Beginning
Climate change has been seriously affecting the world in various ways, such as polar ice melting, rising sea levels, extreme drought, and severe heatwaves. These rapid changes have led many countries to experience more extreme weather and abnormal atmospheric behavior. Australia, in particular, experienced one of the most destructive wildfire seasons in recorded history during 2019 and 2020. Known as the “Black Summer,” the fires killed at least 25 people, millions of animals, and destroyed more than 2,000 homes. The Australian wildfires were not only a national tragedy but also a clear warning to the world about the growing impact of climate change on human life, wildlife, and the environment.
Data Content
This data set includes latitude and longitude coordinates, scan and track pixel sizes, confidence, and brightness measurements, interval data for acquisition time, ordinal data for acquisition date, and nominal data for satellite and instrument type. Scan pixel and track pixel sizes describe the approximate width and length of each detection made by the satellite. These measurements show the areas where heat was detected but not the exact boundaries of the wildfire. This data can be used to track the spread of the 2019 Australian fires over time. Accordingly, the data can be used in reference to other geographical data to determine the impact of the fires on different communities and wildlife ranges. However, because that information is missing from the dataset itself, this data cannot illuminate those phenomena on its own.
Original Sources
The initial data and imagery sources from the Kaggle dataset come from NASA’s FIRMS (Fire Information for Resource Management System) MODIS and VIIRS Fire/Hotspot systems. FIRMS provides nearly real-time, accurate fire and thermal irregularity data from satellites. MODIS data stems from Terra and Aqua satellites, while VIIRS 375m uses data from the Suomi NPP and NOAA-20 satellites. MODIS and VIIRS complement each other in fire detection but have different spatial resolutions and detection capabilities. Due to this, they may record different observations for the same fire and produce different results. Although Kaggle presents the data, it is not the original producer; instead, it hosts a cleaned-up, packaged version of the data compiled by Carlos Paradis.
Information Left Out
The data captured by the MODIS C6 satellite and the VIIRS 375m satellite are limited to what they can observe from space in real time, such as longitude, latitude, acquisition date and time, and brightness. Because these satellites utilize thermal sensors to detect wildfire acquisition, anything that produces detectable thermal radiation will also be captured as data. Thus, information regarding what caused the thermal radiation or what caused the fire is a crucial piece of data that is left out, as it can be used to distinguish each heat signature. A detected anomaly also does not necessarily represent an uncontrolled wildfire. Controlled burns, industrial activity, or other heat producing events may also cause a detection. The confidence measurement helps indicate the likelihood of a detection referring to a wildfire, but it does not identify the true source of the heat. Furthermore, potential wildfire data often goes undetected due to cloud cover, the inability to detect small, low-heat-signature fires, and the fact that satellites can only pass over a specific area a few times a day. Information regarding human and ecological impacts, such as homes and ecosystems destroyed and populations displaced, is also left out and is rather reduced to a series of numerical measurements that focus on what can be seen from space, instead of what is actually experienced on the ground.
Funding of the Data
The Kaggle dataset titled Fires from Space: Australia does not specify a specific funder for its creation. It was compiled and uploaded by a Kaggle user named Carlos Paradis, who repackaged publicly available satellite data rather than generating new data funded by a grant or institution. However, the actual fire-detection data comes from NASA’s Fire Information for Resource Management System (FIRMS), which draws on data from the MODIS (Moderate Resolution Imaging Spectroradiometer) and VIIRS (Visible Infrared Imaging Radiometer Suite) satellite instruments. In short, while the dataset was not funded by a grant or institution, the satellite information is built and funded through government agencies such as NASA and NOAA.
Ideological Effects of the Ontology
The dataset’s ontology frames the Australian Wildfires as a detection problem rather than a social analysis problem. By using categories such as brightness, instrument, and confidence, the dataset presents fires as satellite-detected hotspots rather than personal, social, and environmental stories. As a result, this framing pushes questions of social responsibility and interpretation outside of the data. This ontology also gives the dataset a sense of scientific neutrality, even though measuring heat signatures rather than causes and experiences already biases the conclusions that can be drawn. By doing so, it privileges information that satellites can measure, ignoring experiences such as fear, cultural loss, or other social factors.
If this dataset were the only source, we would lose context about which communities were most severely affected, how state policy contributed to the fires, how emergency planning affected the disaster, or how climate change played a role. The severity of a fire could be misinterpreted: how could we differentiate a fire that burned through an unpopulated area versus a densely populated Aboriginal reserve? The dataset strictly shows that a fire happened and was detected; It does not explain why it happened or who was affected most by it.