Masters Thesis Part 1
Hart, H.; Perrakis, D.D.B.; Taylor, S.W.; Bone, C.; Bozzini, C. Georeferencing Oblique Aerial Wildfire Photographs: An Untapped Source of Fire Behaviour Data. Fire 2021, 4, 81. https://doi.org/10.3390/fire4040081
Wildfires, despite their important role in maintaining healthy forest ecosystems, can have significant economic, environmental, human health and safety impacts. In recent years, these impacts have become more common and severe, calling for the advancement of wildfire behaviour modelling to better understand and predict fire behaviour.
In the province of British Columbia (BC), Canada, for example, the 2017 wildfire season was one of the most severe on record with over 1.22 million ha burned. Fire suppression costs reached over $649 million that year, and approximately 65,000 people were evacuated from communities. The province set a record for area burn yet again in the following year (1.35 million ha), with suppression costs estimated at $615 million.
Currently, wildfire growth modelling in Canada is accomplished by rate of spread (ROS) models for various fuel types in the Canadian Fire Behaviour Prediction (FBP) System. Yet, the current ROS models lack high ROS-ISI wildfire data which represents the majority of large scale and high risk wildfires. For wildfire management decision makers, misrepresented ROS models could lead to critically underestimating the growth of a wildfire.
Example ROS Models in the FBP System
One potential source of large scale wildfire data is an archive of historical wildfire imagery from the BC Wildfire Management Provincial Air Tanker Centre (PATC). This archive contains thousands of wildfire images from 2001 to 2019, however as raw images they cannot provide quantitative fire behaviour data, such as ROS. To address this challenge, the objective of this study is to explore a novel method of extracting fire behaviour data from oblique images taken during wildfire operations using a photogrammetric approach called monophotogrammetry.
Monoplotting has the ability to assign real world coordinates (X,Y,Z) to every pixel coordinate in the oblique aerial image, thus georeferencing the image. The principle of this theory is given that the original camera system, digital image, and DEM lie in a straight line in the world space, it can be said that a ray starting from the camera centre will pass through a set of pixel coordinates in the digital image and intersect with the DEM at the corresponding world coordinates.
Monoplotting Principle
For example, the following wildfire images were georeferenced using the WSL Monoplotting Tool (MPT). These images are from the same wildfire, on the same day, but at different times. The targets in the images represent Control Points (CP) used during the georeferencing process.
With the wildfire image georeferenced, spatial data such as the fire front position and burned area can be drawn on the aerial image using the MPT.
Image 1 @ 18:33
Image 2 @ 19:04
Image 1 Burned Area
Image 2 Burned Area
This custom spatial data can exported from the MPT and imported into GIS Programs, such as ArcGIS for further geo-spatial analysis.
Image 1 Fire Front Position
ArcGIS Services Unavailable
Image 2 Fire Front Position
ArcGIS Services Unavailable
Image 1 Spatial Data
ArcGIS Services Unavailable
Image 2 Spatial Data
ArcGIS Services Unavailable
Using the fire front positions from two successive wildfire images, the spread distance can be calculated, in this case with the Measure Tool. Try it for yourself below!
ArcGIS Services Unavailable
With the spread distance, and time stamp of each fire front, a preliminary spread rate calculation can be made.
time = (19:04 - 18:33) = 31 minutes
distance = 860 metres
ROS = distance/time = 27.74 m/min
In addition, further exploratory analysis has led to the extraction of additional fire behaviour characteristics using the MPT, such as flame depth, flame height, and smoke plume dimensions.
Similarly, the import/export routines of the MPT allow this data to be viewed in GIS programs, such as ArcGIS Pro, for further 3D spatial analysis. For example, 3D smoke column points.
Using the extracted wildfire behaviour from the wildfire image archive will not only help improve and validate existing ROS models, but will increase the overall understanding of wildfire behaviour as it known today.