PREDICTION OF EELGRASS RECOVERY AFTER TSUNAMI BASED ON THE CASE OF THE 2011 GREAT EAST JAPAN EARTHQUAKE AND TSUNAMI

Abstract

 A prediction model for the eelgrass recovery after a tsunami was constructed using machine learning, as the model has not yet been established. Its usefulness in considering countermeasures was also investigated. The objective variable of the model is the rate of change in the eelgrass area relative to the area before a tsunami in each year from 1 to 7 years after a tsunami, and the explanatory variables are the eelgrass area before and immediately after a tsunami and information on the marine environment. A prediction model was constructed from information on the eelgrass distribution in Matsushima Bay after the 2011 Great East Japan Earthquake and Tsunami, using regression with a neural network to express the recovery trend. The model was applied to Ago Bay, Mie Prefecture, to predict the recovery of eelgrass. As a result, the relative recoverability of each location against future tsunamis was obtained. The result indicates that recovery prediction can be useful information in considering effective and strategic countermeasures before and after a tsunami.