Disturbance types play a key role in post-disturbance vegetation recovery in boreal forests of Northeast China

• Integration of various spectral indices can detect and attribute forest disturbance. • Disturbance traits affect vegetation recovery more than climate and topography. • Climate variables in one year after disturbance more affect vegetation recovery. Understanding the patterns and effects of disturbances is important to predict post-disturbance vegetation recovery trajectories and inform adaptive recovery strategies. The inclusion of disturbance type into influential variables, and then to identify and rank the main drivers for vegetation recovery deserves further attention. Utilizing the multiple spectral indices, together with the Landsat-based Detection of Trends in Disturbance and Recovery (LandTrendr) algorithm, we integrally detected and attributed forest disturbances into fire, harvest, and urbanization (forest converted to urban, rural housing, road, etc.) across the Great Xing’an Mountains (GXM) of Northeast China from 1986 to 2022. By integrating disturbance type, disturbance severity, climate, and topography variables, we employed a random forest (RF) model to explore their impacts on short-term post-disturbance vegetation recovery. The recovery was quantified as the percentage of normalized burn ratio (NBR) recovery five years after disturbance. Our findings indicated that approximately 3.4 × 10 6 ha (SE = 1.4 × 10 5 ha) of forest was disturbed, accounting for 46.6 % (SE = 2.1 %) of the total forested area in the GXM. Fire disturbance constituted the largest share at 49.4 % (SE = 0.4 %), followed by harvest disturbance at 45.3 % (SE = 0.4 %) and urbanization disturbance at 4.8 % (SE = 0.3 %). The RF model revealed that disturbance type exerted the most significant influence on vegetation recovery, followed by disturbance severity (dNBR), summer Palmer drought severity index one year after disturbance (PDSI_summer_1), elevation, and summer temperature one year after disturbance (T_summer_1). Fire disturbance exhibited the most rapid vegetation recovery, followed by harvest disturbance, while urbanization disturbance displayed the slowest recovery, aligning with expectations. Vegetation recovery demonstrated both concave and convex relationships with dNBR and elevation, respectively. Generally, faster vegetation recovery corresponded with higher PDSI_summer_1 and T_summer_1. Our study provides baseline information on forest disturbances and disturbance types, which is valuable for the management of forest resources. Our study also underscores the key role of disturbance type in shaping vegetation recovery, which is in favor of identifying poor forest recovery areas and implementing targeted recovery management strategies.