Wildfire Size Disparity on Tribal Lands in California and Associated PM2.5 Burden, 2000–2018

Study Overview This dataset supports a quantitative investigation of wildfire size disparity between federally recognized tribal lands and non-tribal lands in California from 2000 to 2018, and the associated PM2.5 air quality burden borne by tribal communities. Using spatial analysis, panel econometrics, and five convergent causal identification strategies, the study documents that fires on tribal land are on average 15.7 times larger than fires on adjacent non-tribal land — a disparity that persists at 6.5× even after excluding the catastrophic 2003 and 2007 fire years. The theoretical argument connects this disparity to the historical suppression of traditional Indigenous burning practices under federal fire management policy, which produced structural fuel load accumulation on tribal territories. Data Contents This archive contains all inputs, intermediate datasets, and outputs necessary to fully replicate the analysis from raw public data sources. Seventeen files are included: seven publication-quality figures (Figures 1–7), eight CSV data files, and two regression summary text files. The complete replication script (tribal_fire_full_study.py, 1,560 lines, available at [repository URL]) reproduces all results in a single execution and downloads source data directly from public APIs. The primary data files are: the county × year PM2.5 panel (816 observations, 43 California counties, 2000–2018); the classified California fire perimeter dataset (4,496 fires with tribal land indicator); the tribal nation fire burden summary (20 nations ranked by total acres burned); event study coefficients with 95% confidence intervals; county-level lightning flash density from the NASA LIS/OTD HRFC climatology; and county × year lightning panels from the NOAA Storm Events database. The Smoke.ipynb notebook contains exploratory analysis of the Cedar Fire 2003 PM2.5 event. Methods Summary Fire perimeter data were obtained from the NIFC Historical GeoMAC Perimeters archive (2000–2018) via ArcGIS REST API. Tribal boundaries are from the Census TIGER/Line AIANNH 2025 vintage (185 California areas). PM2.5 concentrations are from EPA AQS daily FRM/FEM monitoring data (parameter 88101; 580,066 California daily observations). Lightning data combine NOAA Storm Events county-year incident counts and the NASA LIS/OTD 0.5-degree High Resolution Full Climatology (HRFC) flash density grid. All spatial operations use the California Albers Equal Area projection (EPSG:3310) for area and centroid calculations. Causal identification employs five strategies: (1) regression discontinuity at tribal land boundaries (four bandwidth specifications, 10–100 km); (2) propensity score matching with caliper 0.05 (112 matched fire pairs, all post-match SMDs < 0.05); (3) event study with county and year fixed effects and pre-trend joint F-test (F = 2.39, p = 0.104 — no pre-trend); (4–5) lightning strike instrumental variable estimation using NOAA and NASA FIRMS sources. The PM2.5 baseline regression uses county-clustered standard errors to account for serial correlation (Durbin-Watson = 0.399). Both lightning instruments have weak first stages (F = 4.03 and F = 1.17); reduced form estimates