Burley Cemetery

Burley Cemetery is a small (.16 ha) pioneer cemetery located at  47.41174, -122.63635  in Kitsap County. It spans two parcels that are owned by the Burley Association. The cemetery is maintained by the association, which is documenting its history and improving the grounds for future generations. This includes hiring a consulting firm (Aqua Terra Cultural Resources Consultants) to conduct an archaeological assessment using ground penetrating radar (GPR) and a certified Historic Human Remains Detection firm (Canine Forensics Foundation). On August 20th, 2024,  Gig Harbor NOW  published a comprehensive news article describing efforts to locate 38 additional bodies believed to be buried in the cemetery.

The objectives of this site characterization and field survey are to:

• Help the Burley community understand where specific individuals are currently buried.
• Characterize the area in advance of ground penetration radar (GPR) and cadaver dog surveys of the cemetery.
• Explore digital terrain models (DTM) and terrestrial Lidar (TLS) to identify potential subsidence patterns.
• Analyze results and share data with the public.

The project is organized into four phases: 1) define the historical context, 2) plan for field data collection, 3) collect field data, 4) analyze and share data using a web application.

The cemetery has changed over time, both in terms of land cover and land usage. Understanding these changes requires detecting environmental changes and understanding the patterns of interment from the earliest burial to the present. To detect the environmental changes we contrast a historical aerial photo, NAIP satellite imagery from 2021 and contemporary orthomosaics captured using an entry-level professional drone (DJI Mavic 3 Enterprise). We performed semantic segmentation on the orthomosaic to determine the feasibility of supervised classification in the future.

To understand human activity we reviewed an inventory of interred individuals and assigned each to a decade based on their date of death. This allows us to see the peak of activity at the cemetery as well as the subsequent decline in later years. In the analysis we examine burial patterns by time and location.

The earliest aerial photograph of the cemetery is from 1951. This black and white photograph reveals that the two cemetery parcels are largely clear of vegetation (figure 1). A “J”-shaped stand of trees defines the cemetery’s boundary to the south and east, while the western boundary has been cleared of ground cover up to the primitive dirt road to the west. The yellow polygons outline the legal boundaries of the two plats that make up the cemetery.

The second image comes from the 2021 NAIP survey of Western Washington (figure 2). NAIP stands for National Agriculture Imagery Program, a program run by the U.S. Department of Agriculture's Farm Production & Conservation Business Center (FPAC-BC) Geospatial Enterprise Operations (GEO) Branch in Salt Lake, Utah. The differences are easy to see – with the exception of the center of the cemetery, the property is now surrounded by a stand of secondary growth trees. The resolution of the NAIP imagery (.6 sq m per pixel) does not provide much detail about the open area of the cemetery. A Normalized Difference Vegetation Index (NDVI) analysis is also not useful as the tree canopies are already visible in a 2006 NAIP image, and a Near Infrared (NIR) band is not available until 2011.

Given the small area of interest and the need for high resolution imagery and digital elevation models, three options were available: a) task a commercial satellite to cover the area (cost prohibitive), b) conduct an aerial Lidar survey (also cost prohibitive), or c) fly digital drone aerial photography (dDAP) flights. The third approach was chosen and flights were conducted from 1:00pm to 2:00pm on 7/12/24. These resulted in high quality data products. Unfortunately, the height of the trees on the periphery of the cemetery required the flight to be conducted at an altitude of 50 m Above Ground Level (AGL). The resulting orthomosaic image covers the area with a pixel resolution of 6.9 cm2 (figure 3). The details of the cemetery are in clear focus – with a significant limitation. There are numerous marked graves located beneath the tree canopy.

An existing inventory of interred individuals is available from an (apparently reliable) source. The  Washington Department of Archaeology and Historic Preservation  links to the  US GenWeb Tombstone Transcription Project (Washington) , which in turn includes an  inventory  of Burley Cemetery.

This list was processed in Excel and turned into structured data. Individual dates of death were extracted to a separate field and then assigned to its respective decade. The resulting histogram shows the frequency of burials by decade (figure 4):

An observation is that the majority of burials (58 out of 80) occurred before 1950. This is consistent with the observation that the cemetery grounds were maintained (or at least kept cleared) during the period of greatest activity. (Note: at the time of publication an additional grave was discovered - that of Mary Jeanette Herbert. She represents the last known burial in the cemetery (2007)).

