Student Research Reports

View Research Report

The NASA-SEES 2021 Summer Internship affords high school students the opportunity to conduct research that advances Earth science. The SEES Earth System Explorers-Mosquito Mappers team investigates the relationship between mosquitoes and the environment in the context of human health. As apparent during Mosquito Mapper fieldwork, manually counting mosquitoes in a breeding habitat aids in the understanding of mosquito ecology. Absent a scientific approach, however, manual counts are error-prone and are deemed questionable for use in mosquito management models. To ensure that these counts inform meaningful scientific outcomes, the counting process needs optimization. As such, we explored the feasibility of automating the mosquito count process while minimizing error using ImageJ, an open-source image processor. To this end, images captured during the volumetric sampling of mosquito traps and supplemental images obtained from the GLOBE database were processed using ImageJ. A comparison of the manual and ImageJ counts revealed that both count types were largely unreliable as the difference between many of them exceeded a tolerable margin of error, and no count type was consistently more reliable due to citizen scientist technique and software limitations. Thus, the results do not support automation using ImageJ. Rather, they indicate that ImageJ’s performance depends on the “quality” of the image samples, thereby underscoring the need for standardized scientific methods in the mosquito counting process. However, it is improbable that citizen scientists will employ the counting methodologies of expert scientists since citizen scientists generally prize convenience over validity. An optimal solution may therefore involve a more robust algorithm that builds on the strengths of ImageJ and eliminates the citizen-scientist manual count upon integration into the GLOBE Observer Mosquito Habitat Mapper tool. Although more research is needed to assess the cost-effectiveness, such a multi-layered solution would assist scientists’ prediction of mosquito populations and management of mosquito-borne diseases. Key words: mosquito count, automation, data quality, ImageJ, citizen science