Estimation of potato chlorophyll content using composite hyperspectral index parameters collected by an unmanned aerial vehicle

Abstract

ABSTRACT Real-time monitoring of the nutritional status of potato crops enables rational and efficient decision-making about planting patterns and fertilization strategies that maximize yields. Chlorophyll is a useful index for measuring potatoes’ nutritional status. Therefore, rapid and accurate estimation of chlorophyll content can be used to guide efforts to improve potato crop quality and yields. Here, we use hyperspectral potato crop data collected by an unmanned aerial vehicle (UAV) and correlate the data with vegetation indexes, spectral position, and area characteristic parameters, spectral resolution, and other index parameters to comprehensively analyse the chlorophyll content of experimental potato crops at different growth stages. We establish a model for estimating chlorophyll content and verify the accuracy of the model by using partial least squares, stepwise regression analysis, support vector machine, and random forest analytical methods. This study provides a new method for estimating the chlorophyll content of crops by using hyperspectral data. We find that the partial least squares (PLS) model based on hyperspectral reflection characteristic variables is optimal for estimating chlorophyll content during the budding and tuber stages of potato growth. The optimal model during tuber formation and starch accumulation is the stepwise regression model on the basis of vegetation indexes and spectral position and area characteristic parameters. Comprehensive results show that compared with the single index parameter, the comprehensive index parameter can be used to estimate the chlorophyll content of potatoes with higher accuracy and better effect; it can also be used to monitor the nutritional status of potatoes.