A Satellite Based Monitoring of Changes in Mangroves in Krabi, Thailand Radiometric correction Normally, the NDVI value of vegetation object ranges around 0.5-0.8, while the vegetation index value derived from the 2002 image presented the strange characteristic of low value. Although LANDSAT 5 and LANDSAT 7 have the similar orbits and repeated patterns, however detection calibration differences between the various sensors system might be affect pixel brightness value (Lillesand and Kiefer, 2000). Image normalization was performed in order to reduce pixel brightness value variation caused by non surface factors. Image normalization was achieved by applying regression equation to the 2002 imagery which predict what a given brightness value would be if it had been acquired under the same conditions as the 1995 reference scene. A total of 21 radiometric control points was used to normalize the LANDSAT (ETM+) data on 2002 to the 1995 LANDSAT(TM) data. The band 3 of LANDSAT (ETM+) was normalized to the 1995 data using 3 common wet targets found within the sea area and 8 common dry points extracted from a bare soil area. Thus, linear regression was computed (figure 4). Same as band 3, the repeating procedure was done to the band 4 of the 2002 dataset on 9 wet targets and 6 dry target points. The linear equations computed were applied to the 2002 data using the ENVI system. Consequently, the new NDVI image on 2002 data was transformed (figure 5).  Figure 4 Radiometric correction of LANDSAT (ETM+) data set band 3 and band 4  Figure 5 NDVI histogram on 2002 image after normalized to 1995 image The new histogram of NDVI on 9 February 2002, presented the NDVI value ranged from -0.4031 to 0.7524. Although, vegetated area also appeared the highest NDVI value than other objects, varied from 0.3189 to 0.7524. Barren lands and built-up areas ranged from -0.0299 to 0.3189. Water Bodies had also shown up the lowest NDVI value of -0.0299 to -0.4036. Considering on histograms among 3 images of different time series found than urban area was expansion by give the higher curve ranged around 0.17 on the 2002 histogram than the previous years. RGB-NDVI classification Several methods apply for change detection monitoring by using NDVI classification (Fuller, 1998). Among those techniques, RGB-NDVI classification is the fastest and easiest method to perform (Sedar et al., 2001; Sedar and Winne, 1992) The NDVI which computed from each of the three years (1995-2000-2002), were applied to ISODATA unsupervised classification. Change detection was taken an account on the changes which happen on 1995 to 2002. A statistic test (t-test single factors) was applied in order to compare 2 means on NDVI value among years. Change areas were significantly different at 95 % (p<0.05) (Table 1). Table 1 mangrove change determined at significantly different at 95 % (p<0.05)
Considerate on only mangrove area, five categories were detected which could be grouped into 2 main categories of no change area, and change areas (Figure 6). The changes monitoring derived from NDVI technique presented the stable status within mangrove area covered an area approximately of 43.3 square meters. Changing occurred surrounding the territorial edge of the mangrove which major caused from shrimp aquaculture activity. Total of 9.3 square kilometer of mangrove area were destroyed during 1995-2000, while non-detection on mangrove area decreased happen between 2000-2002. A total of 7.4 square kilometers of urban was remained in the study area. Since 1995 to 2002, moreover than 200% of Agricultural area was found destroyed for urban expansion and construction the built-up area which includes airport land. Most of changing was detected on territorial part of agriculture section. However, the changing seems to be caused from an alternative with in crop cycle of agricultural plants which occurred in this area such as para rubber, oil palm and paddy field.  Figure 6 Changes monitoring during 1995-2002 using NDVI Conclusions Several techniques of change detection could be implementing according to the objective. In this study, NDVI-RGB classification were conducted in order to looking the fast and easy method which could be provide change information. The result from NDVI-RGB could be providing useful information as a draft of changing. It did not provide information in deep stage. However, this method could be detected the changes around the mangrove area. Acknowledgement The study is a part of studying Master degree on the Integrated Tropical Coastal Zone Management field of study at Asian Institute of Technology for 20 months which awarded from the Danish International Development Assistance. The author wishes extremely thank to her advisor Dr. Somsak Boromthanarat, for their valuable advices and guidance towards the completion of this thesis. Special thank is due to Dr. Ole Pederson and Dr. Thongchai Charuppat for the suggestions and for serving as members of her thesis committees. Sincere acknowledgments are extended to the personnel of Mangrove Management Division, and Department of Marine Resources and Environment of Ministry of Natural Resources and Environmental, Bangkok, Thailand, for the providing valuable data and advices. Sincere acknowledgment is also extended to the personnel of Natural Resource Assessment Division, Royal Forest Department of Ministry of Agriculture and Cooperative, Bangkok, Thailand, for their suggestions and providing valuable data. Special acknowledgment is extended to Sombat Poovachiranon, Somkieat Khokieatwong, Nalinee Thongtham, Phuket Marine Biological Center for their valuable comments during the field work in Krabi. Sincere thanks are also extended to Phuket Marine Biological Center staff in Phuket for their valuable support and help during the field work, data collection and sampling analysis. The author offers special thanks to manager of Remote Sensing Laboratory and Ms. Wutjanun Muttitanon for their assistance during this study. The author deeply acknowledges award of scholarship to her to attend the higher Degree Program at. Finally, Thanks are due to all the colleagues and friends, Maneerat Suseangrat, Wanawan Channuhong, Lerio A. Agdalipe, Audrie J. Siahainenia, for their help during her study. References
Page 3 of 3
| Prevous | |