Applicability Of NBR And dNBR Indices In Assessment Of Pyrogenic Transformation And Post-Fire Forest Regeneration: Case Study Of Southeastern Siberia Coniferous Forests

This study evaluated the reliability of the Normalised Burn Ratio (NBR) and its differenced variant (dNBR) for assessing burnt areas and post-fire forest recovery. The research was conducted in the pine forests of the Tunka Depression (Southwestern Cisbaikalia), focusing on areas affected by a 2010 wildfire. Field data consisted of annual geobotanical observations from 2014 to 2022, which documented plant community regeneration across varying degrees of fire severity. Remote sensing analysis utilised Landsat 7 imagery (30 m resolution) between 2009 and 2022. Approximately 500 cloudfree NBR values were extracted from the USGS Landsat 7 Level 2, Collection 2, Tier 1 dataset using the Google Earth Engine platform. We assessed the spatiotemporal dynamics of these indices alongside the geobotanical parameters. The results confirm the effectiveness of NBR and dNBR for mapping burnt areas and determining initial fire severity. For detecting recent burns, mid-growing season imagery was most informative. Regarding post-fire regeneration analysis, springtime dNBR data were most reliable, as the influence of herbaceous cover on the spectral signal is minimised compared to the peak growing season. However, field observations revealed that the recovery of NBR values to pre-2009 fire levels by 2021 does not indicate that plant communities have regenerated to a near-natural state. This trend of rapid NBR recovery underscores the limitation of using NBR/dNBR indices alone for assessing long-term regeneration prospects. In conclusion, the results of the synthesised analysis of geobotanical and geoinformation materials showed that while remote sensing data effectively corroborate landscape-forming processes in disturbed ecosystems, their utility in detailed regeneration studies requires calibration with field data. The findings contribute to refining the application of NBR/dNBR indices and highlight the necessity of integrated approaches for calibrating remote sensing data.