Analysis of wind damage caused by multiple tropical storm events in Japanese Cryptomeria japonica forests

This study analyzed wind damage caused by tropical storms from 1991 to 2007 to Japanese forests mainly consisting of Cryptomeria japonica. Statistical analyses based on logistic regression and Cox regression models were conducted in relation to conditions at the forest and stand level. Known damage to forests managed by Kyusyu Rinsan Corporation (KR forests), located on the Kyushu Island, was analyzed at the forest level, using tropical storm characteristics such as air pressure, precipitation and periods when the forests were within the storm zone as predictors. Wind damage was also examined at the stand level (150 analysis points) using Cox regression models, according to stand age, site index, terrain conditions, management practices and wind velocity indicators (horizontal and vertical velocity vectors). The results indicated that at the forest level, higher maximum hourly wind speed and longer periods of >15 m s^-1 of wind speed were significantly correlated to damage occurrence. At the stand level, indicators of upward vertical velocity, thinning treatments and site index were positively associated with the probability of wind damage. For instance, stands receiving higher upward vertical velocities and thinning treatment within 2 years were more likely to have reduced stability against tropical storms. Stands with higher and lower site index than average also showed lower stability in our analysis.