Unmanned aerial vehicle-based remote sensing in monitoring smallholder, heterogeneous crop fields in Tanzania

ABSTRACT

Obtaining information to characterize smallholder farm fields remains elusive and has undermined efforts to determine crop conditions for food security monitoring. We hypothesize that unmanned aerial vehicles (UAV) would provide high-resolution spectral signatures for effectively discerning agronomic and crop conditions, management practices, and yields in smallholder farms for crop yield outlooks. The current study explores potential in using UAV-mounted sensor spectral signatures for monitoring crop conditions in smallholder agriculture. Images were collected using a 4-band multispectral camera mounted on a small fixed wing UAV, flown at 8-day interval over maize–pigeonpea experimental plots at Sokoine University of Agriculture and maize monocrop in farmers’ fields nearby, during 2015/2016 growing season. Four spectral vegetation indices (VIs) namely; normalized difference vegetation index (NDVI), wide dynamic range vegetation index (WDRVI), red edge chlorophyll index (CI_red-edge), and the green chlorophyll index (CI_green), were evaluated under maize monocrop, maize pigeonpea-intercrop, fertilizer and non-fertilizer and two maize varieties conditions. VIs were used also to detect differences in farm management practices of two farmers’ maize fields. The response of the spectral VIs varied depending on phenological stage of the crop and imposed treatments or management practices. In experimental plots, NDVI was able to distinguish fertilized from non-fertilized plots at all times, distinguish between two maize varieties at 52 days after sowing (DAS), and differentiate monocropped maize from maize–pigeonpea intercrop at 60 DAS. CI_red-edge could detect effect of maize–pigeonpea intercrop and maize varieties at 44 DAS, whereas CI_green could detect variety differences at 44 DAS, intercropping effect at all times and fertilizer effects at 60 and 68 DAS. WDRVI could only detect variety differences and maize–pigeonpea intercrop at 44 DAS. Moreover, NDVI was slightly associated with maize yield in non-fertilized plots (coefficient of determination – R² = 0.58) and CI_green was associated with leaf area index (LAI) (R² = 0.62) in fertilized plots and in monocropped plots (R² = 0.61). CI_green could also differentiate well managed from poorly managed farmer’s fields. We conclude that UAV-derived spectral signatures can provide detailed information for characterizing agronomic and crop conditions under smallholder agricultural settings and aid food security monitoring efforts.

Acknowledgements

The authors are grateful to Mr. J. Kaingo (a PhD student at SUA) for technical support in UAV image collection and processing. We thank Prof. N. Kihupi and other farmers who permitted access to their maize fields during the study.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This study was generously funded by the Spurring a Transformation for Agriculture through Remote Sensing (STARS) Project [Sub award Number Q2549001] and the Climate Change Impacts Adaptation and Mitigation (CCIAM) programme.