Remote sensing of surface fires in boreal forests

Forest fires threaten carbon storage but are vital to boreal ecosystem dynamics. While crown fires are well-studied, low-intensity surface fires, common in Fennoscandia, are less understood. This thesis used remote sensing techniques to examine surface fires across eight Scots pine-dominated test sites (~1 ha each) in southern Finland, with controlled burnings simulating surface fires. Terrestrial laser scanning (TLS) reconstructed forest structure before and after these fires for change detection. Study I utilized bitemporal TLS to identify burned areas and estimate volumetric changes in ground vegetation. A surface differencing-based classification method was developed, achieving high accuracy (recall, precision, F1-score = 0.9). On average, 85% of the test site areas were burned, with a mean reduction in ground vegetation volume of 1200 m3/ha, though variability was observed. Study II examined the effects of ground vegetation on TLS-derived digital terrain models (DTMs) and tree/forest attributes. In burned areas, post-fire DTMs averaged 10 cm lower than pre-fire DTMs, with greater changes and root mean square differences compared to unburned controls. A 10 cm overestimation in DTMs led to underestimates in tree/forest attributes: 1.3 mm (0.6%) in diameter at breast height, 4.8 dm3 (3.1%) in stem volume, and ~3 m3/ha (1.3%) in total stem volume. Study III assessed the normalized burn ratio (NBR) index from Sentinel-2 data for detecting surface fires. Breakpoint analysis identified most fires, with undetected cases linked to sparser vegetation loss and denser canopy cover. A moderate negative correlation (r = –0.5) was found between NBR changes and TLS-derived volumetric changes in ground vegetation. Variations in NBR were explained by vegetation changes, canopy cover, and site conditions (R2 = 84%). This thesis demonstrates the potential of remote sensing to identify surface fires and quantify their effects on ground vegetation, supporting method development and advancing understanding of their role in boreal forest ecosystems.