Do wildfire-prevention power shutoffs fall disproportionately on California's officially-designated disadvantaged communities (SB 535 DACs)? Descriptive analysis of CPUC de-energization polygons (Oct 2021 +) joined to census tracts. Click any tract.
PSPS events, High Fire Threat District & IOU territories: California Public Utilities Commission (CPUC). SB 535 Disadvantaged Communities: CalEnviroScreen 4.0 (OEHHA/CalEPA). Non-DAC baseline tracts: US Census TIGERweb 2010 (CES 4.0 vintage).
Event polygons (De_Energization = "Yes", one composite key per utility/event/start-date) joined to 2010 census tracts. Per tract: number of intersecting events, summed event durations (exposure hours), ever-affected. The primary universe is IOU-covered tracts (representative point inside a territory covered by this CPUC event source; publicly-owned-utility tracts cannot appear in the source). DAC vs non-DAC compared within that universe and, as a restricted comparison, within HFTD Tier 2/3 ∩ IOU tracts (restriction is a robustness comparison, not a statistical control). p-values are two-sided Mann–Whitney U, plus a county-stratified DAC-label permutation (centered on the permutation null).
Polygons are dissolved at the event level, not circuit — intersection ≠ experienced outage. exposure_hours is a lower bound (some events lack a restoration time; those tracts are flagged). Customer counts are event-level; the per-tract figure is a labelled population-weighted ESTIMATE, not measured. p-values are descriptive only (spatial autocorrelation, MAUP). Data begins Oct 2021. DAC is a state policy designation, not ground truth. Descriptive only — no causal or liability claim about any utility.
Source: California Public Utilities Commission (CPUC). CPUC states: "Commercial use of the Dashboard, and the data within it, is strictly prohibited." This is non-commercial academic research (UW Humanistic GIS lab); CPUC is cited as the data source. Census data is public domain.