A Multimodal Spatial and Temporal Features-Based Dataset for Wildfire Risk Prediction

Forest fires are an increasing threat to ecosystems and communities, exacerbated by climate change driving increased frequency and intensity. Accurate fire risk assessment and early detection are essential for effective mitigation and response. Intelligent predictive models are currently at forefront of this effort, yet they are highly dependent on the quality, quantity, and diversity of data. A key challenge in wildfire prediction lies in the integration of spatial and temporal features, making multimodal data fusion critical. In this paper, we propose a methodology to create multimodal datasets designed to support wildfire risk prediction. Our approach integrates meteorological, topographical, and fire history data across various formats, spatial resolutions, and temporal scales. Datasets include fire-related variables, such as, burned area occurrence, ignition events, and fire causes, alongside input features like temperature, humidity, wind conditions, land use and land cover classification, vegetation indices, and terrain attributes. Data sources include Copernicus Earth Observation data, satellite imagery, IPMA meteorological records, and ICNF fire reports, offering a comprehensive view of wildfire risk factors. The resulting dataset enables a wide range of applications, including machine learning-based fire prediction, risk mapping, and informed land management. This work contributes a valuable resource for researchers, policymakers, and emergency response teams, advancing data-driven strategies for forest fire prevention and management.

• Publication year

  2025

• Venue

  Experiment@ International Conference

• Publication date

  2025-09-03

• Fields of study

  Not labeled

• Identifiers
  DOI 10.1109/exp.at2565440.2025.11348433
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Machine Learning Algorithms Applied to Wildfire Data in California's Central Valley

  2024cited by this paper
• Wildfire Risk Prediction: A Review

  2024cited by this paper
• The Portuguese Large Wildfire Spread database (PT-FireSprd)

  2023cited by this paper
• Wildfire Prediction Model Based on Spatial and Temporal Characteristics: A Case Study of a Wildfire in Portugal’s Montesinho Natural Park

  2022cited by this paper

• No citing papers are available for this paper.