The use of spectral indices in environmental monitoring of smouldering coal-waste dumps

This study demonstrates that while satellite data can reconstruct the fire history of coal-waste dumps, UAV-based remote sensing combined with field validation is essential for effectively monitoring small-scale thermal anomalies and assessing vegetation stress caused by smouldering in post-industrial environments.

The Gist

Imagine a giant, man-made hill in Chorzów, Poland, made of leftover coal waste. For years, this hill has been secretly "cooking" from the inside. It's not a roaring bonfire you can see from space; it's a slow, smoldering burn deep underground, like a slow-cooking pot that never turns off. This hidden heat is the main character of this study.

The researchers wanted to figure out how to "see" this invisible fire and understand how it affects the plants growing on top of the hill. They used two main tools: satellites (like a bird's-eye view from very high up) and drones (like a hawk flying just above the trees).

Here is the story of what they found, explained simply:

1. The "Thermometer" Problem: Satellites vs. Drones

Think of the satellites as a person looking at a city from a plane. They can see the big picture and tell you, "Hey, that whole neighborhood is a bit warmer than usual." They successfully tracked this coal hill's "fever" going back to 1999, showing that the fire has been slowly moving around for decades.

However, satellites are like looking at a mosaic from far away; the individual tiles (pixels) are too big to see the small details. The fire spots on this hill are small and shift around. The satellites were too blurry to see exactly where the fire was or how it was changing the plants right next to it.

Enter the drones. These are like a detective with a magnifying glass. Flying low, they captured incredibly sharp images (seeing details as small as a coin). This allowed the researchers to see the exact edges of the fire and how the plants were reacting to the heat right under their feet.

2. The "Winter Garden" Mystery

The most surprising discovery was about the plants. Usually, in Poland, when winter hits and the air gets freezing, plants go to sleep (dormancy). They turn brown and stop growing.

But on this coal hill, the underground fire acts like a hidden radiator.

• The Trap: In the autumn and winter, the air is cold, but the ground under the plants is warm because of the fire.
• The Result: The plants get confused. They think it's still summer! They keep their green leaves and keep growing while everything around them is brown and dead.
• The Twist: This isn't a sign of a healthy, happy garden. It's a sign of stress. The plants are being tricked by the heat. They are burning the candle at both ends, trying to grow in conditions that are actually too harsh.

The researchers found that if you just looked at the "greenness" (using a standard tool called NDVI), you might think the plants were thriving. But in reality, they were struggling against the heat and toxic soil. It's like seeing a person sweating in a heavy coat and thinking they are exercising, when they are actually just overheating.

3. The "Green Zones" Map

The study mapped out the hill like a weather map, but for heat and plants:

• The Fire Core: The hottest part. No plants can survive here; it's a barren, dead zone.
• The "Green Ring": Just outside the fire core, the heat is warm but not deadly. This is where the "winter garden" happens. Plants grow here year-round, creating a strange green halo around the dead center.
• The Normal Zone: Farther away, where the fire's heat doesn't reach, plants follow the normal rules: green in summer, brown in winter.

4. The Takeaway

The main lesson from this paper is that context is everything.

• Satellites are great for seeing the big history of the fire (where it started, how it moved over 20 years).
• Drones are necessary to see the small, dangerous details (exactly where the fire is now and how it's hurting the plants).
• The Warning: Don't just trust the "greenness" of a plant. A plant can look green and healthy because it's being forced to grow by a hidden fire, even though the soil is toxic and the ecosystem is damaged.

In short, the researchers used high-tech eyes to prove that this coal hill is a place where nature is being tricked by a hidden fire, and to understand the truth, you need to get close enough to see the details that satellites miss.

Technical Summary

Technical Summary: Monitoring Smouldering Coal-Waste Dumps Using Spectral Indices

Problem Statement
Subsurface fires in coal-waste dumps, particularly within the Upper Silesian Coal Basin, present significant environmental and safety challenges. These smouldering fires are difficult to detect early due to their hidden nature and often require costly, complex interventions. While remote sensing has become a standard tool for environmental monitoring, conventional satellite imagery often lacks the spatial resolution necessary to detect small-scale thermal anomalies and subtle vegetation stress patterns associated with these localized fires. Furthermore, the application of standard vegetation and fire indices to subsurface fire environments is not straightforward, as the thermal dynamics differ significantly from surface forest fires. The study addresses the need to evaluate the applicability of environmental spectral indices for monitoring these specific post-industrial degradation processes and to determine the optimal scale of observation (satellite vs. UAV).

