Contact
Hi! My name is Tiffany and I’m a geospatial analyst who is passionate about climate resilience and data for social good. I am also interested in exploring remote sensing for biodiversity monitoring, urban analytics for smart cities, and how the realms of natural resource management and sustainable agriculture intersect in the context of community food systems. I grew up in coastal Los Angeles and am now based in New England.
Broadly, I am interested in using geospatial data science and environmental modeling to address planetary challenges. As an analyst on the R&D team at Blue Marble Geographics, a geospatial software company in Maine, I have focused on data processing for ML/AI projects. This includes a deep learning land cover classification initiative for Global Mapper, one of our flagship GIS products, and can be found in our new Insight and Learning Engine. At Blue Marble, my current responsibilities include working on a custom PyTorch build with expanded CUDA architecture support, transforming objects in pixel space into georeferenced coordinates for model validation with high-resolution imagery, and gaining familiarity with C++ builds and debugging. My role also involves automating image processing, feature extraction and geometric operations in Python and leveraging tools such as Google Earth Engine to analyze satellite data. Additionally, I conduct internal research on 3D and point cloud modeling.
Before graduate school, I was the lab manager and a research assistant at the Global Policy Lab, a research group using remote sensing, data science and econometrics to study climate impacts at UC Berkeley’s Goldman School of Public Policy. The lab’s work expanded to public health matters, and I was a member of the team that wrote a Nature paper (Hsiang et al., 2020) which analyzed how government non-pharmaceutical interventions in six countries affected the spread of COVID-19. In addition, I have worked with the U.S. EPA, NRDC, Stockholm Environment Institute, Metropolitan Area Planning Council and Arup on sustainability, data analysis and climate strategy.
I earned my master’s in Environmental Planning with a concentration in spatial data analytics from Tufts University’s Department of Urban and Environmental Policy + Planning (UEP) and my bachelor’s in Animal Science with a minor in environmental sustainability from Cornell University. My graduate thesis was on the predictive modeling of eelgrass in coastal Maine.
Below is some recent work from my portfolio. This site is best viewed on desktop. Thanks for visiting!
Languages (Programming):
- Python (Proficient)
- Powershell (Intermediate)
- R (Intermediate)
- C++ (Learning)
Python Geospatial/ML Libraries:
- GDAL/OGR
- GeoPandas
- Rasterio
- Fiona
- Shapely
- PyTorch
- Torchvision
Frameworks and Tools:
- CMake
- Docker
- Git
- Jenkins
- OpenMVG/OpenMVS
- Visual Studio
- VSCode
Languages (World):
- English (native)
- Mandarin (native)
- Taiwanese (fluent)
- Spanish (proficient)
Education:
- Cornell University, BS, 2018
- Tufts University, MS, 2024
Project 1
Graduate Thesis
2024
Eelgrass (Zostera marina) is a critical component of coastal ecosystems, providing essential services such as habitat for marine species, nutrient cycling, and shoreline stabilization. Understanding the spatial and temporal distribution of eelgrass is vital for conservation and management efforts, particularly in the face of environmental changes. This study employs spatiotemporal modeling techniques to predict the distribution of eelgrass habitats in Casco Bay, Maine. Using MaxEnt, a species distribution modeling tool, I incorporate variables such as temperature, salinity, dissolved oxygen, total nitrogen, turbidity, and bathymetry to generate a presence-only predictive model of the region.
Spatial statistics, including Local Moran's I and Emerging Hotspot Analysis (EHSA), are utilized to identify and analyze patterns of eelgrass presence and distribution across different temporal periods. This research contributes to the understanding of eelgrass ecology and provides a framework for predicting presence under varying environmental conditions. The findings can inform conservation strategies and management practices aimed at preserving and restoring eelgrass habitats in coastal Maine, ensuring their resilience in the face of climate change and anthropogenic pressures.
Thesis: Link
Slide deck: (upon request)
This independent project is ongoing in collaboration with the Maine Department of Environmental Protection and Casco Bay Estuary Partnership.
