Details

Autor(en) / Beteiligte

• Meiyan, Shu
• Mengyuan, Shen
• Qizhou, Dong
• Xiaohong, Yang
• Baoguo, Li
• Yuntao, Ma

Titel

Estimating the maize above-ground biomass by constructing the tridimensional concept model based on UAV-based digital and multi-spectral images

Ist Teil von

• Field crops research, 2022-06, Vol.282, p.108491, Article 108491

Ort / Verlag

Elsevier B.V

Erscheinungsjahr

2022

Link zum Volltext

Quelle

Elsevier ScienceDirect Journals

Beschreibungen/Notizen

• Above-ground biomass (AGB) is an important basis for the formation of crop yield. The accurate estimation of maize AGB based on unmanned aerial vehicle (UAV) images is important for superior varieties selection, field management and maize yield prediction. The previous studies mainly focused on constructing empirical models of AGB by using spectral vegetation indices (VIs), plant height (PH), texture, and is lacked of universality. We conducted the field experiments of maize breeding materials for three years, and obtained UAV digital and multi-spectral images. Considering that the maize AGB before tasseling stage was composed of stem and leaf, we constructed a tridimensional concept model to predict maize AGB coordinated by integrating leaf area index (LAI) and PH, in order to improve the accuracy and universality of UAV data on monitoring maize AGB at multiple growth stages. Firstly, the maize PH was estimated based on the maize canopy height model constructed using the UAV digital images. Secondly, the maize LAI was estimated based on UAV multi-spectrum images and the modified Beer-Lambert law. Finally, the tridimensional concept model of maize AGB was constructed by integrating PH and LAI, and compared with the AGB regression model based on the normalized difference vegetation index (NDVI). The results showed that the maize PH could be estimated well, and the R², RMSE and rRMSE of the measured and estimated PH were 0.87, 11.17 cm and 16.04% respectively. The LAI could be estimated effectively, and the R², RMSE, and rRMSE of the sample set were 0.78, 0.49 and 30% respectively. Compared with the maize AGB estimation model based on NDVI (R² = 0.79, RMSE = 41.95 g/m², rRMSE = 31.79%), the tridimensional concept model could better estimate the maize AGB (R² = 0.82, RMSE = 38.53 g/m², rRMSE = 29.19%). Testing the tridimensional concept model by stand-alone data of 2019 and 2021 years, the accuracy of the AGB estimation model based on the tridimensional concept was much higher than that of the NDVI model. In conclusion，the tridimensional concept model of maize AGB proposed in this study effectively improved the accuracy, stability and universality, which could provide a reference for the estimation of maize AGB by UAV technology at plot scale of the breeding materials.