Methodology
The research employed a multi-method approach combining field surveys, Unmanned Aerial Vehicle (UAV) campaigns, and satellite data analysis at a coal-waste dump in Chorzów, Poland.

• Data Acquisition:
  • Satellite: Historical and recent data from Landsat (4/5 TM, 7 ETM+, 8 OLI/TIRS) and Sentinel-2 were utilized to reconstruct fire history and analyze seasonal trends. Landsat data provided thermal context, while Sentinel-2 offered higher spectral resolution for vegetation indices.
  • UAV: Two survey campaigns (Autumn 2023 and Spring 2024) were conducted using a DJI Matrice 210 v2 RTK (thermal + RGB) and a DJI Mavic 3 Multispectral (Green, Red, Red-Edge, NIR). These provided high-resolution imagery (approx. 3 cm/pixel for optical, 20 cm/pixel for thermal).
  • Field Validation: Ground truthing included direct subsurface temperature measurements (using a 30 cm probe and pyrometer), vegetation mapping across four distinct zones (fire spot, death zone, transition zone, inactive areas), and meteorological data collection.
• Analysis:
  • Vegetation and fire indices were calculated using QGIS and ArcGIS. Key indices included NDVI, GNDVI, GRVI, SAVI, EVI, and the Burn Area Index (BAI).
  • The study focused on correlating index values with thermal activity zones and analyzing seasonal variability in vegetation response to subsurface heating.

Key Results

• Satellite Limitations: Satellite data (Landsat and Sentinel-2) successfully reconstructed the fire history, revealing thermal activity dating back to at least 1999. However, the spatial resolution (30m for Landsat, 10-20m for Sentinel-2) was insufficient for detailed monitoring of small-scale thermal anomalies. The contrast between active and inactive zones was often blurred in satellite imagery compared to UAV data.
• UAV Diagnostic Potential: UAV-based multispectral and thermal imagery provided high-resolution insights into the spatial dynamics of the fire. The data clearly delineated the thermally active zone, which forms an arc around the dump's center, and captured fine-scale variations in vegetation stress.
• Vegetation Anomalies and Seasonality: A critical finding was the disruption of natural vegetation cycles in thermally active zones.
  • Winter Greening: In moderately heated root zones, vegetation exhibited "winter greening," where plants remained active or green during meteorological winter due to elevated subsurface temperatures. This resulted in positive NDVI values in areas where surrounding vegetation was dormant.
  • Summer Stress: Conversely, during summer, the same zones showed signs of stress or dieback due to excessive heat and reduced soil moisture, leading to lower NDVI values despite the presence of vegetation.
  • Index Performance: Standard vegetation indices (NDVI, EVI, SAVI) showed consistent patterns of low values in the fire core but revealed complex, non-linear responses in transition zones. The Burn Area Index (BAI) proved effective in distinguishing charred/low-vegetation areas (values >200) from vegetated zones, serving as a viable alternative to NBR when SWIR bands were unavailable at high resolution.
• Fire Migration: Sentinel-2 data from 2015 to 2024 indicated a gradual north-eastward migration of the fire front, suggesting the thermal activity is slowly approaching recreational areas of the site.

Significance and Claims
The paper asserts that while satellite data is valuable for long-term historical reconstruction of subsurface fires, it lacks the resolution required for detailed, operational monitoring of small-scale thermal anomalies in coal-waste dumps. The study highlights the diagnostic potential of UAV-based remote sensing in post-industrial environments, offering the necessary spatial detail to identify specific thermal zones and vegetation stress patterns.

Crucially, the authors caution against the automatic interpretation of high NDVI values as indicators of ecosystem health in these contexts. The study demonstrates that "greening" in thermally active zones during winter is a stress response driven by artificial microclimates (a "mini-heat island") rather than successful reclamation. This phenomenon can mask underlying ecological degradation, such as soil contamination or moisture deficits. Therefore, the paper concludes that accurate environmental assessment of smouldering dumps requires high-resolution UAV data combined with field validation to correctly interpret spectral indices and avoid misleading conclusions about site recovery.