|Pilot Project with USGS National Map Corps|
The National Map Corps Colorado Pilot Project
GISCorps volunteers assist in mapping emergency services, schools, and other structures for The National Map
By David Litke, GISCorps volunteer
In November 2012, GISCorps received a request from the U.S. Geological Survey for assistance in creating and editing data for The National Map. The goals of the project were to: 1) Work towards completion of The National Structures Dataset for Colorado; and 2) Develop a framework for self-organizing of volunteers to promote efficiency and accuracy of crowd-sourced data. Twenty-three volunteers were recruited for the project which started in December 2012 and was completed in February 2013. As a result of this project, the number of structures data points reviewed increased from 2600 to 4678, and the number of topographic quads finalized increased from 30 to 253. The accuracy rate for edited data was close to 100 percent. Techniques used to coordinate the crowd included a Google Group forum for asking and answering questions, Google Documents cloud storage of guideline documents and spreadsheets to manage and track tasks, and Skype Group Conversations for live discussion among project members.
The National Geospatial Program of the U.S. Geological Survey (USGS) manages The National Map (figure 1) which includes 11 primary digital geospatial data themes for the United States and its territories. These datasets are maintained internally by the USGS and through a variety of partnerships with other governmental agencies. The USGS also has sponsored various forms of volunteer map data collection through a program called The National Map Corps (TNMC). In 2012, TNMC began a pilot project to test the efficacy of crowd-sourced data using internet-based mapping tools; this pilot project enables editing of data from the National Structures Dataset within the State of Colorado.
Figure 1. The National Map viewer, one product of the National Map.
In August 2012, the TNMC pilot project was opened to the public. This USGS website describes the project, provides guidelines and training materials, and has an editing interface (figure 2) based on the Potlatch editor developed by the OpenStreetMapproject. The USGS began an outreach program to various organizations to interest the public in citizen mapping. As part of this outreach, the USGS contacted GISCorps in November 2012, and a project was designed to 1) Work towards completion of The National Structures Dataset for Colorado; and 2) Develop a framework for self-organizing of volunteers to promote efficiency and accuracy of crowd-sourced data.
GISCorps sent out a recruitment letter to its 2000 members who are located in the United States and received interested responses from 60 volunteers; of these 23 were selected to take part in the project. The volunteers who participated are: Micah Babinski, Tanachy Bruhns, Karen Chadwick, Timothy Cooley, Chuck Failing, Cassie Follett, Matt Goodman, Chuck Gooley, Lara Holimon, Chris Kleinhofs, Chen Li, David Lok, Yang Lui, Richard Monteiro, Anders Olson, Katie Panek, Keera Pullman, Matthew Sloane, James Smith, Joy Straley, Lisa Templeton, and Carlos Vallejos. The GISCorps project manager was David Litke, and the USGS TNMC project manager is Greg Matthews.
Figure 2. The editing interface for the National Map.
The first task for GISCorps volunteers was to read the USGS editing guidelines and to become familiar with the editing interface, and then to make progress towards completion of the Colorado test dataset. The dataset to be edited was the National Structures point dataset; this dataset is one of the 11 primary themes of the National Map. Structures data support disaster planning and response and homeland security organizations and topographic mapping and resource planning needs. The National Structures Dataset has a comprehensive data model and specifications document; however, for the pilot project the data model was simplified to 10 (out of 44) structure types and 8 structure attributes. Each structure is represented as a point located at the approximate centroid of the feature (in most cases a building centroid).
The simplest and surest way to verify structure information is for a local volunteer to do an on-site visit. However, this is very time consuming, and can only be done where local volunteers are available. This project tested the efficacy of remote crowd-sourced work. Remote editing of structure data is essentially an exercise in data mining using the Internet, which is a skill commonly needed for crowd- sourced projects. This project made use of a volunteer community of GIS professionals, whose skills in GIS data analysis can improve the efficiency and accuracy of data review. For example, volunteers used desktop GIS to construct maps (figure 3) which provide a reference for editing structures data within Colorado counties.
Figure 3. Map showing Morgan County, Colorado, school district boundaries, cities, and quadrangle boundaries.
