Human-ignited fires result in more extreme fire behavior and ecosystem impacts

Climate change and socio-economic factors have significantly altered global fire regimes, resulting in increased wildfire activity, particularly megafires in temperate regions. California exemplifies this trend, with a more than doubled mean annual burned area over the last decade. Research has focused on the roles of historical fire suppression and climate change in this increase, but less attention has been paid to how these factors influence fire behavior and subsequent ecological impacts. While burned area is a readily available metric, it provides an incomplete picture. Fire intensity, determined by fuel type, fuel consumption, and spread rate, is crucial in understanding ecological effects. This study aims to bridge this gap by linking meteorology, ecosystem status at ignition, fire behavior, and post-fire impacts, specifically comparing human-caused and lightning-caused fires to test the hypothesis that human-caused fires are more devastating due to a higher probability of ignition during extreme fire weather.

Numerous studies have examined the drivers of increased burned area in California, citing historical fire suppression leading to fuel buildup and climate change increasing temperatures and vapor pressure deficit (VPD), a key factor in fuel moisture content. However, these factors alone cannot fully explain the observed increase. Increased population density in the wildland-urban interface (WUI) also contributes to increased fire occurrence. While current burned areas are likely below Holocene levels, the contemporary fire regime is characterized by high-intensity, high-severity fires resulting in fatalities and substantial infrastructure and ecosystem damage. Past research has explored individual linkages between these factors, but comprehensive studies exploring the entire causal chain from ignition to ecological impact, particularly comparing ignition sources, are lacking.

This study utilizes a novel dataset of daily fire spread for large, multi-day wildfires in California (2012–2018) derived from 375 m resolution VIIRS active fire detections. The dataset includes 214 fires, encompassing 2939 fire days and 21,558 km² of burned area. Ignition sources (human or lightning) were determined from the California Fire Resource Assessment Program (FRAP) database. Daily fire rate-of-spread was calculated using the 95th percentile of the distance between active fire detections along the fire line and the previous day's line. Environmental conditions at ignition (potential evapotranspiration, windspeed, forest biomass, forest cover) were extracted from GridMET and LEMMA datasets. Fire severity was assessed using remotely sensed indicators: differenced normalized burn ratio (dNBR), relative differenced normalized burn ratio (rdNBR), and tree mortality (percent reduction in tree basal area). Statistical analyses, including Welch Two-Sample t-tests and Kolmogorov-Smirnov tests, were employed to compare fire characteristics and impacts between human- and lightning-caused fires.

Human-caused fires exhibited significantly larger sizes than lightning-caused fires from day one onwards. Human-caused fires grew 6.5 times larger on average by the end of the first day (18.8 vs. 2.9 km²). After five days, human-caused fires were more than three times larger. This difference stems from significantly higher daily fire spread rates for human-caused fires (1.83 km/day) compared to lightning-caused fires (0.84 km/day). Human-caused fires were initiated under significantly more extreme fire weather conditions (higher potential evapotranspiration, windspeed) and in areas with lower tree cover and biomass. A strong positive relationship was observed between fire rate-of-spread and tree mortality; fast-moving fires (>2 km/day) resulted in a threefold increase in tree mortality (48.3 ± 19.7%) compared to slow-moving fires (<0.5 km/day) (15.3 ± 18.0%). This relationship held true for both human- and lightning-caused fires. Human-caused fires showed significantly higher tree mortality compared to lightning-caused fires. The top 10% of days with the fastest fires accounted for 55% of the burned area, highlighting the disproportionate impact of extreme fires.

The findings strongly support the hypothesis that human ignitions, particularly during extreme fire weather, lead to faster-spreading, more intense fires with greater ecosystem impacts. The increased ignition frequency throughout the year increases the likelihood of ignitions coinciding with extreme weather conditions. While lightning ignitions can occur under similar conditions, they often coincide with higher humidity or rainfall, limiting spread. Furthermore, the location of lightning strikes frequently limits the effectiveness of initial suppression efforts. The skewed distribution of fire spread rates, with frequent slow spread and infrequent fast spread, indicates the importance of considering non-linear interactions between fire weather, fuels, and topography in fire models. Fuel type significantly impacted initial spread; faster rates were observed in areas with dry grasses and shrubs compared to denser forests. The strong link between fire rate-of-spread and fire severity (via tree mortality) suggests the need to incorporate spread rate into fire intensity and severity models.

This study reveals a human-ignited fire syndrome where ignitions during extreme weather lead to rapid fire spread, high intensity, and severe ecosystem impacts, notably high tree mortality. In contrast, lightning-caused fires, while potentially large, tend to have lower severity due to often less extreme weather at ignition and remoteness limiting early suppression. Reducing human ignitions during extreme weather is crucial for mitigating the escalating impact of wildfires in California. Future research should focus on the differential allocation of fire suppression resources between human- and lightning-caused fires and further refine the complex relationships between fire weather, fuels, and topography to improve fire behavior and ecosystem impact models.

The study focused on large, multi-day fires, representing a subset of all fires in California. Data availability for some fire severity indicators limited the analysis. The study focused solely on two large California ecoregions and results might not be generalizable to other ecosystems. The dataset only reflects fires not contained on the first day, potentially excluding smaller, quickly suppressed events. Finally, the use of remote sensing data relies on certain assumptions on data accuracy.