Project 2
Agrivoltaics GIS Suitability Analysis
2022
In the realm of renewables, solar energy is rapidly expanding, with more than 2,500 utility-scale projects in the United States alone today. The utility of these large scale operations can be increased by having livestock graze on the land, allowing it to serve dual purposes. While Washington State has been a vanguard in the decarbonization sector, the majority of its efforts have been focused on hydroelectric power thus far. Growth in solar energy and indeed, solar grazing could be a novel way to strengthen the state’s renewable energy efforts while providing livestock with healthy pasture.
The suitability analysis was conducted by reclassifying each layer and utilizing the raster calculator. The slope layer was mosaicked from USGS DEM tiles. When these four input layers were combined to form the final suitability analysis, it was clear that the flatter parts of Washington east of the Cascade Mountains, particularly in the Columbia Basin and Palouse region near the Idaho border, were optimal for solar grazing. In fact, this area is currently an agricultural powerhouse for the state and produces wheat, barley, and legumes, among other crops. They received lots of sunlight, were close to transmission lines, and contained appropriate vegetation. The results of this study demonstrate how solar grazing or agrivoltaic projects in this region – or converting existing operations to accommodate such initiatives – are a viable and potentially useful way for farmers to promote sustainability while generating additional revenue, without the need to fundamentally restructure present land use. These findings could also be useful for policymakers and utility companies in Washington as they find ways to adapt to rising energy demands and research renewable energy alternatives to hydroelectric or geothermal power.
Data sources: NatureServe, NREL, Oak Ridge National Lab, USGS, U.S. Census Bureau
Project 3
Spatiotemporal AQ Analysis
2022
Air pollution in today’s globalized, industrialized society poses a major environmental health risk, and millions of premature deaths from respiratory disease or heart attacks can be attributed to ambient pollution. This problem is particularly acute in urbanized areas. As part of the effort to identify and combat air pollution, air quality monitors have been deployed by governments to measure pollutants with high levels of accuracy. Advances in technology suggest that low-cost mobile monitors can help augment existing fixed monitors in understanding air quality levels on a more granular level, such as in a neighborhood or street by street. Wind speed can also impact the spread of pollutants, with suspended particulates responsible for smog conditions observed in London and Los Angeles.
Using existing data from a Tufts University study (Jiang, et al. 2020) on urban traffic and air quality, this project analyzed how atmospheric conditions have impacted levels of ultrafine particles (UFP), black carbon (BC) and nitric oxide (NO) through the binning of wind speeds and wind directions and comparing the mean concentration of the three pollutant types under each condition. Statistical data was analyzed through the Seaborn and Plotly libraries for visualization, and k-means clustering was done to compare unsupervised learning with categorization from the Beaufort Scale for wind speeds. Spatial data was saved as a GeoDataframe and uploaded to kepler.gl.
Data sources: Tufts Air Pollution Lab (TAPL)
Project 4
work in progress
2022, 2024
In the coastal plains of the Arctic, climate change impacts on ecosystems are increasingly visible. The Greater White-Fronted Goose (Anser albifrons) is a circumpolar species of goose that spends its spring breeding season in tundras across North America, Greenland and Russia. Because the white-fronted goose migrates to wetlands along the Gulf Coast during the winter, their distribution is broad and there are various factors that go into their survival and overall productivity. Using one such study with USGS data, I analyzed whether certain trends in these greater white-fronted geese in Alaska, such as weight and wing molt, were observed among individuals.
The primary findings of the Stata analysis indicated that the change in mean weights of greater white-fronted geese were statistically significant and increased each year. Assuming that the weights of adults served as a proxy for their overall health, this suggested the population was doing well, confirming the USGS Alaska Science Center’s observation that they were growing in population despite a changing climate.
As a follow-up to the original project, I plan on analyzing spatial characteristics of the Arctic Coastal Plain using open-source Python libraries and Global Mapper with a touch of satellite image processing. These steps are intended to better understand the topography of the region and availability of food sources.