Ian Helfrich / Selected spatial work
Investment geography, classified LiDAR and reconstructed terrain. Research and working models at the scale of a neighborhood, a region and a country.
Explore the 3D work National investment atlas
Econometrics · Network analysis · Point clouds · Terrain
Selected visual work
Ground, buildings and vegetation from USGS 3DEP returns.
Ground surfaces, contours and building context.
Rotate and inspect a land–seabed mesh with 525,474 triangles.
02 · Spatial investment
The national atlas reports where New Markets Tax Credit financing was observed. The Asheville scenario asks a separate, conditional question: how a fixed incentive budget changes road-network access to jobs once offers, additionality and relocation take stated values.
Observed New Markets Tax Credit projects across the United States, built from the public CDFI Fund FY2025 release updated August 2026. Dollar totals are qualified low-income community investments (QLICI).
Limit. Financing records show where investment was observed. They do not measure jobs or outcomes, or what would have happened without the credit.
A 3D scenario over Asheville terrain. It separates supported incentive offers from induced site employment and additional study-area jobs, then compares an equal split of the budget with an access-maximizing allocation across six existing employment clusters. Inputs are 2023 Census primary-job counts and a directed OpenStreetMap road network.
Limit. A conditional scenario, not a causal estimate. Behavioral parameters are illustrative; the model has no fiscal return, wage or general-equilibrium component.
Compares rural and urban investment within the same Community Development Entity, using 19,907 transactions across 8,024 projects, with intermediary fixed effects and an order-invariant decomposition.
Limit. The paper uses an earlier data vintage than the national atlas and its estimates are not restated with the 2023 data. Its results are descriptive and noncausal.
03 · LiDAR and terrain
The portfolio holds eight selected artifacts, including academic exercises and group work that are credited as such. The terrain diagnostic and interactive mesh use different source measurements.
An independently processed ground-only surface was used to check a reported raster streak. The streak was absent from the comparison hillshade, which narrowed the diagnosis. Refining the grid could not recover ground returns that were not collected: missing cells rose as cells shrank.
Limit. Terrain diagnostics only. Rainfall, infiltration, discharge and calibrated flood depth are outside the analysis.
An original triangulated surface joining land elevation and seabed bathymetry, loadable in the browser for rotation, wireframe and vertex inspection. Sources are GMRT bathymetry (CC BY 4.0) and Mapzen terrain tiles built from SRTM, GMTED2010 and ETOPO1.
Limit. The mesh vertices are gridded elevations, not LiDAR returns, photogrammetry or navigation data. A 100 m grid does not establish measurement accuracy.
04 · Applications
Urban property evidence, ocean dynamics and county-level spatial statistics.
Application
City and County geometry, official City property evidence and rental scenarios in a 3D city model. Scenario outputs remain separate from recorded parcel data.
Ocean data · Live
Depth-resolved HYCOM and NOAA current fields, Sentinel-2 water imagery and a 3D tank model with published numerical checks. The tank is a qualitative laminar model, not laboratory-calibrated.
Spatial statistics · Live
County-level spatial statistics, regression diagnostics, empirical-Bayes shrinkage and a California redlining case study. Associations remain distinct from causal effects.
05 · More work
Spatial analysis & technical instruction
Tell me what you need to understand, visualize or build, and which data you have.