Future agricultural and food production systems must make better use of limited resources to ensure that farmers can economically produce more high-quality food for the growing world population while minimizing the environment's impact. Grapevines have lower mineral nutrient requirements than many other crops; still, vine nutrient status should be monitored to help prevent deficiencies that could reduce vine capacity, yield, and fruit quality. Monitoring nutrient status can also help growers prevent unnecessary fertilizer applications, which needlessly increase production costs and cause unintended nutrient imbalances or environmental contamination.
A nitrogen (N) management plan helps growers make the best use of N fertilizers, and an important part of such plans involves monitoring vine N status. Tissue sampling is one of the best ways to monitor nitrogen in grapevine. However, it is an imperfect monitoring method because it is based on a destructive, costly, and time-consuming laboratory assessment. This is a laborious practice that, and most of the time, provides an average value for the entire vineyard. In contrast, remote sensing can provide more detailed data about entire vineyards, potentially for every vine.
The recent introduction of sophisticated, more affordable, and user-friendly drone-based aerial imaging equipment has made it possible for growers to consider collecting frequent large-scale data for monitoring vineyards. However, a well-established aerial image processing pipeline, a comprehensive calibration procedure, and accurate data interpretation models are lacking. Until those exist, it will be difficult for growers to capitalize on aerial imaging applications fully. This project aims to develop a data-driven decision support tool that will enable growers to monitor vine nutrient status accurately. This tool will enable growers to use drone imagery to obtain site-specific insight about nutrient status throughout the entire vineyard. Over the last few seasons, we have been pairing aerial imaging and proximal sensing with laboratory analysis of tissue samples to develop calibrated models that should enable us to use remote sensing to estimate vine N status. The tools that will be developed in this study will be accessible via a web application that will automate most processing and data analytics steps and provide per-vine nutrient status at different stages of the growing season.
In 2018, we created an experimental plot design with three different N application rates of (0, 19.3, and 48.3 g N/vine) applied in either two (Slugs) or 10 (Spoonfeed) separate applications throughout the season and with 5 repetitions. In total, we had 30 plots; each included 5 vines.
Tissue samples were collected from vines in the experimental plot and assessed in the laboratory for true nitrogen status. The nitrogen content percentage were used as labels to calibrate the multispectral data interpretation model.
Study in 2020
Our objective in 2020 is to understand temporal variabilities in aerial data and how they correlate to actual N content at three important stages of the growing season: bloom, version, and harvest. We are developing and calibrating a robust N content estimation model that can be practically used by vineyard managers.
Study in 2019
Results published in:
Moghimi, A., Pourreza, A., Zuniga-Ramirez, G., Williams, L. E., & Fidelibus, M. W. (2020). A Novel Machine Learning Approach to Estimate Grapevine Leaf Nitrogen Concentration Using Aerial Multispectral Imagery. Remote Sensing, 12(21), 3515.
Our main objective in 2019 was to investigate nitrogen distribution within each vine further. We collected and assessed tissue samples from each vine (instead of combining 5 vines into one sample). Similar to 2018, we have 30 plots, each included 5 vines, and the lab analytics provided us with 150 nitrogen labels. In 2019 we had 5 times more labeled samples than in 2018, which would improve the N estimation model's robustness.
We collected very high-resolution multispectral imagery (altitude ~10 m) from each plot. The images were processed to segment each vine vegetation from the soil and adjacent vines. We used thresholds in NDVI and EGI indices for removing background pixels, then separated each vine using the markers we placed on the ground at the time of aerial imaging.
The following histogram shows the distribution of N content for the 150 vines in this experiment. This histogram suggests that N application rate is not necessarily correlated to nitrogen uptake in vine since we had N deficient vines that received maximum N application rate and vines with excess N that did not receive any N fertilizer.
Using the N content labels, we trained several supervised learning methods with either excess N or extremely deficient samples. Then we tested these trained models with all other vines in the middle. The following figure shows the prediction results using the top two supervised learning methods: XGBoost, and Random Forest; both provided very similar R-squared and RMSE.
We compared the performance of a supervised-learning method and vegetation indices to identify N content and determine the spatial distribution of N in vines. The figure below shows the supervised-learning method's results [P(excess|X)] compared to NDVI and NDRE. In the top, the comparison of histograms of two vines (one deficient and one excess) shows a clear separation between the two vines, while in NDVI and NDRE, we see many overlaps of pixel values. Also, P(excess|X) clearly illustrates N's distribution at different parts of each vine, while these details are not visible in vegetation index representations.
Study in 2018
Our main objective in 2018 was to investigate aerial multispectral imaging's potential for detecting grapevine Nitrogen status. We flew over a Flame Seedless vineyard in Selma, CA, once during the bloom and once before harvest. Tissue samples collected from each plot (of 5 vines) were mixed, and the lab analytics provided us with 30 nutrient labels for 30 experimental plots (1 label per 5 vines in a plot). Various potential vegetation indices (identified by other researchers for monitoring N) were extracted from aerial multispectral data and tested to check if different Nitrogen levels in the vine will lead to distinctive vegetation indices values. The following figure lists the vegetation indices tested in this study:
Correlation between leaf nitrogen and vegetation indices was calculated and plotted in the figure below. The blue columns show the R-squared value, and the orange line shows p-values for each index. The best correlation was achieved when R-squared was high and the p-value was low; this means the model significantly explains most of the variation within the data.
Based on the correlation results, NDRE, CGM, and MTCI showed the best correlation to vine nitrogen in table grape. This graph also showed that NDVI, i.e., the most commonly used vegetation index, was not as good as NDRE, CGM, or MTCI for detecting N content. Among the 20 indices tested in this study, NDRE was the best vegetation index at both growing stages (bloom and harvest).
The following figure shows the top three vegetation indices and NDVI for an experimental flame seedless vineyard at harvest time. Almost all four indices indicated the Nitrogen-deficient zones, but based on the regression results, NDRE is the most reliable.
These results confirmed that vineyard managers could obtain an insight into spatial variability of nitrogen throughout the entire vineyard using drone-based multispectral imagery.
Principal Investigator: Alireza Pourreza
Collaborator: Matthew Fidelibus
Project Sponsor: California Table Grape Commission