The potential of UAV and very high-resolution satellite imagery for yellow and stem rust detection and phenotyping in Ethiopia

Abstract

Very high (spatial and temporal) resolution satellite (VHRS) and high-resolution unmanned aerial vehicle (UAV) imagery provides the opportunity to develop new crop disease detection methods at early growth stages with utility for early warning systems. The capability of multispectral UAV, SkySat and Pleiades imagery as a high throughput phenotyping (HTP) and rapid disease detection tool for wheat rusts is assessed. In a randomized trial with and without fungicide control, six bread wheat varieties with differing rust resistance were monitored using UAV and VHRS. In total, 18 spectral features served as predictors for stem and yellow rust disease progression and associated yield loss. Several spectral features demonstrated strong predictive power for the detection of combined wheat rust diseases and the estimation of varieties' response to disease stress and grain yield. Visible spectral (VIS) bands (Green, Red) were more useful at booting, shifting to VIS-NIR (near-infrared) vegetation indices (e.g., NDVI, RVI) at heading. The top-performing spectral features for disease progression and grain yield were the Red band and UAV-derived RVI and NDVI. Our findings provide valuable insight into the upscaling capability of multispectral sensors for disease detection, demonstrating the possibility of upscaling disease detection from plot to regional scales at early growth stages.

Figure 1

Location of the study site and experimental setup (experiment site: UAV false color composite using near infrared, red and green bands from 2020-10-29; heading stage / DAS 80; study site: Pleiades satellite scene from 2020-10-16). UAV imagery processed using Pix4Dmapper and figure prepared using QGIS version 3.14.16-PI.

Figure 2

Overview of the data collection timeline (red dashed line: temporal focus window; light red box: sampling date; In. SR: inoculation of SR; F: fungicide application; DAS: day after sowing; YR: yellow rust; SR: stem rust; PL: Pleiades; SS: SkySat; UAV: unmanned aerial vehicle; int: interpolated).

Figure 3

Disease progression per wheat variety and treatment block for disease merging scenarios—(a) Scenario X (simple sum) and (b) Scenario W_0.7 (weighted sum).

Figure 4

Box plot showing recorded yield per variety and treatment block.

Figure 5

Correlation between grain yield and AUDPC—(a) Scenario X (simple sum) and (b) Scenario W_0.7 (weighted sum) (AUDPC_X: AUDPC for Scenario X; AUDPC_W: AUDPC for Scenario W_0.7).

Figure 6

Scenario W with tested weight coefficients – Pearson correlation between DI_norm and spectral features derived from (a) UAV, (b) SkySat, and (c) Pleiades at DAS 80 (non-fungicide treatment block) (note: Scenario W_1.0 with the weight coefficient of 1.0 is the same as Scenario X; Scenario W₀ with the weight coefficient of 0 excludes the SR impact).

Figure 7

Predicted versus measured DI_norm values at DAS 60 based on the regression equation of the linear model using (a) G derived from UAV, (b) R derived from SkySat, and (c) G derived from Pleiades data (black line: one-to-one line) (note: the results are the same for both scenarios).

Figure 8

Scenario W_0.7—Predicted versus measured DI_norm values at DAS 80 based on the regression equation of the linear model using (a) RVI derived from UAV, (b) RVI derived from SkySat, and (c) RVI derived from Pleiades data (black line: one-to-one line) (results for Scenario X are presented in Supplementary Data S Fig. 1).

Figure 9

Scenario W_0.7—Predicted versus measured AUDPC_W values based on the regression equation of the linear model using (a) RVI-AUC derived from UAV and (b) R-AUC derived from SkySat data (black line: one-to-one line) (results for Scenario X are presented in Supplementary Data S Fig. 2).

Figure 10

Predicted versus measured grain yield values based on the regression equation of the linear model using (a) RVI-AUC derived from UAV and (b) R-AUC derived from SkySat data (black line: one-to-one line).