Developing automated detection and monitoring of peat fires in Indonesia with thermal infrared sensors under drones

Climate change presents the single biggest threat to sustainable development across the world. Its widespread, unprecedented impacts disproportionately burden the poorest and most vulnerable. The Intergovernmental Panel on Climate Change concluded it is extremely likely that human influence has been the dominant cause of climate change since the mid-20th century. The global rise in atmospheric concentration of CO2 has been a major driver in the change of climate.

Scientists estimate that peat forest fires can contribute up to 15-20% of all yearly global anthropogenic CO2 emissions -- more than the global transport sector. Currently the severe (dry) conditions leading to these extreme fires occurs roughly every decade, but with climate change their frequency may increase. The annually re-occurring peat forest fires in Indonesia represent a substantial fraction of the global anthropogenic CO2 emissions. Efforts are underway to find and contain the fires, but the challenges faced in achieving this are considerable. For example, to completely extinguish a peat fire with an area of only 1 square meter requires 200-400 litres of water(!). Identifying the precise location of fires when they are small and more easily contained and extinguished is therefore vital. Once a peat fire has become established, it can burn for weeks and is incredibly difficult to extinguish, leading to large areas of forest loss, plantation and property damage, huge carbon emissions, enormous economic losses, and prolonged exposure of local people to toxic haze pollution. This haze, plus the sometimes thick vegetation at fire sights and the fact that peat fires can burn underground, means fire-fighters frequently struggle to ascertain where best to focus their efforts to control the fires, in addition to creating health and safety issues for fire-fighting team members.

Working closely in collaboration with key stakeholders in Indonesia, our pilot study in July 2018 showed that our drone plus thermal infrared camera system can easily identify hot spots in peat fires burning both underground and through the smoke -- current fundamental bottlenecks in stopping the fires. We now propose to use GCRF funding to hire a PDRA to develop our system to achieve automated, real-time detection of above- and below-surface fires on the drone, so that real-time fire detection and monitoring in the field can be achieved. We will train local NGOs, Government partners and local fire-fighting teams in the usage of the system, to facilitate local uptake of the technology and deliver technological knowledge transfer, thus improving longer-term local fire-fighting capacity and reducing the negative impacts associated with these fires. At the end of the funding we aim to: (i) have demonstrated the system's use in the field during two full fire seasons; (ii) have quantified a significant increase in firefighting efficiency by statistically comparing the extent and duration of the fires tackled with/without the system; and (iii) estimate the resulting long-term reduction in atmospheric CO2 emission and haze (which we estimate will be a factor >10). We will then be in a position to apply for future funding to take the system to fire zones around the world, concentrating on those with the biggest global impact on increasing atmospheric CO2 levels and ill health due to toxic haze.

Forest fires in Indonesia and particularly on peatlands are a huge, annually re-occurring problem for development in the region. In severe years they lead to economic losses for tens of millions of Indonesian people amounting to USD billions (USD 16.1 billion in 2015 alone). The World Bank estimates that tens of millions of people suffer health problems from these fires with associated economic costs, and a peer-reviewed study of the 2015 fires estimated 69 million people were exposed to smoke for 3 months and 100,300 deaths arose from these fires. In addition to the human tragedy this represents, losing so many lives represents a huge long-term economic loss for the country. The high carbon content of peat makes it one of Nature's most efficient carbon stores, with the result that peat fires are a major contributor to global warming. In severe years, peat forest fires contribute up to 13-40% of all yearly global anthropogenic CO2 emissions, thus greatly impacting climate change, which presents the single biggest threat to sustainable development everywhere.

Building on our previous experience in designing/using drones for ecology, we will develop a drone with combined thermal imaging and visual spectrum camera for detection of above- and below-surface fires through smoke. Our pilot study shows our drone system with thermal-infrared camera can detect the location of (even small) underground fires and humans through the smoke, overcoming the fundamental problems in fire-fighting. We are working with the key local stakeholders who have all the logistic networks in place to immediately implement the system to tackling real fires. The thermal footage will provide the location, temperature and size of each fire 20 times more quickly and much more reliably than traditional methods on foot, allowing fire-fighters to prioritise tackling the most serious (parts of fires) in order. We conservatively estimate that the factor ~20 improvement in hot spot detection efficiency per person, which essentially allows all person effort to be directed into containing and extinguishing the fires with maximum efficiency, will reduce the burn area by at least an order of magnitude, while simultaneously increasing fire-fighter safety. Scaled up, the resulting reduction of similar magnitude in CO2 emissions and toxic haze will have an immediate impact on atmospheric CO2 levels, and therefore climate change, and improve the health and long-term economic growth of the people in the region.