Student Led Area Measurement Assessments Using Virtual Globes and Pictometry Web-based Interface within an Undergraduate Spatial Science Curriculum

The use of Virtual Globes and Pictometry continues to expand and develop in undergraduate spatial science education. Spatial science undergraduates measured the area of 30 rectangles on the earth’s surface and compared them to Pictometry hyperspectral imagery measurements within a web-based interface and the Google Earth interface compared to ArcGIS Explorer, Map Developers and ArcMap using the ArcMap 10.5.2 interface. An analysis of variance of the absolute mean area errors (p-value of 0.009271) concluded the accuracy of the five area measurements were statistically different at the 95% confidence interval. A Tukey pair-wise test found that the Pictometry and Google Earth methods were more accurate than the ArcGIS Explorer, Map Developers and ArcMap methods. The lowest standard deviation of errors (72.6 sq. ft.) for Pictometry was the most accurate and precise method for on-screen area measurement, followed by Google Earth (SD = 205.0 sq. ft.). The high variation of area measurement error from ArcMap, Map Developers, and ArcGIS Explorer made them less reliable as an alternative to field measurements with ArcMap the worst (SD = 915.1 sq. ft.). The results indicate that Pictometry and Google Earth could both be used to accurately estimate area using on-screen measurements compared to in situ area measurement assessments.


Introduction
Virtual Globes and Pictometry, which represent ways to display earth spatial features within a computerized environment, assisted undergraduate spatial science students in a capstone course in the Arthur Temple College of Forestry and Agriculture (ATCOFA) at Stephen F. Austin State University (SFASU). Undergraduate spatial science students used Virtual Globes and Pictometry to visualize the world in virtual terms and to process the three content standards of geography. These standards include how to use maps and acquire and process spatial information, how to organize information in a spatial perspective, and how to analyze spatial organization of people, places and the environment. This allows spatial science students to explore the virtual world to facilitate learning essential for long-term learning of spatial science measurements (National Academy of Sciences, 2006;Schultz et al., 2008).
Spatial science students in ATCOFA perform well in applying analytical skills to measure and predict, use geospatial technologies to measure natural resources, use math and statistics for analysis and problem solving, and analysis of priorities to solve real-world problems (Bullard et al., 2014). These are desirable traits in training students for real-world geospatial applications to promote critical thinking, technological knowledge and hands-on engagement (Unger et al., 2016a;Sattar et al., 2017).
Virtual Globes and Pictometry use geobrowser technology to integrate multiscale and temporal data from multiple sources to develop tools for planning for stakeholders (Sheppard & Cizek, 2009;Schroth et al., 2011). The spatial concepts of onscreen measurements combined with in situ ground truthing combines the accuracy and precision of the methods for training in natural resource management.
Criteria for evaluating landscape visualization for Virtual Globes and Pictometry include the accuracy of measuring features in the landscape, is easily accessible to the public, and be presented in a clear and concise fashion.
Undergraduate spatial science students in ATCOFA at SFASU measured ground rectangles in an urban environment to evaluate tools in Virtual Globes and Pictometry to accurately assess earth surface measurements. After training the students in each method in a senior capstone spatial science course, students measured 30 rectangles clearly visible across remotely sensed imagery in an urban setting. The rectangles were each identified, then measured for length and width on-screen for each of the five techniques analyzed (e.g., Pictometry, ArcGIS Explorer, Google Earth, ArcMap, and Map Developers) to compare the effectiveness of each method. To obtain the actual area of each surface feature identified, as a way to validate each method's area assessment, a measuring tape was used in situ to measure the real-world length and width of each rectangle.
Working together, students developed the methodology for spatial science research projects to prepare for their careers (Newman et al., 2007;Bullard et al., 2014;Unger et al., 2016a). The students also learned via one-on-one faculty interactions that while working collaboratively students can acquire and retain information better than working individually (McBroom et al., 2015;Viegut et al., 2018).

Virtual Globes
Virtual Globes display the earth as a three-dimensional globe that one can "fly" above. Virtual Globes run software with spatial data that is streamed from the internet including commercial or public domain satellite, aircraft or Unmanned Aerial Systems (UAS) data. Virtual Globes began in 2001 with EarthViewer from Keyhole, Inc. that was acquired by Google in 2004. This platform allowed users the ability to visualize, locate and navigate through their own geospatial data (Bailey & Chen, 2011). These data can be measured as polygons for area or perimeter assessments (Kerski, 2008;Schultz et al., 2008;Viegut et al., 2018).
These digital spatial data can format directly into ArcGIS Explorer or Google Earth. Virtual Globes are critical for learning to think spatially for points, lines, areas and volumes (National Academy of Sciences, 2006) and are an alternative to traditional Geographic Information Systems (GIS) with data sharing at the global scale (Yu & Gong, 2012). Virtual Globes were downloaded over 500 million times by Google Earth by 2010 and now is a standard method of on-screen data measurement.

Google Earth
Google Earth can assist in learning spatial skills while students work independently or collaborating with others to solve a real-world question (Patterson, 2007). Google Earth is not a GIS but can incorporate data sets incorporated by users . By using Google Earth, students expand their thinking and analytical skills, important in high impact teaching (Unger et al., 2016b).
Google Earth enhances the four E's of the learning cycle proposed by Cates et al. (2003) including: engaged in the lesson, explore the earth, explain what they identify, and evaluate the information.
Google Earth on-screen data measurements of a surface object can be ground truthed for validation (Yu & Gong, 2012) and then used in additional locations (Potere et al., 2009). Once polygons are identified and measured on-screen in Google Earth, they can be exported as vector layers into Keyhole Markup Language (KML) format for use in natural resource measurements. These polygons can then be compared overtime for change in the shape or land use of these features (Yu & Gong, 2012).
For measurement in Google Earth, latitude and longitude can be taken at each point on a polygon and used as reference for either distance or area calculations. Since the data are spatially referenced, the polygon can be located either visually in another measuring system or referenced spatially with the coordinate systems for use in the next generation Digital Earth (Goodchild et al., 2012).

