EPIDEMIC RISK ASSESSMENT OF ACUTE WATERY DIARRHEA FOR THE 2011 AYUTTHAYA FLOOD DISASTER USING REMOTE SENSING AND WATER QUALITY

摘要

The flood disaster generally leads to infectious disease outbreaks especially diarrhea frequently having a maximum rate of patients during flood occurrence. Flood water contaminated with pathogen closely relates to the epidemic risk of waterborne infectious diseases, therefore measuring water quality in inundated area can lead to estimate the epidemic risk due to flood. We focus to utilize dissolved oxygen (DO), which is a well-known chemical indicator of environmental water quality. Remote sensing especially radar imagery has an effective capability to detect water variables and also widely uses for determining flood parameters. This study aim to assess the epidemic risk of acute watery diarrhea by using remote sensing and the water quality sampling based on multilayer perceptron (MLP) function. The multiple temporal Radarsat2 images are used to determine flood area and duration during September to December 2011 in Ayutthaya province located on Chao Phraya river basin most severely affected by the 2011 major flood in South East Asia. The reference of epidemic risk derives from the weekly report of diarrhea morbidity from entire hospitals in study area when assume that the more than 50th percentile of weekly morbidity rate are epidemic period. To define input layer of MLP, the main parameters composing of flood duration, population density and dissolved oxygen are simulated to 50-m resolution and divided by each district area. The input and output layer are randomly divided into two parts: 66% for training and 34% for testing. After model training, the predicted epidemic risk of each district has correlation coefficient between 0.70 to 0.91 and RMS error between 0.012 to 0.045. Therefore, we conclude that based MLP algorithm we can assess the epidemic risk of acute watery diarrhea due to flood and the output risk map should also assist the decision maker to prevent and relieve the epidemic.