ABSTRACT

Forest disturbances have been altering the ecological properties of ecosystems; meanwhile, disturbance events of varying sizes create different structures and functions for a forest landscape. Therefore, size and frequency are important attributes of disturbances, and their relationship should be studied. We present a hierarchical method through the modeling of the overall trend of the size–frequency distribution and the characterization of the non-constant variances of disturbance sizes at each frequency level. This method was demonstrated to accurately model the sizes as well as the corresponding frequencies; thus, the total disturbed area and number of disturbance patches were both accurately estimated. By applying the method to 13 provinces in China, consistent patterns were revealed by the modeling results and remote-sensing-based product, showing that between 2000 and 2005, forests in most provinces were dominated by moderate disturbances (10 ha < size < 100 ha). Southeastern provinces contain the largest proportion of small disturbances (size < 10 ha), whereas most of the very large disturbances (size > 1000 ha) occurred in the northeastern and northwestern provinces. This study concludes that the proposed method can improve the representation of the size–frequency distribution of forest disturbances.