ArcGIS Explorer
ArcGIS Explorer is a virtual globe software that is part of the ESRI ArcGIS software suite. ArcGIS Explorer brings GIS analysis capabilities to a free virtual client with geo-processing support (Kienberger & Tiede, 2008) that can be loaded onto handheld computers (Kerski, 2008). ArcGIS Explorer is part of the Virtual Globes that display spatial data on top of a three-dimensional globe on a computer and is periodically updated with increasing resolution and scale. This includes high spatial resolution satellite data that students can access to integrate and utilize GIS services and geographic content (Kerski, 2008).

Pictometry
Area measurement in Pictometry uses the Area tool by clicking on each vertex of a polygon. Pictometry, classified as hyperspatial resolution remotely sensed digital imagery, can be acquired at three or four band options. Images are taken by aircraft include both nadir and oblique angles up to 40 degrees with a spatial resolution ranging from 3 to 6 inches. Pictometry imagery depicts fronts and sides of vertical ground features and the ground surface features in a web-based interface. A composite Pictometry image can be used to accurately measure a polygon size and position using the patented Pictometry web-based interface (Wang et al., 2008). Previous assessment of height indicated no difference between Pictometry and a measuring pole for baldcypress tree height , building height ) and a UAS DJI Phantom 3 for measuring light poles when the UAS was not landed between measurements (Unger et al., 2016a).

ArcMap
Area measurement within ArcMap, also part of the ESRI software suite of spatial analysis software, starts by first loading a digital image of the area in question into an ArcMap data frame. Once loaded, the scale of the digital image can be adjusted by zooming in and out of the area in question until the area assessment location boundary can be easily identified. After identifying the boundary area in question via a visual assessment, the Measure tool in ArcMap is opened allowing the user to outline the area via the polygon measure tool to measure and record area in the user's choice of real-world units.

Methodology
Five measuring tools were used by seniors in a capstone spatial science course to measure the area of 30 rectangles in an urban setting on a computer screen (Figure 1). The tools used were Pictometry, ArcMap, Google Earth, Map Developers, and ArcGIS Explorer. The rectangles were located and measured on-screen in square feet. The actual ground truth measurements were taken with a measuring tape. Measurements were recorded in an Excel spreadsheet and compared using a visual assessment of errors, and then the average and standard deviation distribution of errors for each method were calculated and compared. The average of absolute errors and root mean square error (RMSE) for each method were also calculated and compared. A one-way analysis of variance (ANOVA) was conducted comparing the absolute errors for each method with an alpha of 0.05. If a significant difference among the measurement methods was found a Tukey's pair wise comparison on the average of absolute errors was calculated.

Results
A total of 30 rectangular polygons were identified on a computer screen and measured for their area using an on-line area measurement tool. This process was repeated on the same 30 polygons using five different programs (e.g., Pictometry, ArcMap, Google Earth, Map Developers, and ArcGIS Explorer).
The actual dimension of each rectangle was measured on the ground using a measuring tape. The area values (square feet) were calculated by multiplying length with width of each polygon for the five indicates that Pictometry performed better with errors clustered and centered on zero, followed by Google Earth while the remainder of the methods had widespread errors (Figure 2). The average of errors for each measuring method indicates whether or not a method is either overestimating or underestimating the area as a whole (Table 2). To test the statistical significance on the accuracy of on-screen area measurements, an analysis of variance (ANOVA) was conducted on the absolute errors for each measurement method analyzed (  varied from 29 to 64,865 sq. ft. in an urban setting indicating the usefulness of area measurements over these dimensions.

Figure 2. Distribution of Area Measurement Errors by All Five Methods Analyzed
These results indicate that care needs to be taken in measuring spatial area with available tools. Google Earth as a Virtual Globe product performed as well as Pictometry with a 4 inch spatial resolution.
Undergraduate spatial science students produced accurate onscreen digitizing of rectangles with minimal instruction, crucial for measuring quantitative data for natural resource planning Viegut et al., 2018).
The continued improvement in on-screen measurements indicates their importance in hands-on instruction in these techniques for ATCOFA spatial science students. These techniques can be applied within SFASU and at other universities. The reinforcement of the importance of use of high end technology coupled with analytical skills, use of geospatial technologies and ability to solve problems validates the use of spatial science for multidisciplinary problem solving (Bullard et al., 2014).
The students learned via one-on-one faculty interactions that while working collaboratively students can acquire and retain information better than working individually. Working together spatial science students learned to develop methodology to explore the virtual world to facilitate learning essential for long-term retention of spatial science applications throughout their entire careers.

Conclusions
The use of remotely sensed area measurements achieved by undergraduate spatial science students to measure earth surface features within the web-based Pictometry interface, Google Earth interface, Arc Map, Map Developers, and ArcGIS Explorer indicates their use in lieu of in situ assessments. The Tukey pair-wise test indicates that the remotely sensed Pictometry web-based interface and the Google Earth interface provided the most accurate measurements while ArcMap was least accurate.
This indicates that both Pictometry and Google Earth can be used for on-screen digitizing of polygons.
These results mirror the results of Viegut et al. (2018) that found no significant difference between Pictometry and Google Earth Pro, but both were statistically more accurate than the DJI Phantom 4 Pro UAS. This study reinforces the accuracy of Pictometry and Google Earth for area measurements and the value of student-driven hands on research in spatial science assessments. This interactive hands-on approach of learning with real-world data will prepare the students for their future careers.