To prepare for formal field data collection it is helpful to understand the topology of the site as well as the spatial patterns available from data collected during the first phase.

Based on the Washington State Lidar data we see that the site has a pronounced slope (-7.3 m) from west to east (figure 5). While easy to navigate on foot, it complicates maintenance (e.g. mowing).

DroneLink was used the design the flight plan and transmit it to the DJI Mavic 3 Enterprise. The orthomosaic flight took thirteen and a half minutes, covered 1.8km and resulted in 298 pictures (figure 6):

To provide a 360-degree view of the cemetery, Drone2Map was used to create an interactive 3D web scene:

Digital Terrain Models (DTM) (sometimes called Digital Elevation Models) are topographic models of the bare Earth that can be manipulated by computer programs. The data files contain the elevation data of the terrain in a digital format which relates to a rectangular grid. Vegetation, buildings and other cultural features are removed digitally - leaving just the underlying terrain. DTMs are used especially in civil engineering, geodesy & surveying, geophysics, geography and remote sensing. [Source: European Environment Agency]

To characterize the site's topography we examined three different digital terrain models. The first is based on the Pierce County DTM Hillshade (.3 sq m / pixel) obtained from the WA State Lidar portal. The second (6.9 cm ²  / pixel) and third (.56 cm ²  / pixel) were generated from drone flights at 50 m and 2.2 m (AGL), respectively. They reveal interesting patterns with significantly more detail being visible in the two DTM shaded relief outputs generated from lower altitudes. See the map in the Appendix for a side-by-side comparison and explanation.

A FARO Focus 3D Laser Scanner was borrowed from the University of Washington's  Remote Sensing and Geospatial Analysis Lab . Data was collected in four locations near the center of the cemetery. The FARO SCENE application was used to automatically register the scans and create a 3D colored point cloud (figure 10). The ground resolution falls short of what was achieved with the low-altitude drone flights; however the effort was constrained by available time and the value of TLS remains an open question.

By taking aerial pictures of a small section at a very low level (2.2 m AGL) we were able to generate a Digital Surface Model (DSM) showing possible subsidence in the southern section of the cemetery (figure 11). This is consistent with community reports that approximately 38 burial sites are unmarked.

As the site survey progressed, we connected with Burley Library Association which resulted in a request to make the gravesite inventory suitable for historical research, public education and ongoing updates as new information becomes available. To achieve this the data collection requirements became:

• To accurately identify and document each gravesite in the cemetery. This included well-marked and partially marked gravesites as well as locations where a burial was suspected (e.g., a rock or stake).
• To define a database schema that met the needs of the association. This was done in consultation with Bruce Morse, association president, and his contacts in the community.
• To accurately geotag each gravesite record using a GNSS receiver with a goal of +/- 25 cm absolute accuracy.
• To enable demographic analysis of the cemetery as it was populated across the decades.

Survey123 Connect was used to design the form for mobile data collection. Data was grouped into six categories:

• Cemetery (name, date & time)
• Gravesite (photo(s), GNSS coordinates, # of people buried at the site, gravesite type, border, visibility, vegetation presence, comment)
• Marker (type, condition, material, name visibility, symbols/iconography, epitaph, comment)
• Individual (last name, first name, middle name, gender, year born, year died, age at death, additional date information, veteran, service details, comment)
• Accuracy information (source type, accuracy type, PDOP, satellites visible, satellites in use, fix type)
• Coordinate information (UTM zone, UTM easting, UTM northing, latitude, longitude, altitude)

In addition a number of hidden attributes were added at the request of the community association. These will be used in the future to build more detail about the individuals buried in the cemetery.

An Emlid Reach RS3 GNSS receiver was used with the UNAVCO NTRIP service to correct latitude and longitude in real time. Both FIX and FLOAT accuracy was observed during data collection, with an average horizontal accuracy across 122 readings of 9 cm (95% confidence interval).

To complete the picture, the Burley Community contracted with third-party consultants to identify possible human remains. Data from the K9 cadaver search was added to the project; the ground penetrating radar data may be added in the future.

