Landslide Exposure Database (LED)

Landslide Exposure Database Data and code for the study: "Landslide Exposure in the United States." Overview: This repository contains the analysis pipeline and code used to develop the Landslide Exposure Database (LED). The LED is an integrated, high-resolution building inventory designed to characterize the physical and social dimensions of landslide exposure across the conterminous United States, Alaska, and Hawaii. Files included: data.zip US_LandslideExposureDatabase_points.gpkg -- Dataset containing 128 million building points US_CensusTractsMaster.gpkg -- Dataset containing 84 thousand Census Tracts polygons Respective README files for attribute descriptions. folder: z_supporting_data HurricaneHelene_Landslides_Complete.gpkg -- Data for the Supporting Dataset S1 Seattle_knwonSlideAreas_assessed_Complete.gpkg -- Data for the Supporting Dataset S2 code.zip 0_input: Data Acquisition This directory contains the scripts responsible for fetching the foundational building, hazard, and demographic datasets. Script 1: Download_building_data.ipynbFetches raw building footprint geometries from Overture Maps and retrieves structural and occupancy attributes from the National Structure Inventory (NSI). Script 2: Download_census_data.ipynbDownloads foundational geographic boundaries, including: Physiographic Provinces, Physiographic Divisions, US States, Census Tracts, and Census Blocks. Fetches socioeconomic and demographic datasets: CRE: Community Resilience Estimates; ACS: American Community Survey (Demographics, Poverty, Income); FFIEC: Federal Financial Institutions Examination Council (FFIEC) income level data. Script 3: Download_USGS_landslideSusc_n10.ipynbRetrieves the USGS N10 Landslide Susceptibility Model. 1_process: Data Processing and IntegrationThis directory contains the core analytical engine of the project, which merges spatial data, applies machine-learning filters, and aggregates metrics. The scripts follow a sequential order, with the workflow progressing from a to e. Script 4: a_LandslideExposureDatabase_part1.ipynbPart 1: Building Inventory Integration. Includes: Spatial Comparison: Analysis to perform spatial join and nearest-neighbor; Geometry & Proximity Filters: Cleaning and aligning raw footprint data; Machine Learning: Implementation of the Decision Tree classifier to distinguish primary structures (e.g., houses) from secondary outbuildings (e.g., detached garages). Landslide Susceptibility Attribution: Intersecting building footprints with hazard rasters. Script 5: b_LandslideExposureDatabase_part2.ipynbPart 2: Geographic Classification. Includes: Spatially joining buildings to their respective Physiographic Divisions, Census Divisions, States, Census Tracts, and Census Blocks. Script 6: c_LandslideExposureDatabase_part2.ipynbPart 3: Socioeconomic Attributes. Includes: Assigning population estimates to buildings (daytime/nighttime); Incorporating error metrics, Community Resilience Estimates (CRE), and poverty/income data at the structural level. Script 7: d_CensusTractsMaster.ipynbBuilding Aggregation: Rolling up structure-level landslide exposure metrics to the Census Tract level, Socioeconomic aggregation: aggregating CRE, poverty, and income data to Census Tracts. Script 8: e_LISA_cluster_LED-CRE.ipynbLISA Clusters: Calculating Local Indicators of Spatial Association (LISA) to identify statistically significant clusters of exposed and vulnerable populations. 2_output: Analytics and Visualization This directory contains the scripts used to generate the final analytical figures, maps, and plots for publication. Script 9: Results_figures.ipynb National Landslide-Susceptibility Exposure Assessment. Generates all visual assets based on the LED and Master Database, including: National Exposure Maps: High and Moderate-to-High susceptibility areas; Geographic Summaries: State-level and Tract-level exposure plots; Regional Breakdowns: Analytics sliced by Physiographic Divisions and Census Divisions; Socioeconomic Characteristics: Plots correlating landslide exposure with Income, Poverty, and LISA cluster result.