Data Dives are conversations with Columbia Climate School researchers to learn more about their work and explore trending topics through the lens of data science and visualization.
The Scientist
Carolynne Hultquist is a postdoctoral research scientist at the Center for International Earth Science Information Network (CIESIN) at the Columbia Climate School. Hultquist specializes in developing computational methods for fusing and validating spatial data sources to better understand complex environments, especially vulnerability in relation to hazards. She holds a Ph.D. in Geography and Social Data Analytics from the Pennsylvania State University.
The Data
Carolynne developed the data for the map below, which builds on a CDC index of vulnerability factors—such as poverty and not having access to a vehicle—that indicate increased sensitivity to hazards.
Interactive Map: U.S. Social Vulnerability Index Grids, 2018
» Learn more / View large version of map | Data & documentation
The Dive
Your research is focused on “fusing sources of geographic information to better understand complex environments.” What kinds of questions motivate your research? What topics or issues do you hope to better understand?
I’m a geographer who integrates spatial data from physical and human systems to help decision making during disasters. Rooted in the geographic information sciences, my research uses observations from satellite, aircraft, and drone remote sensing, physical models, and high-resolution population estimates. I’m very interested in exploring the viability of volunteered geographic information that ordinary people contribute from mobile devices during disasters, for monitoring flooding, radiation, and air quality.
For my postdoctoral research at CIESIN, I’m developing a flood risk map in collaboration with Andrew Kruczkiewicz, who is leading the NASA Geo Flash Flood project. The map will indicate not just where flooding occurs, but the severity, exposure, and vulnerability of populations.
Why can it be important to include a social component in a scientific research project? How can a hybrid approach help us better understand complex issues?
Modeling the physical aspect of hazards is only part of the picture. Disasters by their nature involve exposure. This means we need to consider data that characterizes the populations impacted by hazards.
Geography has a rich tradition of studying both physical and human processes, separately, but also by examining interactions between humans and the physical environment, which has evolved into a core sub-discipline of the field. Studying these interactions is critical to facing global problems, from climate change to food security, biodiversity, environmental justice, and disasters. While many approaches may be applied to studying these topics, I specialize in using data science techniques at the intersection of human-environment interactions.
Can you give an example of how social vulnerability data has informed scientific insight and public policy?
Social vulnerability data has been used before and during hazards to inform preparation and response in areas of need. For example, Colleen Neely worked with Andrew Kruczkiewicz and me on a case study in Houston. Stakeholder Iris Gonzalez, director of the Coalition for Environment Equity & Resilience said, “Everyone in Harris County deserves flood protection, but we don’t have any tools in place to be able to understand where those worst hit areas are, and where those pockets of highest vulnerability are.” So we developed a flood vulnerability index that captures social factors influencing flood vulnerability, at the highest spatial resolution, for the Greater Houston area of Harris County.
In evaluating our index we considered one community more closely: Kashmere Gardens, a Black and low-wealth community in Houston. During 2017’s Hurricane Harvey, the people in this area placed among the highest number of high-water rescue calls—a potential indicator of flood vulnerability. Our flood vulnerability map corresponds by showing the community also ranks among the highest in relative vulnerability within Greater Houston. [See “Understanding Flood Vulnerability: A Case Study of Greater Houston.”]
Ultimately, we need to ensure equity and justice in the policies, programs, and decision-making processes that determine which communities get invested in—made more flood-resilient—and which do not. Using high resolution data and engaging community stakeholders to select vulnerability indicators representative of conditions within a community can help advance social justice.
What jumps to your mind when you look at the SVI map above? Do you notice any high-level trends?
The Centers for Disease Control and Prevention’s (CDC) Social Vulnerability Index (SVI) has long been a key resource for identifying communities that may need support before, during, or after hazardous events or disease outbreaks. Our SVI increases the utility of the CDC’s by gridding the input data and removing uninhabited areas. This gives us more accurate population estimates, makes it easier to calculate the social vulnerability for user-defined areas, and facilitates integration with hazard and other geospatial data.
The SVI maps show that broad swaths of the US have high levels of social vulnerability, with the southeast in particular facing many challenges. I grew up in the Carolinas, and I have seen vulnerable populations inland struggle for years to recover from hurricane damage. Sometimes people aren’t able to get on their feet before the next storm hits. As you zoom in closer on the map, you’ll see throughout the country populations with high vulnerability, including pockets within most urban areas.
Like the CDC’s Index, our SVI was produced with four sub-themes—Socioeconomic, Household Composition and Disability, Minority Status and Language, and Housing Type and Transportation—in addition to the overall vulnerability value. Comparing the sub-themes may reveal underlying contributions to vulnerability in different parts of the country and can inform a wide scope of decisions, from recognizing the need to produce early warnings in non-English versions to providing inexpensive evacuation options in areas of socioeconomic deprivation.
What do you see on the horizon in your field? Do you think any particular technologies or techniques could yield advances in the future?
My field of spatial data science is expanding as big social data analytics takes off due to the increased availability of data from a multitude of sources. The focus on GeoAI—geospatial artificial intelligence—has potential as well as pitfalls as we evolve methods that can be interpreted and generalized. More work is needed to evaluate statistical and visualized output, prove the effectiveness of the model in other areas, and ensure reproducibility.
In the midst of these developments is a conversation on ethics; there are privacy reasons that high resolution data on social factors are not provided by the government. Data scientists should take measures to ensure the data we produce is guided by ethics, not just technological advancements, to ensure that it is not misused.
Be sure to leave your questions and ideas for future Data Dives in the comments section below! ↓
Informative and educational. Thank you.
Good read. Currently working on a GIS degree.