A second way to improve efficiency and accuracy, and a second goal of this project, was to develop tools for volunteer community interaction and organization. Tools used included Google Drive, Google Groups, and Skype; these tools are freely available, although the Google applications require establishing an on-line Google account, and Skype requires installation of software and establishment of a Skype account.
Google Drive was used to store documents of general interest. For example, one volunteer wrote and made available on Google Drive an “At-A-Glance Document for TNMC Volunteering” which listed in concise steps how to begin TNMC editing. Editable spreadsheets also were stored on Google Drive, and were used to assist workers in coordinating their work. For example, a School Districts spreadsheet (figure 4) was made available on Google Drive where volunteers could “sign-out” a School District for editing, and then when finished indicate in the spreadsheet that all the schools in that School District had been edited.
Figure 4. Colorado School Districts spreadsheet used to coordinate editing of school structures.
A Google Group (figure 5) was established as a forum where volunteers could: ask and answer questions; post links to documents of interest; and document decisions made on proper editing protocol.
Figure 5. Google Group used for the TNMC project.
A Skype Group Conversation was set up so that volunteers could have Internet Messaging (IM) meetings. Text-only meetings are the preferred channel because with large groups too much band-with is required for audio/video meetings. Volunteers can enter the chat room at any time and chat with other volunteers who happen to be online. Scheduled meetings also were announced to maximize the number of volunteers present at a given time.
As a result of this project, the number of structures data points reviewed increased from 2600 to 4678, and the number of topographic quads finalized increased from 30 to 253. The accuracy rate for edited data was close to 100 percent. The USGS viewed this project as a proof-of-concept regarding crowd-sourced work, and a decision was made to roll-out the scope of this project to the entire United States by the end of 2013.
Community tools were found to be a critical part of this crowd-sourcing project. The availability of free internet tools enables crowds to organize themselves. We chose to use Google apps due to their integrated nature, but many other choices are available; Skype was chosen for communication because it is free and offers a variety of communication channels.
Our Google Group was set up under the personal Google account of the GISCorps project chief as an invitation-only group. The Group worked well as a question-and answer forum and central location for access to community documents (through links embedded in posts); volunteers ranked this Group as an important project asset. It is important to provide structure to the Group by initially setting up principal Topics; we did not use the new Google Group capabilities of Tags, Categories, and Group Categories, but this would be another way to structure the Group. Although Google Groups has a good Search capability, volunteers felt that posts relating to policy should periodically be consolidated and appended into the Guidelines document for ease of reference.
Google Drive proved to meet our needs for document storage. The GISCorps project chief set up a project folder within his own personal Google Drive account and documents within the folder were shared (with edit rights) to anyone with the link; the links to documents were provided through posts in the Google Group. Most of our documents were spreadsheets stored in Google Drive format which therefore did not count against the 5GB size limit of a personal account.
Skype Group Conversations met our needs as an IM mechanism. The Skype requirement of desktop software installation is a barrier for some volunteers, and the mechanics of setting up and managing Skype groups are not completely intuitive. Although most volunteers desired real-time communication, scheduled chats did not meet those needs because attendance was poor; perhaps this was because the size of our volunteer group was relatively small, but a more intuitive and non-intrusive IM solution might be more effective for both scheduled meetings and ad-hoc meetings.
During the course of the project, volunteers had many suggestions for improving the TNMC website, the editing interface, and the TNMC documentation and guidelines; many of these suggestions were subsequently acted upon by the USGS. As GIS professionals, volunteers sometimes were frustrated by the inability to use advanced GIS techniques to streamline their tasks; for example, additional map themes could not be introduced as overlays into the editing interface via outside web mapping services. Also, authoritative lists of structures (such as lists of schools available from the Colorado Department of Education) could not be compared in batch mode to existing structures through proximity analysis, and edits could not subsequently be made through attribute transfer. Despite these frustrations, most volunteers found this project to be fulfilling. Their GIS knowledge enabled them to work more efficiently, and the development of skills in internet-based data mining provided a challenge. Volunteers also found gratification in donating their time towards a beneficial national purpose.