New Zealand's landscapes are increasingly vulnerable to wildfires as climate change brings hotter, drier conditions and more frequent extreme weather events. A groundbreaking study co-authored by Dr. Helen de Klerk from Lincoln University highlights how advanced fire modelling techniques, proven effective overseas, can be adapted to predict wildfire behavior here, significantly boosting national resilience.

This research arrives at a critical time. Wildfire risks are intensifying across the country, with projections indicating extreme fire danger could nearly double by century's end in some areas, driven by warming temperatures and prolonged droughts. Urban expansion into rural, fire-prone zones further heightens threats to homes, communities, agriculture, and biodiversity. Lincoln University's contributions position it as a leader in addressing these challenges through innovative, data-driven solutions rooted in higher education research.

Understanding Fire Refugia and Their Importance in Wildfire Management

At the heart of Dr. de Klerk's study is the concept of fire refugia—persistent unburnt areas within fire-affected landscapes that serve as safe havens for flora and fauna. These zones are vital for ecosystem regeneration post-fire, protecting seed banks and wildlife during blazes. Traditional fire prediction models often overlook these micro-scale safe spots, but machine learning changes that by mapping them accurately over vast forested regions.

In New Zealand, where South Island terrains mirror those studied in South Africa's Southern Cape—steep mountains, undulating foothills, variable winds, and coastal influences—this modelling holds immense promise. Complex wind patterns around peaks create unpredictable fire spread, but by factoring in slope steepness, aspect (sun-facing vs. shaded), temperature, and wind direction, the models identify wind channels and natural barriers. "When landscapes burn, it's important to understand which areas may remain unburnt," Dr. de Klerk notes, emphasizing their role in safeguarding biodiversity.

This approach not only enhances firefighting strategies but also informs land-use planning, helping prevent development in high-risk zones and promoting resilient landscapes.

The Machine Learning Models: A Step-by-Step Breakdown

The study, titled "Predicting persistent forest fire refugia using machine learning models with topographic, microclimate and surface wind variables," employs ensemble and K-Nearest Neighbour algorithms. Here's how it works step by step:

• Data Collection: Gather topographic data (elevations, slopes), microclimate metrics (temperature, moisture), and surface wind patterns from satellite imagery and ground sensors.
• Variable Integration: Combine factors like aspect, which influences moisture balance and vegetation flammability, with predominant wind directions to simulate fire exposure.
• Model Training: Train machine learning on historical fire data from South Africa, validating against real burn scars to refine predictions.
• Refugia Mapping: Output high-resolution maps pinpointing unburnt zones, even in large-scale fires.
• Validation: Cross-check with field observations for accuracy in complex terrains.

Published in the International Journal of Geo-Information, the paper demonstrates superior performance over conventional methods, offering actionable insights for New Zealand's fire agencies.Read the full study

Lincoln University's Broader Fire Ecology Research Portfolio

Dr. Helen de Klerk's work builds on Lincoln University's robust fire research ecosystem. Her expertise in conservation fire ecology and spatial analysis spans GIS, remote sensing, and management-driven studies. Complementing the refugia model, PhD student Georgia Stevenson's experiments on beech forest litter flammability mark the first field-based ignition tests in New Zealand native forests. These reveal when damp litter transitions to self-sustaining wildfires, critical as seen in the 2024 Bridge Hill fire.

Additionally, Associate Professor Tim Curran's 12-year project has cataloged 470 low-flammability plant species using a "plant BBQ" ignition tester. This directory, partnered with Fire and Emergency New Zealand, guides defensible space planting around homes and farms. Explore the directory

A new PhD scholarship targets indigenous forest flammability, underscoring Lincoln's commitment to actionable science.

Photo by Marek Piwnicki on Unsplash

Climate Change Amplifying Wildfire Threats in Aotearoa

New Zealand faces escalating wildfire dangers. NIWA models predict more intense dry spells, with extreme fire weather potentially doubling in frequency. Urban fires could rise over 40% by 2100 under high-emissions scenarios, fueled by hotter temperatures. The 2026 National Climate Change Risk Assessment flags wildfires among top threats to economy and ecology.

Forest plantations and native bush, vital for biodiversity and timber exports worth billions, are at stake. Lincoln's models address this by pinpointing vulnerabilities, enabling proactive measures like strategic fuel reduction and community education.

Stakeholder Perspectives: From Firefighters to Policymakers

Dr. de Klerk presented at a local fire forum, sparking keen interest from practitioners. Fire managers value tools for resource prioritization during rapid fire spread, while conservationists see refugia as biodiversity lifelines. "Prioritising where to place resources is essential during fast-moving fire events," she highlights.

Landowners benefit from low-flammability planting advice, reducing insurance premiums and property losses. Policymakers can integrate findings into national strategies, aligning with Te Mana o te Wai principles for Māori land stewardship. Multi-perspective collaboration exemplifies higher education's role in bridging science and practice.

Practical Applications: Enhancing Firefighting and Land Planning

• Fire Operations: Identify defensible spaces for backburning and safe helicopter landing zones.
• Land-Use Planning: Zone developments away from high-spread areas, incorporating green firebreaks.
• Conservation: Protect refugia to accelerate post-fire recovery.
• Agriculture: Safeguard food production by modeling risks around farms.
• Urban Interfaces: Promote resilient planting via Curran's directory.

These applications promise economic safeguards, as wildfires cost millions annually in suppression and lost productivity. For aspiring researchers, Lincoln offers pathways in fire ecology.Explore higher ed jobs in environmental science.

Future Outlook: Scaling Up and Integrating AI

Adapting these models for New Zealand requires local data calibration, but similarities with South Africa suggest quick wins. Ongoing Lincoln projects, like beech flammability thresholds, feed into national fire danger outlooks. Future integrations with drones and real-time satellites could enable dynamic predictions.

New PhD opportunities signal growing investment. As climate pressures mount, Lincoln University's leadership ensures Aotearoa builds adaptive capacity. Students and professionals can contribute via NZ university opportunities or career advice.

Photo by Edgar Chaparro on Unsplash

Careers in Fire Ecology and Resilience Research

Lincoln University's fire research attracts top talent, offering roles in modelling, ecology, and policy. With demand surging, graduates secure positions in government agencies, Scion, and consultancies. Check university jobs, faculty positions, and research assistant jobs. For insights, visit Rate My Professor or higher ed career advice.

This study exemplifies how NZ universities drive real-world impact, fostering resilient futures.