A dynamic vegetation model was adapted to a mixed forest of deciduous broad-leaved trees and evergreen conifers in Hokkaido, the northernmost main island of Japan, located within a transition zone between cooltemperate and sub-boreal forest ecosystems. In this forest, elevation and a terrestrial-wetness index affect tree biomass and the percentage of conifers. We maximized agreement between this observed pattern and model output by calibrating parameters that control drought and excessive soil-moisture tolerance, establishment rate, and background mortality rate. In the simulation, biomass increased as a single peak curve with simulation year, while the percentage of conifers increased until the end of 200 simulation years.The 75-year simulated forest was most comparable to the adapted forest, consistent with the average frequency of catastrophic storm disturbances in Hokkaido. The model also reconstructed a reasonable succession pattern based on temperature and soil moisture. Therefore, the model mechanistically reconstructed the mixed forest via spatial niche segregation and succession after catastrophic disturbances. However, the model did not reconstruct the percentages of broad-leaved trees and conifers of forests prior to the disturbance in independent validation plots, demonstrating that additional processes should be considered in future models.