Griffin Hale
← All projects

Spatial Analysis

Case study

Three GIS studies of flood mapping, water quality, and electric-grid reliability, including two results that rejected the hypotheses.

GISSpatial AnalysisRemote SensingPythonarcpy

At a glance

  • Outcome: Completes three studies; two rejected hypotheses.
  • Status: Completed case studies.
  • Role: Solo: framed questions; gathered, scraped, analyzed, and mapped data.
  • Stack & libraries: Uses ArcGIS Pro, Python (arcpy, pandas), ESA SNAP.
  • Scale: Uses multi-scene Sentinel-1 SAR mosaics.
  • Validation: Reports null results and independent feeder flags.
  • Limitations: Approximates unpublished Pacific Power feeder boundaries with circuit-linework Thiessen polygons; these areas are not ground truth.

SAR flood mapping

Compared flood-insurance FEMA maps with a real high-water event in Coquille River valley, Coos County, OR. Processed Sentinel-1 SAR in ESA SNAP: orbit correction, radiometric calibration, terrain flattening, and correction for the valley’s hills. Classified water via ESRI’s pretrained Water Body Extraction model; referenced ALOS-2 PALSAR and Landsat 9.

Found FEMA zones closely tracked flood-weekend extent across the long-recognized, well-mapped valley-floor floodplain.

Water quality and school performance

Tested Oregon drinking-water quality against standardized test scores. Combined Oregon Department of Education results; scraped, hand-curated water-system demerits; and Census density and income. Mapped the three public sources by region as a bivariate choropleth; water data required scraping before use.

Found a weak direct relationship. Density and income fit better; eastern Oregon schools posted similar scores despite large resource differences.

Electrification and grid reliability

Tested whether Portland EV adoption, heat-pump and electrical permits, and densification track poor reliability. Frequent short North Portland outages during charger and multi-unit construction motivated the hypothesis; a previous deeper Southeast residence had rare outages.

Built an eight-step ArcGIS Pro/pandas pipeline from ~80,000 geocoded permits, tract-level EV registrations, PGE SAIDI/SAIFI Oregon PUC filings, feeder polygons, Pacific Power linework, and ACS estimates. Area-weighted tract data onto feeders, independently scored transition pressure and grid stress, then ranked only areas high in both.

Rejected the system-wide correlation hypothesis. Identified ten engineering-review overlap hotspots, including Sylvan-Barnes feeder and Vernon substation. Four of five PGE picks were already independently flagged as worst-performing circuits. The repeatable method targets overlap despite the null.

Maps & figures

Poster titled SAR Floodplain Modeling in the Coquille River Basin, showing the study area, a flood-event hydrograph, the processing workflow, and SAR-derived water extent compared against Landsat surface water and FEMA flood-hazard zones.
Floodplain study: Sentinel-1 SAR processed in SNAP, open water extracted with ESRI's pretrained deep-learning model, compared against Landsat surface water and FEMA flood-hazard zones for the Coquille River basin.
Map of Portland-area electric distribution geography: PGE feeder polygons across Multnomah County with the constructed Pacific Power pseudo-circuit area highlighted.
Electrification study, study area: PGE feeder polygons across Multnomah County and the constructed Pacific Power pseudo-circuit service areas.
Choropleth map of PGE feeders shaded by priority score, with the five final candidate feeders C06 through C10 labeled.
PGE feeders shaded by priority score, with the five final candidate feeders (C06 to C10) labeled.
Map of the ten final candidate service areas, five PGE in blue and five Pacific Power in orange, across Multnomah County.
The ten candidate service areas where electrification pressure and grid stress overlap: five PGE (blue) and five Pacific Power (orange).
Scatterplot of transition pressure versus grid stress for 548 service areas, with final candidates marked and geometric-mean iso-priority curves overlaid.
Transition pressure and grid stress are essentially uncorrelated across 548 areas (Spearman rho around 0.05). Iso-priority curves identify areas elevated on both.