Authors: Sarah E. Walters*, Oak Ridge National Laboratory, Lauryn N. Bingham Bragg, Oak Ridge National Laboratory, Zachary T. Palmer, Oak Ridge National Laboratory, Rohan Dhamdhere, Center for Computational Imaging and Personalized Diagnostics, Marie L. Urban, Oak Ridge National Laboratory, Dalton Lunga, Oak Ridge National Laboratory
Topics: Population Geography, Cultural Geography, Remote Sensing
Keywords: ORNL, PDT, Mixed-Methods, Population Modeling, Cemeteries, Graves, Burials, Building Occupancy, Culture, Qualitative, Quantitative, Automation
Session Type: Virtual Paper
Start / End Time: 8:00 AM / 9:15 AM
Room: Virtual 19
Presentation File: No File Uploaded
The objective of the Population Density Tables (PDT) project at Oak Ridge National Laboratory (ORNL) is to report high-resolution building occupancy (people/1000 ft2) at both national and subnational levels, globally. PDT does so through the creation of observation models that capture facility-specific use-dynamics. These models - built on a Bayesian framework - serve to update and refine preexisting baseline estimates to reflect patterns of use more accurately. While PDT is ever-expanding, not all facility-types currently are (or can be) equally reported due to a dearth of available open-source data. Cemeteries are one such category and present specific challenges. A strength of PDT is the ability to incorporate qualitative data – to include a growing repository of culture-specific funerary practices. However, patterns of burial and rituals related to the disposition of the dead are driven by aspects (e.g., cultural belief, religion, geography) that, however informative, may not fit within existing observation models. To ensure that these models accurately reflect real-world cemetery use, they must be consistently updated and expanded to better exploit available data. Additionally, grave counts are a fundamental proxy for estimating cemetery use. However, capturing these counts has proven problematic since cemeteries often contain large numbers of burials and the manual quantification effort required is resource intensive, making identification and subsequent updates challenging. Results of addressing both challenges will be presented, including the development of an automated feature counting tool that, within the confines of the designated test area, has detected and quantified graves – with little human oversight.