Graduate Research Coffee Hour
- Ruth Buck, “Making Compactness Matter: Evaluating a Multiscalar, Population-Based Compactness Metric in Pennsylvania”
Abstract:
District shapes and compactness have historically been a major signal, both to lawmakers and the public, that some kind of malfeasance, or gerrymandering, has occurred. We argue, however, that much of the discussion of compactness in policy and academic spaces neglects a clear theory of why compactness matters for fair representation, choosing to take a purely geometric understanding of compactness as sufficient. In this paper, we extend DeFord et al.'s (2021) Partisan Dislocation metric to measure the compactness of districts not based on their geometry but how they separate voters into different districts at different spatial scales. We conceptualize the harms of noncompact districts as politically isolating individuals from their neighbors, making activities like canvassing more difficult and decreasing awareness of which districts voters belong to. We develop a multiscalar, population-based measure of compactness using geocoded voterfiles. We evaluate this new metric in Pennsylvania, examining how compactness is operationalized in a collection of 15 redistricting proposals submitted to the PA Supreme Court in 2022.
- Carolina Carrión-Klier, “What a Machine Learned about the Galápagos Islands”
Abstract
To help address the threats posed by invasive species in the Galápagos Islands, we conducted an expedition to the uninhabited island of Santiago during the summer of 2023. In this presentation, I will demonstrate how we used machine learning techniques on remote sensing data to create maps of an invasive species. These maps played a critical role in shaping our excursion protocol and continue to provide valuable information for the development of effective management strategies. Furthermore, I will share some of the insights gained from our expedition and how the acquired data will enhance our understanding of invasive species' spread dynamics and enable the production of more accurate maps.
- Emma Cheriegate, “Investigating the Development of Ghost Forests Due to Saltwater Intrusion along the Savannah River, Georgia Coastline of the United States”
Abstract:
Shallow aquifers along the southeastern US are experiencing saltwater intrusion from rising sea levels, changes in tidal cycles, and groundwater pumping, which are leading to higher soil salinity. Ghost forests, or areas where coastal forests have deteriorated due to salt water, are expanding in the Southeast US. NASA DEVELOP partnered with the USGS, USDA, and Georgia Southern University to investigate saltwater intrusion effects on coastal forests in the lower Savannah River using NASA Earth observation data spanning 2013 to 2023. Our multi-sensor approach used Landsat 7 Enhanced Thematic Mapper Plus (ETM+), Landsat 8 Operational Land Imager (OLI), and Planet Lab’s Scope Rapid Eye and Dove. We aimed to determine the sensitivity of detecting coastal forest health decline proxied by creating a supervised land cover classification, the normalized difference vegetation index (NDVI), and directly link remote-sensing based time-series to in-situ porewater salinity trends throughout the extent of the Savannah River.
Our project ran at both a regional and site-level scale (4 USGS-monitored sites). We found differences in site level NDVI values over 2013-2023 from both Landsat and Planet sensors. At the three sites nearest to the coast, we observed a muted seasonal variation that exhibited an inverse relationship with the increasing levels of river and porewater salinity found in those locations. The results provided here add to the growing body of research seeking to understand saltwater effects on coastal forests using spaceborne remote sensing and emphasize the need for proactive measures to mitigate saltwater intrusion's effects on coastal ecosystems.
- Huan Ning, “Autonomous GIS: the next-generation AI-powered GIS”
Abstract:
Large Language Models (LLMs), such as ChatGPT, demonstrate a strong understanding of human natural language and have been explored and applied in various fields, including reasoning, creative writing, code generation, translation, and information retrieval. By adopting LLM as the reasoning core, we introduce Autonomous GIS as an AI-powered geographic information system (GIS) that leverages the LLM's general abilities in natural language understanding, reasoning, and coding for addressing spatial problems with automatic spatial data collection, analysis, and visualization. We envision that autonomous GIS will need to achieve five autonomous goals: self-generating, self-organizing, self-verifying, self-executing, and self-growing. We developed a prototype system called LLM-Geo using the GPT-4 API in a Python environment, demonstrating what an autonomous GIS looks like and how it delivers expected results without human intervention using three case studies. For all case studies, LLM-Geo was able to return accurate results, including aggregated numbers, graphs, and maps, significantly reducing manual operation time. Although still in its infancy and lacking several important modules such as logging and code testing, LLM-Geo demonstrates a potential path toward the next-generation AI-powered GIS. We advocate for the GIScience community to dedicate more effort to the research and development of autonomous GIS, making spatial analysis easier, faster, and more accessible to a broader audience.
- Shiyan Zhang, “Developing high-resolution PM2.5 exposure models by integrating low-cost sensors, automated machine learning, and big human mobility data”
Abstract:
District shapes and compactness have historically been a major signal, both to lawmakers and the public, that some kind of malfeasance, or gerrymandering, has occurred. We argue, however, that much of the discussion of compactness in policy and academic spaces neglects a clear theory of why compactness matters for fair representation, choosing to take a purely geometric understanding of compactness as sufficient. In this paper, we extend DeFord et al.'s (2021) Partisan Dislocation metric to measure the compactness of districts not based on their geometry but how they separate voters into different districts at different spatial scales. We conceptualize the harms of noncompact districts as politically isolating individuals from their neighbors, making activities like canvassing more difficult and decreasing awareness of which districts voters belong to. We develop a multiscalar, population-based measure of compactness using geocoded voterfiles. We evaluate this new metric in Pennsylvania, examining how compactness is operationalized in a collection of 15 redistricting proposals submitted to the PA Supreme Court in 2022.