Varga et al. (2025) A 32-year species-specific live fuel moisture content dataset for southern California chaparral

Identification

• Journal: Open MIND
• Year: 2025
• Date: 2025-11-24
• Authors: Kevin Varga, Charles Jones
• DOI: 10.5061/dryad.rjdfn2zkw

Research Groups

• University of California, Santa Barbara
• James Madison University
• Earth Research Institute

Short Summary

This study developed a 32-year, species-specific Live Fuel Moisture Content (LFMC) dataset for southern California chaparral by training random forest models with observations and environmental predictors. The resulting dataset successfully captures the annual cycle, spatial heterogeneity, and interspecies differences of LFMC for four key fuel types, providing a valuable resource for wildfire research.

Objective

• To create a historical, 32-year long, species-specific Live Fuel Moisture Content (LFMC) dataset for southern California chaparral.
• To predict the LFMC of four specific fuel types (chamise, old growth chamise, black sage, and bigpod ceanothus) using random forest models.

Study Configuration

• Spatial Scale: Southern California chaparral, spanning from San Luis Obispo, California, to Los Angeles, California, with a spatial resolution of 1 kilometer.
• Temporal Scale: A 32-year period from December 1987 through June 2019, with semi-monthly (1st and 15th of each month) LFMC outputs.

Methodology and Data

• Models used: Species-specific random forest models were trained and tested using 5-fold cross-validation and observation site-specific cross-validation. Quantile mapping bias correction was applied to the chamise LFMC dataset.
• Data sources:
  • Over ten thousand live fuel moisture content (LFMC) observations for chamise (new and old growth), black sage (new growth), and bigpod ceanothus (new growth).
  • Predictor variables from Landsat imagery (e.g., near-infrared reflectance of vegetation, NIRv).
  • Predictor variables from the Weather Research and Forecasting (WRF) model (e.g., 90-day precipitation, 90-day mean temperature, 150-day mean insolation, 7-day mean soil moisture).
  • Day length.
  • Data derived from the National Fuel Moisture Database, Santa Barbara County Fire Department, and Santa Barbara Area WRF Climatology.

Main Results

• A historical, 32-year (December 1987 – June 2019), species-specific LFMC dataset was created for southern California chaparral at a 1-kilometer spatial resolution with semi-monthly outputs.
• The dataset includes LFMC predictions and uncertainty calculations for four fuel types: new growth chamise, old growth chamise, new growth black sage, and new growth bigpod ceanothus.
• The chamise output, which underwent quantile mapping bias correction and is considered the most robust, achieved a mean absolute error of 9.68% and an R² value of 0.76.
• The generated LFMC dataset successfully captures the variability in the annual cycle, spatial heterogeneity, and interspecies differences in live fuel moisture.

Contributions

• This study provides the first historical, 32-year long, species-specific live fuel moisture content dataset for southern California chaparral at a high spatial (1 km) and temporal (semi-monthly) resolution.
• The dataset includes uncertainty estimates for the LFMC predictions, enhancing its utility for fire behavior modeling.
• It offers a valuable resource for understanding varying fire season characteristics and landscape-level flammability, contributing to improved wildfire management and research.

Funding

• NASA Earth Science: 80NSSC21K1630 (Terrestrial Ecology)
• University of California Office of the President: LFR-20-652467

Citation

@article{Varga202532year,
  author = {Varga, Kevin and Jones, Charles},
  title = {A 32-year species-specific live fuel moisture content dataset for southern California chaparral},
  journal = {Open MIND},
  year = {2025},
  doi = {10.5061/dryad.rjdfn2zkw},
  url = {https://doi.org/10.5061/dryad.rjdfn2zkw}
}