I’m Steven Aarons, an Environmental Studies major and Spatial Data Science certificate student at CUNY Hunter College passionate about
using geospatial analytics, programming, and environmental modeling to solve real-world problems. My work bridges science and technology —
from developing automated air quality and noise assessment geoprocessing tools for NYC Government to building deep learning models that
improve precipitation prediction to mitigate extreme weather events. Skilled in Python, SQL, GIS, remote sensing, and data visualization,
I enjoy transforming complex environmental datasets into clear, actionable insights. Beyond coding, I’m drawn to urban sustainability,
landscape design, and mapping how cities interact with the environment. Driven by a love for efficiency: I build tools that make analysis
faster, research smarter, and data more meaningful.
Won first place in the Environmental Science Section at the Emerging Researchers National conference.
Presented research through posters at the American Meteorological Society, American Geophysical Union, City Tech Summer Symposium conferences.
Developed a data-driven precipitation prediction model using a dual U-net architecture with TensorFlow Keras on Google Collab, leveraging RTMA data.
Achieved a Receiver Operating Characteristic Area Under the Curve score of 0.92 for a binary rain/no-rain classification model with a 12-hour lead time.
Gained an deep understanding of Numerical Weather Prediction models and deep learning architectures for image processing.
Automation
ArcPy + ArcGIS Pro (Department of City Planning)
ArcPy, Pandas, Xlsxwriter, Tkinter
Singlehandedly automated multiple complex Air Quality HVAC screening workflows in ArcGIS, reducing week-long analyses to under 30 minutes through custom Python geoprocessing tools, increasing efficency by over 90%.
Built Construction Noise Analysis tools that calculate noise standards from multiple CSVs and automatically export formatted Excel reports with labeled and symbolized map layers.
Ensured longevity of tools through comprehensive ReadMe files, sample table and shapefile inputs, testing, and error handling for incorrect projections or missing values.
Developed clear guidelines to help analysts explain complex construction noise methodology, minimizing consultant follow-up.
Produced an objective assessment of conflict-zone damage, avoiding bias and data gaps in official military sources.
Developed a Damage Proxy Map (DPM) using Sentinel-1 SAR InSAR to detect infrastructure damage from coherence loss, effective even for roof-intact collapses not visible in optical imagery.
Applied advanced remote sensing methods, including Otsu’s thresholding, to isolate conflict-related damage from natural coherence changes.
Performed time-series analysis to trace damage progression and pinpoint peak destruction periods, correlating findings with ground invasion events.
Validated the model against Geoconfirmed.org and UNOSAT (04/20/24), achieving 61.41% and 100% validity, respectively.
R
Tree Growing Season Length vs Land Cover Type and Impervious Surface Percentage
LandsatTS, leaflet, sf, tidyverse, ggplot2
Investigated how urbanization affects tree phenology in New York City using 2021 Landsat imagery, National Land Cover Database, and impervious surface data.
Integrated spatial and temporal analyses in R and Google Earth Engine, linking impervious surface percentages to phenological metrics for nuanced insights on urban heat island effects.
Processed and cleaned multi-sensor remote sensing datasets, calculating NDVI for tens of thousands of tree canopy points while filtering clouds, snow, and water.
Modeled phenological curves with cubic splines, GAMs, and linear trends to estimate start/end of season (SOS/EOS) and growing season length (GSL) for individual trees.
Identified statistically significant differences in phenology between urban, park, and rural trees using ANOVA, revealing longer growing seasons in urban areas despite lower peak greenness.
Developed high-quality visualizations of NDVI trends, phenological curves, and relationships between urbanization and growing season length with ggplot2.
Ensured robust statistical comparisons by creating balanced datasets across regions, providing reliable conclusions on the impact of land cover on tree growth cycles.
ArcGIS Pro
Interpolation Analysis Research Poster
Cartographic Work
Expert Proficiency in ArcGIS Pro and Core GIS: Demonstrated expertise in ArcGIS Pro through formal academic training and work and research experience.
