Can AI save our forests?

More than 200 forest fires were reported in 2022 alone

Can AI save our forests?


P

akistan is an arid country with just over 5 percent of the land area covered by forests, compared to the global average of 33 percent. An estimated 15,000 to 20,000 hectares (37,000 to 49,000 acres) of woodlands are lost annually due to an alarmingly ineffective forest fire management system reliant on traditional methods that are no match for the escalating menace of wildfires.

Almost 18,000 forest fire events were reported across the country from 2001 to 2019. 2022 was particularly devastating as 200 fires ravaged forests spread over thousands of acres across Khyber Pakhtunkhwa, Balochistan and the Punjab, destroying 14,430 acres of forests and pastures.

That year, Balochistan witnessed the massive Sherani Forest blaze that raged for 13 days, burning down an estimated 900,000 olive and pine nut (chilghoza) trees. The blaze spread across the KP-Balochistan border and could not be put down by domestic attempts. Assistance was then sought from Iran to fly missions with specialised aircraft to douse the flames. The impact of the fire was estimated to be a staggering Rs 4 billion. It wrought havoc on the local economy dependent on chilghoza which takes anywhere between 25 and 40 years to fruit.

The Chamba Forest, spread over hundreds of acres in the Punjab was similarly reduced to a cinder. Fires in the Margalla Hills forests have become almost an annual feature. 76 of those reported in the 1990s, 187 in the 2000s and 264 in the 2010s.

According to Global Forest Watch, forest fires are the biggest driver of Pakistan’s deforestation. These are caused mainly by human mischief and negligence. The arsonists are suspected to have caused 50 percent of the damage to Margalla forest spread across 200 square kilometers during 2017-18 period.

Forest officials say budget constraints severely limit the human resources that emergency response agencies and woodland management authorities can employ to contain fire. “Human resources are lacking across the board. From insufficient patrol officers on the ground up, there is not enough staff at any level.”

Human irresponsibility and environmental factors are important causes of the wildfires, says Syed Mehmood Nasir, a former inspector general of forests. He says climate change, temperatures, wind speed, duration of dry spells, distance from roads and urban areas strongly influence the spread of fires.

Nomads too start small fires to obtain good grass produce or to scare off wild animals. Cooking or leaving burning cigarette butts can also cause accidental ignition that sometimes gets out of control.

While most fires have been low-intensity ground and bushfires, the risk of high-intensity fires is increasing with little to boast in terms of the fire extinguishing capacity. Furthermore, ignoring the precautionary measures of removing the fuel load (pine needles and pine cones) from the ground causes a build-up of fire risk.

The failure to maintain “forest fire lines” to prevent the spread of fire from one area to another is a serious violation of standard safety practices. This can turn deadly as the climate becomes increasingly hot and dry.

The communities living within the forests of KP and Margalla Hills National Park play a critical role in the current forest management system. Most likely to be first responders to a fire, the community members use conventional techniques to extinguish fires, often risking their lives.

A project funded by the UK is being run by LUMS and the WWF to test an AI platform to detect forest fires early and predict risks in a forest in Khyber Pakhtunkhwa. Earlier in a pilot project, an AI system, named FireGuard was trained on historical data and used successfully to detect a potential fire risk in Margalla Hills and the Punjab.

Last week, a forest fire in Attock district was brought under control by the efforts of the Forest Department staff and the local community. Hanif Parwana, a local, told The News on Sunday: “Our entire families, especially women, were engaged in extinguishing the fire, spread over two kilometers. They brought water from a distance in pans and containers over their heads… The losses are heartbreaking. We believe that our existence and that of our cattle are intrinsically linked to a flourishing forest,” he said.

Efforts to combat forest fires largely employ traditional methods, including bush beating by patrol officers and community members, who often lack modern equipment or firefighting training. While these traditional methods can be effective for small fires, they prove inadequate once a blaze spreads.

In case of recent fires in Margalla, fire fighters had difficulty bringing water bowsers close to the fire locations. CDA Chairman Irfan Niazi said that it was difficult to transport heavy machinery or trucks to mountainous terrains where the fires erupted. “We transport firefighters in our vehicles but then they have to walk miles to reach the fire site(s),” Niazi said.

Mehmood Nasir advocates for the integration of artificial intelligence in forest fire management systems. By leveraging its capabilities to prevent fires, artificial intelligence platforms can predict the likelihood of fire incidents, as well as forecast their spread and intensity. Additionally, AI-driven image analysis enables swift fire detection, even in remote and underserved areas.

Nasir says with limited resources, risk assessment, detecting, responding and recovering from fires is often not thorough. Human patrols can’t monitor everything. This causes delayed responses when time is crucial. He says AI can help the system overcome these challenges, making it easier to manage forest fires and reduce the risks. Turkey, Nepal, Ethiopia and Australia, he says, are using AI platforms for this purpose.

A project funded by the UK is currently being run by LUMS and the WWF to in a forest in the KP to test an “AI platform to detect forest fires early and predict risks.”

Earlier in a pilot project, an AI system, named FireGuard was trained on historical data and used successfully to detect potential fire risk in Margalla Hills and the Punjab. It detected a potential fire risk three days before a blaze broke out in Margalla Hills, allowing firefighters to create firebreaks, conduct controlled burns and deploy resources strategically, containing the fire before it coud spread. The FireGuard was also deployed at the Punjab’s Chamba Forest where it detected a small fire early, enabling firefighters to extinguish it before it could spread dangerously.

As Pakistan struggles to come to terms with the devastating impact of wildfires on its forests, use of AI can revolutionise its forest fire management system. With adequate funding and partnerships in place, the country can safeguard its precious forests for future generations.


The writer is a senior assistant editor at The News, Karachi. He can be contacted at: amjad.siddiqi @thenews.com.pk

Can AI save our forests?