Accuracy of Volunteer-Derived Data from a Single-Day Inventory Event Built Around a Crowdsourced Tree Mapping Application
Keir Hamilton, Andrew K. Koeser, and Shawn M. Landry
Abstract: Freely available ecosystem service models, like those incorporated in the i-Tree suite of tools, have helped scientists and practitioners estimate the environmental functions and economic benefits associated with their urban forest. Traditionally, professional inventory crews have been used to collect the inventory data needed for these models, but several cities have established crowdsourcing platforms to allow volunteers to map and inventory trees. Students in this study hosted and participated in an Arbor Day inventory collection event, using a newly released crowdsourcing application for mapping trees and estimating ecosystem services. The students located, identified, and measured trees on the University of South Florida campus (Tampa, Florida, U.S.) after a brief training session. After the one-day event, a more rigorously-trained field crew attempted to relocate the inventoried trees to assess the accuracy and variability of the data collected. Of the 339 trees inventoried at the original event, only 57.8% (n = 196) had coordinates that were accurate enough to re-measure. Of the 196 re-measured trees, 91.3% (n = 179) were correctly identified. However, only 47.9% (n = 91) of trees had dbh measurements within a one inch (2.5 cm) threshold for accuracy. Results of this experiment offer insights for communities looking to host special inventorying events to increase participation in crowdsourcing tree inventory initiatives.
Keywords: Accuracy of Volunteer-Derived Data from a Single-Day Inventory Event Built Around a Crowdsourced Tree Mapping Application