When your town is burning, every second counts

Tracking fires at the urban-wildland interface with Planet Labs and Radiant Earth Foundation

By Glenn Moncrieff, Data Scientist, Fynbos Node, South African Environmental Observation Network (SAEON)

Cape Town is world renowned for its scenic beauty. Table Mountain looms large over the city, with the urban environment wedged between its steep cliffs and the oceans surrounding the Peninsula. These attractions draw millions of tourists every year and entice locals to build their homes in close proximity to the natural environment. But this proximity comes with substantial risks. The natural vegetation of the region — an evergreen shrubland locally known as ‘fynbos’ — is prone to regular, high severity fires.

Fynbos vegetation around Cape Town, South Africa. Photo: Adam Wilson.

These fires are a natural feature of the local ecology, with many plant and animal species depending on fire to complete their life cycle. Fires contribute to maintaining the spectacular biodiversity of this area. The Cape Floristic Region is home to almost 9000 plant species, making it the most diverse plant community outside of the tropics.

This setting of a fire-dependent ecosystem juxtaposed with human infrastructure requires careful management to ensure that natural processes are allowed to take place without risking lives and homes. Despite a range of preventative action, such as prescribed burns and attempts at fire-safe building and gardening practices, every fire season has numerous incidents of damage to property and harm to people.

In the early hours of the 1st of January 2019, in an ill-conceived New Year celebration, a flare was sent soaring through the night sky in the small seaside suburb of Betty’s Bay on the outskirts of Cape Town. The ensuing fire burnt for two weeks in the adjacent mountain ranges and neighboring towns. When the dust settled, dozens of homes had been destroyed or damaged and thousands of hectares of vegetation burnt, much of which was not yet mature enough to be exposed to fire.

Planet Labs operate a fleet of over 100 microsatellites, reimaging the world at an unprecedented frequency. Users from academic institutions receive free access to a vast image archive. Their daily imagery allows the progression of fires to be tracked closer than ever before, and the extent of damage to be assessed days after the occurrence of a fire. To analyze Planet images, traditionally I would use their online data explorer to visualize and select the images acquired at my desired date, time and location. Alternatively, I could search and download programmatically using their API’s. Then I would download the entire image to my local machine to visualize and analyze the data. But this workflow can be slow and require significant resources. Depending on the number of images you are downloading, the analytical process can take a significant amount of time and storage space. Once downloaded, it can be computationally expensive to perform operations on the high-resolution data provided by Planet.

Using Radiant Earth’s Platform, I was able to search and explore Planet’s entire images database, find the correct image for the January fire in Betty’s Bay within days of its occurrence and visualize the fire damage by viewing the image in False Color Infrared. When viewed in False Color Infrared living vegetation appears bright red, and bare earth black or brown; Comparing images of Betty’s Bay before and after the fire swept through clearly shows where homes and vegetation were burnt. Switching between True Color and False Color Infrared views using Radiant Earth is as simple as a single click.

Eastern edge of Betty’s Bay on January 6, 2019. Image: Planet Labs.
Eastern edge of Betty’s Bay on January 13, 2019. Image: Planet Labs

These images clearly show the extent of the damage and the progression of the fire. Blown by strong north-westerly winds, the fire spread through the surrounding natural vegetation, burnt through gardens and homes, and eventually extinguished at the coastline. Only the vegetation north of the mountain rimming the town remains unburnt. This visual analysis clearly shows the power of False Color Infrared imagery from Planet Labs to determine fire damage in a timely manner.

The cloud-based geospatial imagery exploration and analysis offered by Radiant Earth enables rapid insights into events such as wildfires without users requiring access to powerful computing resources. This increased accessibility of Earth observation data is contributing to better preparedness and improved management of the interface between the human and natural environments.