Decrease in forest above-ground biomass in war-damaged forests of Ukraine: A case study using GE DI, Sentinel-2 data, and the GEE platform

Autorzy

• Ihor Kozak 1 The John Paul II Catholic University of Lublin, Al. Racławickie 14, 20-950 Lublin, Poland, ihor.kozak@kul.pl
  2 Vasyl Stefanyk Precarpathian National University, Shevchenko 57, 76000 Ivano-Frankivsk, Ukraine
• Piotr Kociuba The John Paul II Catholic University of Lublin, Al. Racławickie 14, 20-950 Lublin, Poland
• Myroslava Mylenka Vasyl Stefanyk Carpathian National University
  Shevchenka 57, 76000 Ivano-Frankivsk, Ukraine
• Victoria Gniezdilova Vasyl Stefanyk Carpathian National University
  Shevchenka 57, 76000 Ivano-Frankivsk, Ukraine
• Nadiia Riznychuk Vasyl Stefanyk Carpathian National University
  Shevchenka 57, 76000 Ivano-Frankivsk, Ukraine

Abstract

Armed conflict leads to significant disturbances in forest ecosystems, resulting in structural degradation and biomass loss. In eastern Ukraine, forested areas have been directly affected by military activities since 2014, with intensified disturbances after 2022. This study assesses changes in forest Above-Ground Biomass (AGB) in selected waraffected areas of Luhansk Oblast using spaceborne LiDAR data from the Global Ecosystem Dynamics Investigation (GEDI) and multispectral imagery from Sentinel-2 processed within the Google Earth Engine (GEE) platform. GEDI Level 4A biomass products were used as reference AGB data to train regression models based on Sentinel-2 spectral bands and vegetation indices. Separate models were developed for three forest polygons (Kuzmyne, Metolkine and Bobrove test sites). The results show strong relationships between GEDI reference AGB and Sentinel-based predictions (R² = 0.715 for Kuzmyne, 0.704 for Metolkine and 0.720 for Bobrove test sites). The models demonstrate consistent biomass underestimation in high-AGB stands, suggesting structural simplification in damaged forests. The study confirms that combining GEDI and Sentinel-2 data provides a reliable approach for biomass monitoring in conflict-affected regions where field measurements are unavailable. Further analysis showed that average Normalized Difference Vegetation Index (NDVI) values decreased from 0.67–0.71 in 2020 to 0.22–0.24 in 2022 and remained at similarly low levels (approximately 0.24) until 2025 at the Kuzmyne and Metolkine test sites. In contrast, a decrease in NDVI was already detected in 2021 at the Bobrove test site. This early decrease is associated with the shelling of the region by Russian-backed armed groups since late 2020, which caused large-scale forest fires and extensive destruction of vegetation. Overall, this study highlights the significant potential of integrating GEDI and Sentinel-2 data to reliably estimate forest AGB under extreme conditions caused by military conflict. The proposed approach is a valuable tool for monitoring and assessing forest ecosystem degradation in hard-to-reach or high-risk regions and can contribute to planning for forest ecosystem recovery after war.

Keywords

• loss of above-ground forest biomass
• Russian aggression in Ukraine
• Global Ecosystem Dynamics Investigation
• Normalized Difference Vegetation Index
• Google Earth Engine
• Sentinel-2