top of page

Infrastructure and California Wildfires

Key Words

Infrastructure, Google Earth Engine, SWAT, Electric Lines, Power, Geopolitics

This project investigates the role of electric transmission lines in igniting wildfires in California, highlighting how human infrastructure challenges the traditional view of wildfires as purely natural disasters. Focusing on monopoly providers like PG&E and using tools like Google Earth Engine, it analyzes the 2018 Camp Fire to explore the intersection of infrastructure, climate policy, and post-fire recovery.

Date

Dec 2023

Screenshot 2025-04-28 at 12.55.43 AM.png
Research Statement

This research explores the complexities of wildfires in California, focusing particularly on the human-induced risks posed by electric transmission lines operated by monopoly providers like PG&E. Unlike wildfires sparked by natural causes or remote land-clearing practices, those caused by electrical infrastructure often occur near residential areas, intensifying their impact and underscoring the need to reconsider the traditional view of wildfires as purely natural disasters. The conventional "fire triangle" model—centered on oxygen, heat, and fuel—fails to account for the role of human infrastructure in wildfire ignition, necessitating a rethinking of how we define and address these events. To analyze this more nuanced understanding, the project examines the 2018 Camp Fire using tools like Google Earth Engine and SWAT to assess environmental recovery and long-term impacts. It also employs Virtual Reality to create immersive educational experiences, fostering deeper connections between people and forest ecosystems. By combining spatial analysis, remote sensing, and digital storytelling, the research offers an integrated methodology for understanding wildfire dynamics, addressing climate justice, and promoting sustainable coexistence between humans and the environment.

Precedent Projects

A Google Earth Engine Approach for Wildfire Susceptibility Prediction Fusion with Remote Sensing Data of Different Spatial Resolutions (Tavakkoli Piralilou et al.)
This project combines remote sensing data from sources like Landsat and Sentinel-2 using machine learning (Random Forest and SVM) on Google Earth Engine to predict wildfire susceptibility. By fusing multiple spatial resolutions and applying the Dempster–Shafer theory, it improves prediction accuracy and demonstrates how spatial data can guide fire risk analysis. Its approach aligns with my research on electric-line-induced wildfires, offering valuable tools, datasets (e.g., MODIS), and modeling techniques for assessing wildfire risks around infrastructure.

​

Assessing the Hydrologic Response to Wildfires in Mountainous Regions (Havel et al.)
This study uses the SWAT model to examine the hydrological impacts of the 2012 High Park and Hewlett fires in Colorado, revealing changes in runoff and sediment flow. Through remote sensing and GIS, it assesses burn severity and land cover change—methods relevant to analyzing fire effects from electrical infrastructure. The project emphasizes the ecological aftermath of wildfires and offers techniques I can apply to evaluate water resource impacts in fire-prone areas near residential zones.

​

Is Burying Power Lines Fire-Prevention Magic, or Magical Thinking? (Evans)
This article investigates the push for undergrounding power lines to prevent wildfires in California, highlighting utilities like PG&E and SDG&E’s efforts despite high costs. It also presents cheaper alternatives like line coating and vegetation management. These practical insights help inform the final stage of my project, where I plan to propose new landscape or architectural systems to reduce fire risk based on data from my GEE and SWAT analyses.

Data and Resources

Fire Ignition Data from CPUC (Wildfire and Wildfire Safety)
The CPUC dataset tracks utility-caused wildfires but omits major events like the 2018 Camp Fire, raising concerns about data accuracy and transparency. Allegations of past corruption between PG&E and CPUC add to suspicions. These gaps limit the dataset’s usefulness for analysis and highlight the need for stricter reporting standards.

​

Burn Severity Data (Home | MTBS)
MTBS maps wildfire burn severity across the U.S. using Landsat imagery. It aids in tracking fire impacts and recovery, supporting research and management. The project reflects how federal data collection influences policy, environmental justice, and land use decisions.

 

California Electric Transmission Lines
This CEC dataset maps transmission lines and includes ownership, capacity, and status. It allows filtering by provider, enabling focused analysis on PG&E’s lines—crucial for studying the 2018 Camp Fire.

​

California Wildfire History Map
Covering 20,000+ wildfires since 1878, this dataset reveals trends but excludes smaller fires and lacks updated cause data. These omissions reflect policy choices and raise questions about the completeness and politics of wildfire recordkeeping.

Workflows

Case Study: 2018 Camp Fire
The Camp Fire began on November 8, 2018, in Butte County, CA, killing 85 people and destroying over 18,000 structures. Caused by a faulty PG&E transmission line, the fire rapidly spread across 153,336 acres, devastating communities like Paradise. The event exposed major issues in utility infrastructure and emergency response.

​

Workflow 1: Mapping Wildfire, Infrastructure, and Urban Areas
This workflow maps the spatial relationship between the Camp Fire boundary, PG&E transmission lines, and urban areas to highlight how electric-line-induced wildfires often start near residential zones. Unlike remote forest fires, these pose direct threats to people and property.

 

Steps

  • Use GEE to generate Camp Fire boundary using MODIS (MOD14A1, MYD14A1)

  • Map urban areas using Copernicus Global Land Cover (CGLS-LC100)

  • Extract urban pixels

  • Import PG&E electric line shapefile into QGIS

  • Combine maps (fire, electric lines, urban areas) in QGIS

  • Final editing in Photoshop

​

Technical Resources

Screenshot 2025-05-16 at 1.29.56 AM.png
Screenshot 2025-05-16 at 1.30.26 AM.png

Workflow 2: GEE – Burned Area and Burn Severity Mapping:

Wildfires like the 2018 Camp Fire drastically alter ecosystems, affecting vegetation, soil, water quality, and biodiversity. As vegetation is lost, runoff and erosion increase, disrupting nutrient cycling and damaging habitats. To assess these impacts, fire intensity and burn severity are key metrics: fire intensity reflects the energy released, while burn severity measures ecological damage, including vegetation loss and soil change.