Using ArcGIS Pro, we explored a technique to identify approximately 24 previously unknown locations of human remains. The approach was to find a least 2-meters of separation between the K9 flags and the nearest grave markers.

Steps:

1. Create radius buffers around the survey points (i.e. the lat/lon coordinates). The first screenshot shows 1 meter buffers around positively identified grave markers (green circles) and 1 meter buffers around K9 detection flags (blue circles).
2. Using the "Select By Location" tool (second screenshot), find those flag buffers that don't overlap with grave marker buffers. (The 1 meter buffer size was arbitrary. It was a guess as we don't know the typical circular error rate for the cadaver dogs. The buffer size can be fine-tuned once the GPR data is available.)
3. The red highlighted flag buffers (third screenshot) represent unmarked locations that were identified by the dogs and are likely to have human remains. The grave marker and flag coordinates are separated by at least 2 meters.

This approach (inverted spatial intersection) requires minimal data processing and results in a useful visualization.

An ArcGIS Experience Builder application was developed to provide the Burley Community and other interested parties with the ability to view and analyze the data. The web application is available here:  https://arcg.is/1KyOK52 .

This project demonstrated the feasibility of using digital drone aerial photography and field survey software to characterize a pioneer cemetery in detail. With input from the Burley community we created an asset that can be used to maintain and enhance information about individuals interred at the site, and to add new records as the identity of others becomes available. The project complemented ongoing research at the cemetery, e.g., we had the opportunity to integrate data from a K9 search to identify potential sites of human remains.

Technical accuracy met expectations. By using a commercial GNSS receiver with real-time kinematic corrections, we achieved an average error rate under 10cm in all axes (X, Y & Z). This includes the coordinates for ground control points that were used to adjust the 2D orthomosaic images.

Based on historical photos we knew that the tree canopy had grown extensively since 1950. The height of the tallest tree inside the property line was 45m. This required that the drone be flown at an altitude of 50m. The trees obscured a good portion of the ground, making it impossible to achieve complete photographic coverage of the gravesites.

To overcome this challenge we evaluated the use of terrestrial Lidar scanning (TLS) equipment. Lidar generates 3D point clouds which often require significant post-processing (e.g., georeferencing and format conversion) before the output can be used in GIS systems. Due to time limitations we were not able to complete this task.

A future aerial survey conducted with a Lidar-equipped drone could be productive, as could ground level photo collection using a painters pole.

Burley Association leaders would like to collaborate with university students and staff to conduct additional research and enhance the database. Those who are interested can contact them directly at  burleyassociation@gmail.com .

A number of software packages were used during the project:

• Clip Raster (reduce aerial photo to area of interest)
• Export Features with selection (subset of parcel IDs)
• Classification Tools - Segmentation (using orthomosaic)
• Exploratory 3D Analysis (interactive elevation profile)

• Gravesite inventory and histogram

• dDAP mission planning and flight control

• 2D and 3D dDAP products - all available outputs

• Lidar scan registration and 3D point cloud creation

• Survey Form Design

Map content created using ArcGIS Pro:

Digital drone aerial photography processing (orthomosaics, digital surface models, digital terrain models and 3D web scene).

ArcGIS analysis workflow to identify probable locations of human remains.

Geographic Coordinate System: WGS 1984

Projected Coordinate System: WGS 1984 UTM Zone 10N

Buist, C. Bobby, and Heather Bedle. “Improving GPR Interpretation by Applying Image-Enhancing Attributes and Machine Learning Techniques: A Case Study over Green Hill Cemetery in Frankfort, Kentucky.” The Leading Edge 43, no. 7 (July 1, 2024): 422–35.  https://doi.org/10.1190/tle43070422.1 .

European Environment Agency. “Digital Terrain Model.” Term. Accessed August 23, 2024.  https://www.eea.europa.eu/help/glossary/eea-glossary/digital-terrain-model .

Lange, Jan-Michael, Martin Kaden, and Nadine Janetschke. “Petrographic Investigations in Cemeteries in Dresden: A Scientific View on the Tombstone Inventory Including Historic and Cultural Aspects.” Environmental Earth Sciences 77 (September 6, 2018).  https://doi.org/10.1007/s12665-018-7789-4 .

