[Paper Review] From spatio-temporal data to chronological networks: An application to wildfire analysis

Didier A. Vega‐Oliveros, Moshé Cotacallapa+5 more|Biblioteca Digital da Memória Científica do INPE (National Institute for Space Research)|Dec 4, 2018

Complex Network Analysis Techniques31 references4 citations

TL;DR

This paper proposes a novel chronological network approach to model spatiotemporal wildfire data by linking grid cells when consecutive fire events occur between them, overcoming limitations of correlation-based networks. The method reveals annual fire activity displacement patterns and subregional fire dynamics in the Amazon Basin using MODIS data, demonstrating superior detection of temporal and spatial event sequences.

ABSTRACT

Network theory has established itself as an appropriate tool for complex systems analysis and pattern recognition. In the context of spatiotemporal data analysis, correlation networks are used in the vast majority of works. However, the Pearson correlation coefficient captures only linear relationships and does not correctly capture recurrent events. This missed information is essential for temporal pattern recognition. In this work, we propose a chronological network construction process that is capable of capturing various events. Similar to the previous methods, we divide the area of study into grid cells and represent them by nodes. In our approach, links are established if two consecutive events occur in two different nodes. Our method is computationally efficient, adaptable to different time windows and can be applied to any spatiotemporal data set. As a proof-of-concept, we evaluated the proposed approach by constructing chronological networks from the MODIS dataset for fire events in the Amazon basin. We explore two data analytic approaches: one static and another temporal. The results show some activity patterns on the fire events and a displacement phenomenon over the year. The validity of the analyses in this application indicates that our data modeling approach is very promising for spatio-temporal data mining.

Motivation & Objective

• To address the limitations of correlation-based networks in capturing non-linear, recurrent, and sequential spatiotemporal events in complex systems.
• To develop a computationally efficient, adaptable method for constructing networks directly from event sequences in spatiotemporal data.
• To apply the method to real-world wildfire data in the Amazon Basin to uncover hidden temporal and spatial patterns.
• To compare static and temporal network analysis approaches for identifying active fire regions and migration trends.
• To validate the method’s ability to detect fire activity displacement and community structures linked to climate and land use.

Proposed method

• Divides the Amazon Basin into a spatial grid, with each cell represented as a node in a network.
• Establishes links between nodes only when two consecutive fire events occur between them, forming directed edges.
• Allows self-loops when consecutive events occur in the same grid cell.
• Applies both static and temporal network analysis techniques to evaluate network structure and dynamics.
• Uses time windows to segment data and analyze changes in network properties over time.
• Integrates external data (precipitation, land use) to interpret network findings in environmental context.

Experimental results

Research questions

• RQ1 Can chronological networks better capture sequential and recurrent spatiotemporal events than correlation-based networks?
• RQ2 What temporal patterns, such as fire activity displacement, emerge from chronological network analysis of Amazon wildfires?
• RQ3 How do subregional differences in dry season timing, fire intensity, and land use correlate with network community structures?
• RQ4 To what extent do static and temporal network analysis approaches agree in identifying the most active fire regions?
• RQ5 How does the network structure respond to environmental and anthropogenic factors like precipitation and deforestation?

Key findings

• The chronological network approach successfully identified a clear annual pattern of fire activity displacement, starting in the southeastern subregions and spreading toward the northwest.
• Subregions 4–7 exhibited dry seasons occurring in the same period, with 4 and 5 showing lower precipitation than 6 and 7.
• Subregions 2 and 3 had the longest and most severe dry seasons, while subregion 12 had the least intense but longest dry season, differing from all others.
• The static and temporal network analyses both identified the same set of most active fire cells, validating the robustness of the method.
• Land use patterns such as pasture, agriculture, and deforestation near roads and reserves were strongly associated with high-fire-activity communities in the network.
• The method revealed that fire propagation is not solely driven by proximity but by sequential event dynamics across regions, especially during dry seasons.