Identifying Potential Mosquito Breeding Grounds: Assessing the Efficiency of UAV Technology in Public Health

Abstract

Human ecology has played an essential role in the spread of mosquito-borne diseases. With standing water as a significant factor contributing to mosquito breeding, artificial containers disposed of as trash—which are capable of holding standing water—provide suitable environments for mosquito larvae to develop. The development of these larvae further contributes to the possibility for local transmission of mosquito-borne diseases in urban areas such as Zika virus. One potential solution to address this issue involves leveraging unmanned aerial vehicles that are already systematically becoming more utilized in the field of geospatial technology. With higher pixel resolution in comparison to satellite imagery, as well as having the ability to update spatial data more frequently, we are interested in investigating the feasibility of unmanned aerial vehicles as a potential technology for efficiently mapping potential breeding grounds. Therefore, we conducted a comparative study that evaluated the performance of an unmanned aerial vehicle for identifying artificial containers to that of conventionally utilized GPS receivers. The study was designed to better inform researchers on the current viability of such devices for locating a potential factor (i.e., small form factor artificial containers that can host mosquito breeding grounds) in the local transmission of mosquito-borne diseases. By assessing the performance of an unmanned aerial vehicle against ground-truth global position system technology, we can determine the effectiveness of unmanned aerial vehicles on this problem through our selected metrics of: timeliness, sensitivity, and specificity. For the study, we investigated these effectiveness metrics between the two technologies of interest in surveying a study area: unmanned aerial vehicles (i.e., DJI Phantom 3 Standard) and global position system-based receivers (i.e., Garmin GPSMAP 76Cx and the Garmin GPSMAP 78). We first conducted a design study with nine external participants, who collected 678 waypoint data and 214 aerial images from commercial GPS receivers and UAV, respectively. The participants then processed these data with professional mapping software for visually identifying and spatially marking artificial containers between the aerial imagery and the ground truth GPS data, respectively. From applying statistical methods (i.e., two-tailed, paired t-test) on the participants’ data for comparing how the two technologies performed against each other, our data analysis revealed that the GPS method performed better than the UAV method for the study task of identifying the target small form factor artificial containers.