Differentiating effects of salvage logging and recovery patterns on post-fire boreal forests in Northeast China using a modified forest disturbance index

Abstract

ABSTRACT Forests are resilient to a range of disturbances, but combinations of severe natural and anthropogenic disturbances (e.g. wildfire and logging) may inhibit forest recovery and lead to forest degradation. Recent studies have explored long-term forest-disturbance detection and forest-recovery dynamics by using free and open-access remote-sensing images. However, mapping consecutive multiple disturbance agents is challenging using existing automated change-detection algorithms because the reduced canopy reflectance and the smoothing of consecutive disturbance signals mean that the initial disturbance cannot be spectrally separated from the second disturbance. Furthermore, uncertainty remains about post-disturbance vegetation dynamics and the effects of forest recovery under the interaction of burn severity, biological-legacy management, and active forest restoration (i.e. artificial regeneration and assisted natural regeneration). This contributes to biases in long-term forest-recovery monitoring, which are not conducive to the guidance of post-fire vegetation recovery. Here, we propose a modified disturbance index to separate the spectral characteristics of fire and forest logging using normalized tasseled-cap components (brightness and wetness) and detect the spatiotemporal distribution of post-fire logging by means of an index threshold and image differencing. On this basis, the recovery patterns of the post-fire forest are differentiated by considering the cumulative effect of fire, post-fire logging, and recovery approaches. The method is tested in the burn areas of the 5.6 Fire in the Greater Hinggan Mountain area (the largest forest fire in recorded history in China), giving an overall accuracy of 85% in post-fire forest logging mapping. Our results confirm that biological legacies (i.e. trees, logs, and snags) were removed across many areas in the fire, with activities peaking in the second year after the fire and located chiefly in areas of moderate and high burn severity. By identifying post-fire logging, the fluctuation and high disturbance index of the conventional temporal trajectory in the early stage of forest recovery are explained. The large-scale salvage logging slowed the recovery of the post-fire forest ecosystem and influenced the recovery process through the interaction of burn severity and active forest restoration. In areas of high burn severity, assisted natural regeneration (i.e. natural regeneration with artificial aids such as clearing the snags, weeding, digging pits, and supplemental planting) preserved biological legacies and achieved a higher proportion of forest recovery, second only to non-logged areas where natural regeneration (i.e. forestry recovering naturally without any artificial intervention) occurred. By contrast, salvage logging followed by artificial regeneration (i.e. clear-cutting all dead or damaged trees followed by tree planting or artificial seeding) hindered vegetation recovery in the early stage, but it improved the recovery rate in years 10–20 and approached the recovery proportion of non-logged areas where natural regeneration occurred as the recovery progressed and habitat conditions improved. The proposed method is shown to offer important advantages in detecting post-fire salvage logging, and it provides improved guidance for forest managers in developing strategies for forest recovery.