Ecological consequences of armed conflict: vegetation change in Ukraine from remote sensing and conflict intensity modeling

Abstract

Armed conflict generates extensive and long-lasting environmental consequences, yet large-scale quantitative assessments of these impacts remain limited. This thesis examines the ecological effects of the Russia–Ukraine War by integrating remote sensing time-series analysis with spatial Bayesian modeling to quantify vegetation change and link it to patterns of military activity. Using 22 years (2001–2023) of MODIS Enhanced Vegetation Index (EVI) data, the Breaks for Additive Season and Trend (BFAST) methodology was applied to detect long-term vegetation trends and structural disturbances across Ukraine. The BFAST analysis revealed that approximately two-thirds of the country exhibited positive vegetation trends while localized but substantial negative trends were concentrated in eastern and southern regions experiencing the most intense military activity. Additional analyses show that cropland and urban areas significantly influence national patterns and may mask conflict-related ecological signals. Explosion events recorded in the ACLED database (2018–2023) were modeled as a spatial statistical process using Bayesian Integrated Nested Laplace Approximation (INLA) with a stochastic partial differential equation (SPDE) framework, demonstrating that explosion intensity was strongly clustered near infrastructure, population centers, and front-line regions, with spatial dependence extending approximately 10 km. Predicted explosion intensity surfaces were incorporated into a generalized additive modeling framework to evaluate relationships between conflict intensity, climatic variables, and vegetation trends, indicating that higher modeled explosion intensity is associated with more negative vegetation trends. Together, these findings represent the first national-scale application of BFAST to assess vegetation dynamics across Ukraine and the first spatial modeling of explosion intensity counts using Bayesian methods, advancing methodological approaches for studying environmental impacts in active war zones and providing a replicable framework to support post-conflict environmental assessment, ecological risk analysis, and long-term recovery planning.