Published October 29, 2026 | Version 1.2.0

ZIP-Code-Level Data on Rooftop Solar, Battery Storage, and EV Charging in California, 2023

  • 1. Massachusetts Institute of Technology

Description

A ZIP-code-level panel of distributed energy resources across California, built so that deployment can be compared across technologies rather than one at a time.

Distributed energy data is scattered across agencies, geographies and vintages: rooftop solar sits with one program, storage with another, EV charging in a dashboard export, and none of them share a common set of controls. This record assembles them into a single 2023 panel keyed to Census ZIP Code Tabulation Areas, so questions that require comparing technologies — who has storage but not solar, where charging is dense but residential access is not — can be asked directly.

Coverage. 2,548 California ZCTAs by 50 columns. Rooftop solar PV capacity, battery storage capacity, EV charging counts split by Level 1, Level 2 and DC fast, EV registrations, wind capacity and turbine counts, and power-plant capacity. Each is joined to American Community Survey 2023 five-year demographics and socioeconomics, exhaustive housing structure and tenure, NASA POWER climate and solar resource, utility service territory, annual electricity demand from all three investor-owned utilities, and California Energy Commission energy burden and affordability measures.

Using it. Geography is the 2023 ZCTA, keyed by zip_code as a five-digit string — read it as text, not as an integer. Coverage is not universal on every column: housing structure is complete for 1,717 ZCTAs, tenure for 1,713, and energy burden for 1,669, so expect missingness rather than assuming a balanced panel. The four housing-structure shares are exhaustive and sum to one, with single-family the natural omitted reference.

Beyond the panel. The record also carries the 740 model-output tables (254 coefficient tables with their VIF and fit statistics) behind every figure in the accompanying paper, so results can be redrawn without refitting; the raw inputs that have no stable public download URL, including LBNL's Tracking the Sun file; the exact Census API response and its query manifest, preserved because the Census serves current vintages rather than historical snapshots; and the pinned environment the pipeline runs under. CHECKSUMS.sha256 covers every file.

Underlying sources retain their own citation requirements, in particular LBNL Tracking the Sun, the USGS US Wind Turbine Database, CaliforniaDGStats, NASA POWER, the California Energy Commission and the US Census Bureau. SOURCES.md documents each with its origin.

Notes

Version 1.2.0 corrects the storage outcome. Earlier versions applied the 5 MW per-system DER size limit to ZIP x sector totals from Storage_LatLong.xlsx, which removed all residential storage from the 49 ZIPs whose residential total exceeded 5 MW. storage_capacity_mw is now built from the record-level CEC export (Storage_Map_Zip_Pie_Full Data_data.xlsx), summing residential and commercial systems of up to 5 MW each excluding utility-sector records and, matching the Tracking the Sun cutoff, systems approved after 31 December 2023 (1,614 MW statewide). The ZIP panel drops from 2,549 to 2,548 rows: ZIP 93633 (population 32) had only post-2023 storage. Every storage coefficient table, and the Model 7, O and 9A tables that use storage as a control or predictor, are regenerated; no other outcome changes. Raw third-party inputs are not redistributed in this record. They are public and are cited with retrieval URLs in SOURCES.md, and scripts/fetch_exact_public_data.py downloads the pinned snapshots. Version 1.1.0 changes only file names and documentation; no data or coefficient values changed. The model-output tables for the education and housing-value specifications are renamed Model 2A and Model 2B to match the paper's labels (60 files, byte-identical to v1.0.0), and the record description and README now give the correct table count (740 model-output files: 254 coefficient tables plus VIF and fit statistics) and tenure coverage (1,713 ZCTAs).

Files

zenodo-bundle.zip

Files (230.6 MB)

Name Size
md5:916aefeabf2c089dada7ee45effd52cc
230.6 MB Preview Download

Additional details