RADARSAT-2 2011/11/26 2012/02/13 THAICHOTE 2009/12/13 2011/11/25 GEOEYE-1 2011/11/22 100 Km Figure 1. The Chao Phraya River basin and the areas covered by satellite images used in this study. EXTRACTION OF FLOODED AREAS IN THE 2011 THAILAND FLOOD FROM RADARSAT-2 AND THAICHOTE IMAGES Pisut Nakmuenwai 1 and Fumio Yamazaki 2 1 Graduate Student, Graduate School of Engineering, Chiba University, Japan; Senior Computer Scientist, GISTDA, Thailand, pisut@gistda.or.th 2 Professor, Graduate School of Engineering, Chiba University, Japan, yamazaki@tu.chiba-u.ac.jp ABSTRACT This paper examines an extraction method of widespread flooded areas in the Chao Phraya River basin of the central Thailand during the 2011 monsoon season. RADARSAT-2 imagery data were mainly used to extract affected areas while ThaiChote imagery data were aslo used as optical supporting data by the Thai government. In this study, the same data were used by a somewhat different method in more detail. The extracted results were validated by GeoEye-1, a high-resolution optical satellite image, water height data from gaging stations and a digital surface model (DEM) from LiDAR. Index Terms— 2011 Thailand flood, RADARSAT-2, THAICHOTE, SAR 1. INTRODUCTION Floods happen almost every year in Thailand and have brought dissatisfied situations. Severe flooding occurred during the 2011 monsoon season. It spread throughout the northern, northeastern and central provinces of the country. It caused heavy economic impacts by disturbing industrial production activities of the affected areas and the supply chains of world’s industries. In this study, satellite imagery data, the most effective ways for extracting information of large-scale disasters, are introduced. RADARSAT-2 (RS2), a Canadian SAR satellite with the C-band at a wavelength of 5.6 cm, had been mainly used for flood monitoring in Thailand since 2008 [1]. It can operate in the daytime/nighttime and under all weather conditions, and thus is considered to be very effective in flooded area extraction because water surfaces always show low backscatter [2]. ThaiChote (TH1), the Thai first satellite, was used as optical supporting since 2004. The NDVI values obtained from TH1 images could recognize flooded areas in the open space under a clear sky condition. After the flood situation had ended, a Digital Elevation Model (DEM) from LiDAR was brought into use for the first time in Thailand in early 2012, which would be very helpful to improve the accuracy of flooded areas extraction. These data sets are used in this paper for flood situation monitoring in Thailand. 2. STUDY AREA AND IMAGERY DATA This paper focuses on the central part of Ayutthaya, which includes Ayutthaya Historical Park, Rojana and Hi-Tech Industrial Estates, about 16.5 km in width and 21.0 km in length shown in Figure 1. To detect floods in a large-scale area, ScanSAR mode HH (single) polarization had been used in the most cases by the Thai Government. An in-flood image observed on November 11, 2011 was in SCNA (W1+W2) beam types shown in Figure 2 1A while a post- event image observed on February 23, 2012 was in SCNB (W2+S5+S6) beam types shown in Figure 2 2A. Both of them were observed from the descending path approximately in the duration of one minute, respectively at 6:07 and 6:12 in the local time of Thailand. A GeoEye-1(GE1) image taken during the flood event on November 22, 2011 was also used in this study. A pansharpened GE1 image has 4 multispectral bands with 1- m resolution shown in TH1 and RS2 images were used to extract water body due to Figure 3 1A. TH1 multispectral image with 15-m resolution taken in the flood event on November 25, 2011 is shown in Figure 3 2A and another in the dry season before the flood on December 12, 2009 in Figure 3 3A. 3354 978-1-4799-5775-0/14/$31.00 ©2014 IEEE IGARSS 2014