

Approach

The field performance of four smart phone based methods of estimating vine water status will be compared against measurements of midday stem water potential, stomatal conductance and canopy temperature:

• Thermal camera images will be used to calculate a Crop Water Stress Index. A portable and rapidly deployable system to provide wet and dry reference temperatures will also be developed.
• A SCiO spectrometer will be used to capture reflectance at specific wavelengths, to calculate the Water Band Index.
• Image analysis techniques will be developed to automatically locate the stomata in the magnified image and measure how open they are. 
• A camera-based leaf assessment using stereo vision to build a depth map which represents the scene in three dimensions. 

Sector benefits

This project has the potential to provide inexpensive, easily accessible tools to assist grape growers to manage water stress for positive outcomes, increasing both on farm returns and winery returns as the result of improved and more consistent grape and wine quality.