Spera, Stephanie A., Matthew S. Franklin, Elizabeth A. Zizzamia, and Ryan K. Smith. “Recovering a Black Cemetery: Automated Mapping of Hidden Gravesites Using an sUAV and GIS in East End Cemetery, Richmond, VA.” International Journal of Historical Archaeology 26, no. 4 (December 1, 2022): 1110–31.  https://doi.org/10.1007/s10761-021-00642-3 .

“Build Web Apps with No-Code or Low-Code | ArcGIS Experience Builder.” Accessed August 18, 2024.  https://www.esri.com/en-us/arcgis/products/arcgis-experience-builder/overview .

“Desktop GIS Software | Mapping Analytics | ArcGIS Pro.” Accessed August 18, 2024.  https://www.esri.com/en-us/arcgis/products/arcgis-pro/overview .

Dronelink. “Dronelink. Drone Flight Control for DJI, Autel Drones.” Accessed August 18, 2024.  https://www.dronelink.com .

“FARO ^®  SCENE 3D Point Cloud Software | FARO.” Accessed August 18, 2024.  https://www.faro.com/en/Products/Software/SCENE-Software .

Emlid. “Flow 360 Land Surveying Software.” Accessed August 18, 2024.  https://emlid.com/emlidflow360/ .

“GIS Drone Mapping | 2D & 3D Photogrammetry | ArcGIS Drone2Map.” Accessed August 18, 2024.  https://www.esri.com/en-us/arcgis/products/arcgis-drone2map/overview .

“Online Form Builder for Smart Forms & Surveys | ArcGIS Survey123.” Accessed August 18, 2024.  https://www.esri.com/en-us/arcgis/products/arcgis-survey123/overview .

DJI. “DJI Mavic 3 Enterprise Series - Industrial Grade Mapping Inspection Drones - DJI Enterprise.” Accessed August 18, 2024.  https://enterprise.dji.com/photo .

Emlid. “Reach RS3 RTK GNSS Receiver.” Accessed August 18, 2024.  https://emlid.com/reachrs3/ .

“FARO Focus Laser Scanner | FARO.” Accessed August 18, 2024.  https://www.faro.com/en/Products/Hardware/Focus-Laser-Scanners .

Farm Production and Conservation Business Center. “Geo Site.” Accessed August 18, 2024.  https://www.fpacbc.usda.gov/geo/index.html .

Earth Resources Observation And Science (EROS) Center. “National Agriculture Imagery Program (NAIP).” Tiff. U.S. Geological Survey, 2017.  https://doi.org/10.5066/F7QN651G .

“Real-Time GPS/GNSS Data | Data | GAGE.” Accessed August 18, 2024.  https://www.unavco.org/data/gps-gnss/real-time/real-time.html .

“The USGenWeb Tombstone Project - Washington State.” Accessed August 18, 2024.  http://www.usgwtombstones.org/washington/washing.html .

“Washington Lidar Portal.” Accessed August 18, 2024.  https://lidarportal.dnr.wa.gov/#47.40179:-122.63334:14 .

Washington State Department of Archaeology & Historic Preservation (DAHP). “Researching Washington’s Historic Cemeteries,” April 27, 2011.  https://dahp.wa.gov/archaeology/cemeteriesburial-sites/researching-washingtons-historic-cemeteries .

“Washington State Geospatial Open Data Portal.” Accessed August 18, 2024.  https://geo.wa.gov/datasets/96bd481859444ddd9c9f2a4d0d729570 .

Aqua Terra Cultural Resource Consultants. "Aqua Terra Cultural Resource Consultants." Accessed August 19, 2024.  http://aquaterracrc.com/ .

Burley Association. “Burley Association.” Accessed August 16, 2024.  https://www.burleyassociation.org .

Canine Forensics Foundation. “K9FF.Org |Home.” Accessed August 19, 2024.  https://k9ff.org/ .

Remote Sensing & Geospatial Analysis Laboratory. “Remote Sensing & Geospatial Analysis Laboratory.” Accessed August 23, 2024.  https://sites.uw.edu/rsgal/ .