1 of 18

Spatial Data Lab Internship

Final Presentation

​

Miguel Pires, University of Lisbon

September 2024 - February 2025

​

2 of 18

“Comparison and Analysis of Spatial and Spatiotemporal Clustering of Wildfires in the United States of America”

Gunjan Barua

Professor

Kim Junghwan

Virginia Tech

​

Virginia Tech

Miguel Pires

University of Lisbon

3 of 18

Objectives of the Study

This study aims to perform spatial and spatiotemporal clusters of wildfire occurrences across the United States of America. Then, we look for answers to the following questions:

Results aim to optimize fire management resources, supporting policymakers in creating targeted wildfire prevention and mitigation policies.

  • How do spatial and spatiotemporal clusters differ? Why?

​

  • What should one consider before selecting one of these approaches?

​

  • Does incorporating the wildfire return interval refine the definition of high-risk areas?

4 of 18

Background | Motivation

Upwards trend in wildfires in the US

No prior studies used this method in the USA.

Similar methods can be applied in other areas

Potential to influence

policy / fire management

Currentness

Universality

Novelty

Impactful

5 of 18

Method - Dataset (1899-2024)

Similar methods can be applied in other areas

data-nifc.opendata.arcgis.com

6 of 18

Method - Software used

Similar methods can be applied in other areas

https://www.satscan.org/

7 of 18

Results - June

Similar methods can be applied in other areas

Spatial Cluster

Spatiotemporal Cluster

8 of 18

Results - June

Similar methods can be applied in other areas

Spatial Cluster

Spatiotemporal Cluster

9 of 18

Results - July

Spatial Cluster

Spatiotemporal Cluster

10 of 18

Results - July

Spatial Cluster

Spatiotemporal Cluster

11 of 18

Results - August

Spatial Cluster

Spatiotemporal Cluster

12 of 18

Results - August

Spatial Cluster

Spatiotemporal Cluster

13 of 18

Discussion

  • There are differences between the spatial and spatiotemporal clusters

​

  • Spatial clusters consider areas which had a large amount of wildfire occurrences

​

  • Spatiotemporal clusters identify areas where wildfires are recurrent

​

  • Spatiotemporal clusters reveal hidden patterns in wildfire occurrences, indicating their potential for fire management

14 of 18

Challenges

  • Choice of probability distribution

​

​

  • Choice of SaTScan parameters

​

​

  • Lack of literature on this topic

​

15 of 18

My Role

  • Conducting literature review

​

​

  • Assisting in performing the clustering

​

​

  • Writing the introduction of the paper

​

16 of 18

What I have learned in the SDL

  • Best practises in research, how to conduct literature review, use citation tools, which journal to choose, …

​

  • How to perform spatial and spatiotemporal clustering with SaTScan

​

  • Data Processing in ArcGIS

​

  • Gained knowledge on clustering and scan statistics

​

  • Interdisciplinary collaboration

​

  • This experience further motivated me to pursue a career as a researcher in geospatial sciences

17 of 18

Future Work

  • The paper is to be published in 2025

​

​

  • Possible follow-up projects

​

​

  • I would like to remain engaged with the SDL and CGA

​

18 of 18

Thank You!

Do you have any questions?

fc58678@alunos.fc.ul.pt

+351 918 222 048

CREDITS: This presentation template was created by Slidesgo, and includes icons by Flaticon, and infographics & images by Freepik