Determining leaf nutrient concentrations in citrus trees using UAV imagery and machine learning

Lucas Costa,Ute Albrecht,Sudip Kunwar,Yiannis Ampatzidis

doi:10.1007/s11119-021-09864-1

Lucas Costa, Ute Albrecht + Show 2 more

Open Access

https://doi.org/10.1007/s11119-021-09864-1

Copy DOI

Abstract

Nutrient assessment of plants, a key aspect of agricultural crop management and varietal development programs, traditionally is time demanding and labor-intensive. This study proposes a novel methodology to determine leaf nutrient concentrations of citrus trees by using unmanned aerial vehicle (UAV) multispectral imagery and artificial intelligence (AI). The study was conducted in four different citrus field trials, located in Highlands County and in Polk County, Florida, USA. In each location, trials contained either ‘Hamlin’ or ‘Valencia’ sweet orange scion grafted on more than 30 different rootstocks. Leaves were collected and analyzed in the laboratory to determine macro- and micronutrient concentration using traditional chemical methods. Spectral data from tree canopies were obtained in five different bands (red, green, blue, red edge and near-infrared wavelengths) using a UAV equipped with a multispectral camera. The estimation model was developed using a gradient boosting regression tree and evaluated using several metrics including mean absolute percentage error (MAPE), root mean square error, MAPE-coefficient of variance (CV) ratio and difference plot. This novel model determined macronutrients (nitrogen, phosphorus, potassium, magnesium, calcium and sulfur) with high precision (less than 9% and 17% average error for the ‘Hamlin’ and ‘Valencia’ trials, respectively) and micro-nutrients with moderate precision (less than 16% and 30% average error for ‘Hamlin’ and ‘Valencia’ trials, respectively). Overall, this UAV- and AI-based methodology was efficient to determine nutrient concentrations and generate nutrient maps in commercial citrus orchards and could be applied to other crop species.

Highlights

IntroductionAdoption of best management practices and development of superior food crop cultivars are necessary to cope with pressures imposed by different biotic and abiotic factors (e.g., pests, diseases, drought, water logging, salinity, nutrient deficiencies, extreme temperatures, etc.) and secure food production
Adoption of best management practices and development of superior food crop cultivars are necessary to cope with pressures imposed by different biotic and abiotic factors and secure food production
N, P and S returned low variability, but the data are still viable as these are the normal ranges for each of the nutrients for these trees. As this low variability could force the model to overfit, to evaluate this possibility, the mean absolute percentage error (MAPE) of the model created was compared to the coefficient of variance (CV) of the data itself

Summary

Introduction

Adoption of best management practices and development of superior food crop cultivars are necessary to cope with pressures imposed by different biotic and abiotic factors (e.g., pests, diseases, drought, water logging, salinity, nutrient deficiencies, extreme temperatures, etc.) and secure food production. Recent advances in plant breeding have accelerated the development of new crop cultivars to cope with a rapidly changing production environment affected by diseases and other plant stresses (Lenaerts et al, 2019) These new cultivars require evaluation on a large-scale in a commercial production setting to assess their horticultural traits, physiological needs and economic potential before being released for widespread commercial adoption. Identification of plant physiological needs through field scouting and hand sampling of tissues for laboratory analysis has been the traditional way to evaluate new cultivars and fine-tune management practices As these processes are laborious, costly and prone to human error, new phenotyping techniques are needed to advance crop selection and improve crop production (Li et al, 2014). The correlation of different models and crop indices based on spectral reflectance data has been studied to assess crop yield in wheat (Hassan et al, 2019; Mirasi et al, 2019), biomass in oat (Coelho et al, 2018), leaf area index in wheat (Xie et al, 2014), plant nutrient content in citrus and grapevine (Osco et al, 2019; Moghimi et al, 2020), detection of pests and diseases in citrus and avocado (Abdulridha et al, 2019a; Partel et al, 2019a, b)

Objectives

Methods

Results

Discussion

Conclusion