|STOCKER, MATTHEW - Orise Fellow|
|VALDES-AVELLAN, JAVIER - Universidad De Alicante|
Submitted to: Science of the Total Environment
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 11/24/2019
Publication Date: 11/26/2019
Citation: Morgan, B.J., Stocker, M.D., Valdes-Avellan, J., Kim, M.S., Pachepsky, Y.A. 2019. Drone-based imaging to assess the microbial water quality in an irrigation pond: a pilot study. Science of the Total Environment. 716:135757. https://doi.org/10.1016/j.scitotenv.2019.135757.
Interpretive Summary: One of the key issues in safety of irrigated produce is microbial quality of irrigation water. It is currently evaluated using E. coli concentrations in irrigated water sources. These concentrations vary spatially across irrigation ponds. We attempted to use images taken from small unmanned aerial vehicles (drones) to determine sections of ponds with different conditions for E. coli survival. It was found that images taken with modified GoPro cameras could characterize differences in E. coli habitats across the study pond with the same or better accuracy than a comprehensive set of water quality parameters measured in situ. Results of this work will be of use to irrigation water managers and consultants in that they indicate the possibility of using drone imagery to improve the design of comprehensive water sampling across irrigation ponds.
Technical Abstract: Microbial water quality data is essential for irrigated agriculture to prevent microbial contamination of produce. E. coli concentrations are commonly used to evaluate microbial water quality. Remote sensing imagery was successfully used to retrieve several water quality parameters that can be determinants of E. coli habitats in water bodies. The objective of this work was to carry out a pilot study test the possibility of using imagery data from a small unmanned aerial system (sUAS or drones) to improve the estimation of microbial water quality in small irrigation ponds. In situ measurements of pH, turbidity, specific conductance, and concentrations of dissolved oxygen, chlorophyll a, phycocyanin, and fluorescent dissolved organic matter were taken at the depth of 10-15 cm in 23 locations across a pond in Central Maryland. The pond surface was imaged from the drone with three modified GoPro cameras and a MicaSense Red Edge camera. The GoPro imagery was decomposed into red, blue, and green components, and mean digital numbers from the 2-m radius clips from the 14 images were combined with the water quality data to provide inputs for the regression tree-based data analysis. The accuracy of the regression tree data description with the ‘only imagery’ inputs was the same or better than for the trees with the ‘only water quality parameters’ inputs. From multiple cross-validation runs with ‘only imagery’ inputs for the regression trees, values of 0.793±0.035 and 0.131±0.011 were found for the average ± standard deviation of the determination coefficients and root-mean squared error of the decimal logarithm of E.coli concentration, respectively. Results of this study indicate the opportunity of using the sUAS imagery in obtaining a more accurate delineation of the spatial variation of E. coli concentrations in irrigation ponds.