​

Using remote sensing, the Normalized Burn Ratio (NBR) distinguishes burned from unburned areas by analyzing NIR and SWIR reflectance. The Differenced NBR (dNBR), which compares pre- and post-fire NBR values, produces detailed maps of burn severity—critical for understanding fire effects and planning recovery strategies.

​

Screenshot 2025-05-16 at 1.32.04 AM.png
Screenshot 2025-05-16 at 1.32.33 AM.png

Workflow 3: SWAT – Post-Fire Assessment:

Wildfires alter landscapes by removing vegetation and changing soil properties, leading to increased runoff, erosion, and pollution. The Soil and Water Assessment Tool (SWAT) models these impacts over time, simulating changes in water flow, sediment yield, and chemical transport.

​

SWAT helps assess:

  • Runoff & Erosion: Simulates increased runoff and sediment flow due to vegetation loss and soil changes.

  • Water Quality: Tracks nutrient and chemical loading into water bodies post-fire.

  • Vegetation & Land Use: Models regrowth and land use shifts like reforestation or new farming practices.

  • Hydrological Changes: Reflects altered infiltration, groundwater recharge, and evapotranspiration due to burned landscapes.

 

This workflow is essential for understanding long-term post-fire watershed impacts and planning recovery strategies.

​​

​​

​

SWAT results

​

🌲 Watershed Overview

  • Total Area: 48,683 ha (~120,298 acres)

  • Number of Subbasins: 20

  • Number of HRUs: 169 (Hydrologic Response Units, each representing a unique combination of land use, soil type, and slope)

 

🚜 Land Use Distribution (% of Total Watershed)

  • Pasture (PAST): 52.5% – the dominant land use, suggesting significant agricultural presence

  • Range-Grasses (RNGE): 29.1% – covers natural vegetation areas

  • Deciduous Forest (FRSD): 16.2% – key for soil stability and biodiversity

  • Agricultural Land (AGRL): 0.5% – minimal direct cropping

  • Water Bodies (WATR): 1.8%

 

🧱 Soil Composition (% of Total Watershed)

The watershed comprises diverse soils. Key contributors:

  • Soil 374386: 26.0%

  • Soil 367898: 14.8%

  • Soil 374388: 13.6%

  • Indicates variability in infiltration, erosion potential, and post-fire hydrological response.

 

🏞️ Slope Classification

  • Low Slope (0–1%): 69.0%

  • High Slope (>1%): 31.0%

  • High low-slope coverage means greater surface runoff risks post-fire, especially where vegetation is lost.

 

🔎 Subbasin Example (Subbasin 1)

  • Area: 1,395 ha

  • Land Use: 83.5% pasture, 16% range

  • Main Soils: 374386 and 374388

  • Slope: 55% steep, 45% flat

  • Suggests risk of erosion and sediment transport due to slope and pasture dominance.

 

🌧️ Implications for Post-Fire Assessment

  • High pasture and range-grass coverage indicates vulnerability to erosion and nutrient runoff without vegetative protection.

  • Soil types and slopes across HRUs provide critical inputs for simulating sediment yield, nutrient loading, and hydrologic changes in SWAT.

  • These configurations can now be used to simulate changes in runoff, sediment transport, and nutrient loss post-wildfire—especially in heavily burned subbasins.

​

​

​

 
​

 

Key Insights from the SWAT Output

  1. Pasture and Range-Grasses Dominate the Watershed

    • Over 80% of the watershed is covered by pasture (52.5%) and range-grasses (29.1%).

    • These land uses offer limited canopy cover and are highly vulnerable to post-fire runoff and erosion, especially when burned.

  2. Soil Variability with Erosion Risk

    • The most prevalent soil type is 374386 (26%), followed by types like 367898 and 374401.

    • Some of these may have poor post-fire infiltration or increased erodibility—key for modeling sediment yield and nutrient runoff.

  3. Gentle Slopes Dominate the Terrain

    • Nearly 69% of the watershed has a 0–1% slope, meaning large surface areas are prone to overland flow.

    • While steep slopes typically drive erosion, flat slopes combined with vegetation loss can result in widespread flooding and sediment transport.

  4. Subbasins with High Agricultural Use and Steep Slopes

    • Several subbasins combine pasture use with higher slope classes (>1%), making them hotspots for erosion and nutrient leaching after fire events.

 

Implications for the Post-Fire Wildfire Project

  1. Targeted Risk Zones
    The model helps pinpoint subbasins where fire damage could most severely disrupt water and soil systems—especially areas with pasture on slopes and sensitive soils.

  2. Prioritizing Post-Fire Recovery
    Areas with dominant pasture/range cover and sensitive soils should be prioritized for interventions like:

    • Vegetative restoration

    • Erosion control structures

    • Buffer zones near water bodies

  3. Water Quality Monitoring
    Due to nutrient and sediment runoff risks, monitoring should focus on nitrogen, phosphorus, and turbidity levels in streams flowing from highly affected HRUs.

  4. Landscape Design Opportunities
    You can incorporate ecological landscape strategies (e.g., bioswales, terracing, reforestation) in high-risk subbasins to improve resilience to future fires and reduce downstream impacts.

  5. Policy Recommendations
    The data supports advocating for land use regulations that limit agricultural expansion or improve post-fire land management in erosion-prone areas.

output (1).png
bottom of page