are reported in lieu of 2SLS point estimates. Key Findings Mean fire size on tribal land: 42,572 acres. Mean fire size on non-tribal land: 2,706 acres. Ratio: 15.7× (Mann-Whitney p < 0.0001). The disparity is stable across all sensitivity specifications. The 2003 Cedar Fire burned across nine San Diego County tribal nations, producing a maximum AQI of 328 (Hazardous) — a 125% spike over baseline. The top 10 tribal nations by total fire burden are dominated by San Diego County nations (Barona: 873,341 total acres; Viejas: 546,467 acres) and the Karuk Tribe of Siskiyou County (419,722 acres). The event study post-period F-test is significant (F = 5.37, p = 0.006), confirming elevated PM2.5 following tribal fire years. Lightning reduced form coefficients: NOAA 1.188 μg/m³ per strike (p < 0.001); NASA FIRMS 2.287 μg/m³ per fl/km²/yr (p < 0.001). Computational Environment Python 3.12 · geopandas 1.1.3 · pandas 2.2.2 · statsmodels 0.14.6 · scipy 1.16.3 · linearmodels 7.0 · scikit-learn 1.6.1 · netCDF4 1.7.4 · Executed on Google Colab CPU runtime · Approximate runtime: 8–12 minutes License Creative Commons Attribution 4.0 International (CC BY 4.0). Users are free to share and adapt the material for any purpose provided appropriate credit is given to the authors.研究概况 本数据集支持针对2000年至2018年加利福尼亚州联邦认可部落土地与非部落土地之间野火规模差异，以及部落社区所承担的相关PM2.5空气质量负担开展定量研究。本研究采用空间分析、面板计量经济学与五种收敛性因果识别策略，结果显示：部落土地上的野火平均规模是邻近非部落土地野火的15.7倍；即便剔除2003年和2007年的灾难性火灾年份，这一差异仍维持在6.5倍。本研究的理论逻辑将该规模差异归因于联邦消防管理政策下传统原住民焚火实践被长期压制，导致部落领地内的燃料负荷结构性积累。 数据内容 本数据集存档包含从原始公开数据源完整复现分析所需的全部输入数据、中间数据集与最终输出结果。本次归档共包含17个文件：7份可用于发表的高质量图表（图1至图7）、8个CSV格式数据文件，以及2份回归结果汇总文本文件。完整复现脚本（tribal_fire_full_study.py，共1560行，可于[仓库链接]获取）可一键复现全部研究结果，并直接从公开API下载源数据。 核心原始数据文件包括：县级×年度PM2.5面板数据（共816条观测值，覆盖加利福尼亚州43个县，时间跨度2000-2018年）、分类后的加利福尼亚州野火边界数据集（共4496起野火，附带部落土地归属标识）、部落野火负担汇总表（按总过火面积排序的20个部落）、带有95%置信区间的事件研究系数、来自NASA LIS/OTD HRFC气候数据集的县级闪电闪击密度数据，以及来自NOAA风暴事件数据库的县级×年度闪电面板数据。Smoke.ipynb笔记本包含对2003年雪松野火PM2.5事件的探索性分析。 研究方法概述 野火边界数据通过ArcGIS REST API从NIFC Historical GeoMAC Perimeters存档（2000-2018年）获取。部落边界数据来自美国人口普查局TIGER/Line AIANNH 2025版数据集（覆盖加利福尼亚州185个区域）。PM2.5浓度数据来自EPA AQS每日FRM/FEM监测数据（参数代码88101；共包含580066条加利福尼亚州每日观测值）。闪电数据整合了NOAA风暴事件数据库的县级年度闪电事件计数，以及NASA LIS/OTD 0.5度高分辨率全气候（HRFC）闪电密度网格数据。所有空间运算均采用加利福尼亚阿尔伯斯等面积投影（EPSG:3310）进行面积与质心计算。 本研究采用五种因果识别策略：（1）部落土地边界处的断点回归设计（包含4种带宽设定，范围10-100公里）；（2）卡尺值为0.05的倾向得分匹配（共得到112对匹配野火，匹配后所有标准化均差均小于0.05）；（3）纳入县级与年度固定效应并进行事前趋势联合F检验的事件研究（F=2.39，p=0.104，未检测到事前趋势）；（4-5）基于NOAA与NASA FIRMS数据源的闪电闪击工具变量估计。PM2.5基准回归采用县级聚类标准误以控制序列相关（Durbin-Watson统计量为0.399）。两款闪电工具的一阶阶段均较弱（F值分别为4.03与1.17），因此本文报告简约式估计结果以替代两阶段最小二乘点估计值。 核心研究发现 部落土地野火平均规模为42572英亩，非部落土地野火平均规模为2706英亩，二者比值为15.7倍（Mann-Whitney检验p<0.0001）。该规模差异在所有敏感性分析设定下均保持稳定。2003年雪松野火波及圣迭戈县的9个部落领地，峰值空气质量指数（AQI）达到328级（危险级），较基准值上升125%。按总过火面积排序的前10大部落中，圣迭戈县的部落占多数（巴罗纳部落：总过火面积873341英亩；维耶哈斯部落：546467英亩），以及锡斯基尤县的卡鲁克部落（419722英亩）。事件研究的事后阶段F检验结果显著（F=5.37，p=0.006），证实部落野火发生年份后PM2.5浓度显著升高。闪电简约式估计系数：NOAA数据源为每闪击1次对应1.188 μg/m³（p<0.001）；NASA FIRMS数据源为每平方公里每年每闪击对应2.287 μg/m³（p<0.001）。 计算环境 Python 3.12 · geopandas 1.1.3 · pandas 2.2.2 · statsmodels 0.14.6 · scipy 1.16.3 · linearmodels 7.0 · scikit-learn 1.6.1 · netCDF4 1.7.4；运行于Google Colab CPU运行时环境；单次运行耗时约8-12分钟。 授权协议 本数据集采用知识共享署名4.0国际许可协议（Creative Commons Attribution 4.0 International，CC BY 4.0）。使用者可自由共享、改编本材料用于任何用途，但需向原作者提供适当署名。