Estimation of transpiration coefficient and aboveground biomass in maize using time-series UAV multispectral imagery

Authors: SHAO, GUOMIN - NORTHWEST A&F UNIVERSITY; HAN, WENTING - NORTHWEST A&F UNIVERSITY; Zhang, Huihui; WANG, YI - NORTHWEST A&F UNIVERSITY; ZHANG, LIYUAN - NORTHWEST A&F UNIVERSITY; NIU, YAXIAO - NORTHWEST A&F UNIVERSITY; ZHANG, YU - NORTHWEST A&F UNIVERSITY; CAO, PEI - NORTHWEST A&F UNIVERSITY

Submitted to: The Crop Journal
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 8/19/2022
Publication Date: 8/27/2022
Citation: Shao, G., Han, W., Zhang, H., Wang, Y., Zhang, L., Niu, Y., Zhang, Y., Cao, P. 2022. Estimation of transpiration coefficient and aboveground biomass in maize using time-series UAV multispectral imagery. The Crop Journal. 10(5):1376-1385. https://doi.org/10.1016/j.cj.2022.08.001.
DOI: https://doi.org/10.1016/j.cj.2022.08.001

Interpretive Summary: It is important to estimate maize above ground biomass (AGB) for irrigation and nutrient management during the crop growing season. Remotely sensed crop spectral information has been applied widely for AGB estimation. However, the accuracy of the prediction still needs to be improved due to the limitation of remotely sensed spectral indices, especially, when crop under water stress. Crop transpiration coefficient (CTc), the product of basal crop coefficient and water stress coefficient, according to FAO-56 method, should be highly correlated to AGB, by accounting for both canopy coverage and water stress level. In this study, by using high frequency drone multispectral images, we first developed a model to predict CTc with vegetation indices derived from drone images; and then exam that how the drone mapped CTc estimated maize AGB under different irrigation treatments. We found the accuracy of CTc prediction by drone multispectral vegetation indices was 0.90 and the accuracy of AGB prediction by drone derived CTc is about 0.76. The result could be used to improve algorithms in crop models, such as Aquacrop model, which used CTc for AGB prediction. Drone based imagery will provide spatial distribution of CTc and AGB within a field and support precision irrigation management at farm scales.

Technical Abstract: Understanding maize biomass response to water deficit and obtaining a high-resolution spatial distribution of them rapidly and accurately before harvesting is important for field irrigation management, especially in a semiarid region of Northwest China. Unmanned aerial vehicle (UAV)-based remote sensing provides a reliable tool to estimate crop growth parameters. This study developed and assessed a novel machine learning method to estimate crop transpiration coefficient (CTc) using time-series UAV multispectral vegetation indices (VIs), and then above-ground biomass (AGB) were estimated by the VIs-based cumulative CTc under different irrigation treatments at a field scale. Four machine learning regression methods, multiple linear regression (MLR), support vector regression (SVR), random forest regression (RFR), and adaptive boosting regression (ABR), were applied to correlate CTc and UAV VIs. AGB was then estimated using exponential, logistic, sigmoid, and linear equations based on the optimal CTc estimation model. The UAV VIs derived CTc using the RFR estimation model had the highest accuracy (R2 = 0.90, RMSE = 0.0554, nRMSE = 9.67%). The normalized difference red-edge index, transformed chlorophyll absorption in reflectance index, and simple ratio had high contributions to the RFR-based CTc model. The accuracy of AGB estimation using non-linear methods was better than that using the linear method. The exponential method had the best accuracy (R2= 0.76, RMSE = 282.8 g/m2) during the growing seasons in 2018 and 2019. The study confirmed that the AGB estimation models by cumulative CTc performed well under different irrigation treatments using high-resolution time-series UAV multispectral VIs and can provide effective support for irrigation management with high spatial precision at a field scale.