Remote estimation of leaf area index (LAI) with unmanned aerial vehicle (UAV) imaging for different rice cultivars throughout the entire growing season

Abstract

Abstract Background Rice is one of the most important grain crops worldwide. The accurate and dynamic monitoring of Leaf Area Index (LAI) provides important information to evaluate rice growth and production. Methods This study explores a simple method to remotely estimate LAI with Unmanned Aerial Vehicle (UAV) imaging for a variety of rice cultivars throughout the entire growing season. Forty eight different rice cultivars were planted in the study site and field campaigns were conducted once a week. For each campaign, several widely used vegetation indices (VI) were calculated from canopy reflectance obtained by 12-band UAV images, canopy height was derived from UAV RGB images and LAI was destructively measured by plant sampling. Results The results showed the correlation of VI and LAI in rice throughout the entire growing season was weak, and for all tested indices there existed significant hysteresis of VI vs. LAI relationship between rice pre-heading and post-heading stages. The model based on the product of VI and canopy height could reduce such hysteresis and estimate rice LAI of the whole season with estimation errors under 24%, not requiring algorithm re-parameterization for different phenology stages. Conclusions The progressing phenology can affect VI vs. LAI relationship in crops, especially for rice having quite different canopy spectra and structure after its panicle exsertion. Thus the models solely using VI to estimate rice LAI are phenology-specific and have high uncertainties for post-heading stages. The model developed in this study combines both remotely sensed canopy height and VI information, considerably improving rice LAI estimation at both pre- and post-heading stages. This method can be easily and efficiently implemented in UAV platforms for various rice cultivars during the entire growing season with no rice phenology and cultivar pre-knowledge, which has great potential for assisting rice breeding and field management studies at a large scale.