Remote Sensing Based Model Induction for Drought Monitoring and Rainfall Estimation

摘要: Droughts are natural phenomenon threatening many countries around the globe. In this work, we study regional drought in northeastern area of Thailand. Since 1975 droughts northeast, especially Nakhon Ratchasima province south region, have occurred more frequently than past with stronger intensity. We firstly investigate relationship to cycle El Nino Southern Oscillation (ENSO) and find that cool phase ENSO (or La Nina) shows positive effect increase rainfall, whereas warm Nino) has no clear decrease rainfall province. then further our by inducing a model monitor situation based on historical remotely sensed data. Data six years had been selected from both excessive rain fall due Nina unclear cause. also draw estimate amount lagged two three months The monitoring estimation built decision tree induction algorithm models’ accuracy tested 10-fold cross validation 68.6 % 72.2 %, respectively.