Geospatial Research and Tech

On the last day of the Geospatial Gateway Forum, attendees heard from several academic and government groups that showcased the latest in geospatial technology and research innovation. Their ideas ranged from distinguishing the effects of geographic mobility and unemployment during the coronavirus pandemic amid rural and urban counties to using computer-generated point cloud data to train deep learning algorithms for scene segmentation.

These are just some of the presentations during the technology showcase:

Geospatial Repository and Data Management System (GRiD) Demo

The Geospatial Repository and Data Management System (GRiD) is the National Geospatial-Intelligence Agency’s (NGA) identified authoritative dissemination platform for elevation and 3D content throughout the community.

“NGA teamed up with the U.S. Army Corps of Engineers’ Cold Regions Research and Engineering Laboratory in order to come up with a better solution. From this partnership, GRiD was born,” said Jon Darveau, NGA. “GRiD has deployed its capability into the various AWS cloud-based architectures, allowing the system to take advantage of cloud technologies, to provide for a more stable and responsive system for its user base.”

GRiD is different from other IT systems deployed across the various networks. Primarily, GRiD is a U.S. government program. There is no big contractor behind GRiD. The intellectual property, and continued management and direction of the program, is owned by the government. GRiD utilizes a responsive and open development process in order to take feedback solicited from users and integrate it into future releases of the software.

Latino Neighborhoods and Home Ownership in Los Angeles County

Disparities are often explained through the lens of housing discrimination. In particular, residential security maps created in the 1930s are often used to show the legacy of systemic racism in the U.S. housing market. This study examined factors that contribute to disparities in Latino homeownership rates across neighborhoods in Los Angeles County.

Pedro Ruiz, Ph.D. student at St. Louis University (SLU), analyzed maps of LA County and various datasets, such as numerous census reports, to better understand differences in neighborhood-level Latino homeownership rates and examined their implications.

According to Ruiz’s findings, historical redlining was only associated with modest increases in neighborhood-level Latino homeownership rates. Poverty and education play increasingly important roles in Latino homeownership rates, and high racial diversity is associated with modest decreases in Latino homeownership rates. Immigrant characteristics also play an important role in Latino homeownership rates.

“What policy implications does this have? First, historical redlining plays a relatively small role in Latino homeownership rates, even though it receives a lot of attention in the public. Second, poverty and barriers to citizenship have to be addressed if you want to increase the Latino homeownership rate,” Ruiz said.

Those are some of the most important factors that influence homeownership rate, but more research needs to be conducted to understand why racial diversity and privileged neighborhoods play a small or insignificant role in Latino homeownership rates, he added.

LiDAR Classification Using Synthetic Data

This demonstration covered how NGA uses computer-generated point cloud data to train deep learning algorithms for scene segmentation. “Using computer-generated data on point clouds for training is not something new,” said Steven Spiegel, visiting scientist, NGA.

The data set that Spiegel and his team used for training is called Synth City globally registered point cloud that was developed in a program called Blender. Spiegel and his team trained a convoluted neural network on a computer-generated urban scene. Additionally, they added varying levels of noise to the dataset to simulate sensor noise found in LiDAR scanners. Then they tested the trained algorithm on an urban collection in Lille, France.

Distinguishing the Effects of Geographic Mobility and Unemployment on the Coronavirus Pandemic Amid Rural and Urban Counties

Steve Scroggins, Ph.D. student, SLU College for Public Health and Social Justice, presented a study he completed that reframed how to think about geography and employment when they become the forefront of the pandemic. A fraction of the more than 30 million jobs lost have been regained, and most business closures and position loss are permanent, unable to withstand this economic shock.

“While a global pandemic might not have been unexpected, the scale and effect it had on our economies certainly were from the large national financial systems to individual households,” Scroggins said. “From our model, we determined some profound ecological differences between rural and urban counties, and how each reacted to the unemployment surge during the pandemic.”

According to Scroggins’ results, in rural counties, as unemployment increased, COVID-19 decreased; as the percent of individuals staying at home increased, so did COVID-19 cases. But in urban counties, the opposite occurred. In urban counties, as unemployment increased, so did COVID-19 cases; when the percentage of individuals staying home increased, COVID-19 cases decreased.

To understand these results, Scroggins looked at the contextual difference between rural and urban areas. For example, in many rural areas, large families congregated into one home. When one person does not leave the home, it does not mean that another doesn’t risk exposure to others.

“We have to put into context what we know about geographical differences and how they respond to unemployment through available means,” said Scroggins. “Through integrating geography mobility and these innovative big data collection methods, we’re able to get a clear picture of how individuals are reacting within communities that are changing due to COVID-19 and the associated economic shock.”