Forests (Dec 2023)

Autoregressive Forecasting of the Number of Forest Fires Using an Accumulated MODIS-Based Fuel Dryness Index

  • Daniel José Vega-Nieva,
  • Jaime Briseño-Reyes,
  • Pablito-Marcelo López-Serrano,
  • José Javier Corral-Rivas,
  • Marín Pompa-García,
  • María Isabel Cruz-López,
  • Martin Cuahutle,
  • Rainer Ressl,
  • Ernesto Alvarado-Celestino,
  • Robert E. Burgan

DOI
https://doi.org/10.3390/f15010042
Journal volume & issue
Vol. 15, no. 1
p. 42

Abstract

Read online

There is a need to convert fire danger indices into operational estimates of fire activity to support strategic fire management, particularly under climate change. Few studies have evaluated multiple accumulation times for indices that combine both dead and remotely sensed estimates of live fuel moisture, and relatively few studies have aimed at predicting fire activity from both such fuel moisture estimates and autoregressive terms of previous fires. The current study aimed at developing models to forecast the 10-day number of fires by state in Mexico, from an accumulated Fuel Dryness Index (FDI) and an autoregressive term from the previous 10-day observed number of fires. A period of 50 days of accumulated FDI (FDI50) provided the best results to forecast the 10-day number of fires from each state. The best predictions (R2 > 0.6–0.75) were obtained in the largest states, with higher fire activity, and the lower correlations were found in small or very dry states. Autoregressive models showed good skill (R2 of 0.99–0.81) to forecast FDI50 for the next 10 days based on previous fuel dryness observations. Maps of the expected number of fires showed potential to reproduce fire activity. Fire predictions might be enhanced with gridded weather forecasts in future studies.

Keywords