|
|
|
Education/Research
|
Case studies for landcover chance analysis using Micro-Asean
Abd-alla Gad A. G.
National Research Center &
Remote Sensing Center,Cairo, Egypt
hoji Takeuchi,Kauhei Cho,Mitsunori Yoshiura
Remote Sensing Technology
Center of Japan, Tokyo 106, Japan
Abstract
MICRO – ASEAN, the image analysis software for a personal computer developed through the join research project between ASEAN countries and Japan, has now become available for various studies for monitoring of landcover change using remote sensing. Because of its high cost performance and easy-handling, MICRO-ASEAN has been used as one of the core systems for the training courses for remote sensing data analysis. This paper reports some results obtained from the case studies conducted using MICRO-ASEAN at the training course as well as at the joint research. Two kinds of case studies are reported, one is the analysis using MOS-1/MESSR and Landat/MSS for landcover change at Tama Area, one of new urbanized areas in Tokya Capital Region, which was conducted as the subject for the remote sensing training course. The other is the case study for monitoring of the wide-range landcover condition is Thailand using NOAA/AVHRR data combined with MOS-1/MESSR data. Through these case studies, MICRO-ASEAN was proved to be used effectively and practically for landcover monitoring by remote sensing.
Introduction
Since a personal computer technology has made a remarkable progress, the image analysis software using a personal computer for practical use of remote sensing data has become possible to be realized. MICRO-ASEAN (MICRO computer-based Advanced System for Environmental Analysis with remote sensing data) was developed on the above background through the joint research project with ASEAN countries by Special Coordination Funds of the Science and Technology Agency of Japan. This software was developed based on the software package called MICRO-TIPE developed by Tokai University and extended to the system including the functions for GIS and the analysis of multi-temporal and multi-stage remote sensing data.
MICRO-ASEAN has now become one of the important systems used for the international cooperation for promoting remote sensing data utilization in developing countries, such as the training course for remote sensing technology. This paper reports some examples of the case study using MICRO-ASEAN, which will be very useful as the subject for remote sensing data analysis at the training courses as a typical study for monitoring of landcover change. Two kinds of case studies are reported, one is the analysis using MOS-1/MESSR and Landsat/MSS at Tama Area, one of new urbanized areas in Tokyo Capital Region, which was conducted as the subject at the training course held by JICA (Japan International Cooperation Agency). The other is the case study for monitoring of wide-range landcover condition in Thailand using NOAA/AVHRR data combined with MOS-1/MESSR data, which was conducted as a part of the joint research described above.
Landcover Change Extraction of Tama Area Using MESSR and MSS Temporal Images
- Background and Location of Study Area
The increase of the population in Tokyo Capital Region has urged the society to extend its residential area. This trend has an impact on the environment. The purpose of this study is to test the applicability of MICRO-ASEAN system for detecting the change of landcover through a particular case. Two satellite images for the study area were used; MOS-1/MESSR data Nov. 1990 and Landsat/MSS dated Oct. 1981.
The test site “Tama Area” is located 30-40 km from the center of Tokyo on the Tama Hill to the west. A new planned city “Tama New Town” has been developed in this area. The total area of the New Town is about 3,016 hectares. Development of Urban facilities such as roads, water works, etc. is carried out for a planned population of 370,000 people, while housing constructing is planned for an expected residential population of 310,000.
- Methods
The data analysis flow is shown in Fig. 1. Two approaches were applied to investigate both of landcover condition and landcover change. The first approach as landcover classification procedures by supervised maximum likelihood classification method. The second one was the change detection using normalized vegetation index (NVI).
In the second approach, a normalization procedure was applied for both of MESSR and MSS images to follow the change in NVI. Then, the discrimination of vegetated and non-vegetated land was performed for both temporal images. Landcover change between vegetation and non-vegetation was extracted using the result of discrimination by NVI.
- Results and Discussions
The system allowed to select 14 landcover categories including total 30 training areas. It was possible to confirm a particular spectral signature of each object. The result of confusion matrix of the training data set showed a significant correlation in classification of the following categories; Factories (85.8%), Forest 1 (84.8%), forest 2(92.3%), Bare soil (93.9%), Water (97.6%), City 4(81.%), City 1(88.5%) and Graves (84.9%). However, weak coincidence were found in other different categories (e.g. Golf Courses and Secondary Forest). This can be attributed to the mixing of these categories in feature space and also to the seasonal effect of vegetation cover.
The result of landcover classification of Tama Area is represented by the color-coded image shown in Fig. 2. The result of the area measurement of classified data is shown in Table 1. The result shows an emphasis on creating new urban areas. The old residential areas (City 1 and City 2) represent 6.7 and 2.9% respectively of the total area. However, less dense built-up areas record higher percentage of the total study area (i.e. City 3=19.6% and City 4 =15.5%). It is also remarkable to notice a significant wide coverage by forested area (20.3% total area of Forest1, Forest2 and Secondary Forest).
Fig. 3. shows the result of change detection using NVI. This image was obtained though the filtering of resultant image by change detection procedures between vegetation and non-vegetation. Table 2. shows the result of area measurement of three kinds of changing patterns, 0;no change, 1;change from vegetation. The result shows a significant change from vegetation to vegetation. The result shows a significant change from vegetation to non-vegetation (type-1). By comparing Fig. 2 with Fig.3, the changed area of type-1 almost corresponds to the surrounding urban areas of Secondary Forest. This suggests that significant change from
the secondary forest to urban occurred during 9 years difference between the observation dates of MESSR and MSS images.
|
|
|
|
|
|
|
