THE JAPANESE SOCIETY OF IRRIGATION,DRAINAGE AND RURAL ENGINEERING

Damage to agricultural land and agricultural facilities is becoming more severe due to the frequent occurrence of heavy rains and widespread areas. In large-scale disaster areas, national financial support is required by designating a heavy disaster, but it takes about one month after the disaster. However, early designation and judgment of expected designation are desired. In this research, a nonlinear multivariate estimation model was constructed by deep learning using past disaster data, rainfall data, stock data, etc., at the early stage of the heavy disaster. We estimated the damage amount of the disaster and examined the judgment prediction method of the prospect of heavy disaster designation.