Forest Monitoring Framework At Regional Level Using Multi-Resolution Satellite Data
With Comdination Of Optical And Thermal Bands
Dr. Surat Tertlum*
Chair Prof. Shunji Murai**
Dr. Kiyoshi Hinda**
*Computer Science Department
Chulachomklao Military Academy, Thailand
E-mail: surat@cs.ait.ac.th,
surat@crma.ac.th
** Space Technology Application and Reserch (STAR) Program
Asian Institute of Technology, Thailand
Abstract
In order to study the regional scale reforestation process in Indo-China Peninsula. Suitable forest classification system is a main requirement for such the study. Since high resolution data set is not suitable of the data volume. A solution is by using low-resolution remotely sensed data (NOAA AVHRR LAC 1.1km) as the main data source, and resolution-remotely data (Landsat TM30m) as the ground truth data for monitoring and analysis of primary forest, secondary forest, and other vegetation resources. A vegetation classification methodology from the integration of low and high resolution remotely-sensed data with the combination of optical and thermal bands has been developed and proposed for this forest classification system.
This paper describes the vegetation classification methodology developed from the integration of low and high-resolution remotely sensed data with the combination of optical and thermal bands using multi-resolution satellite data, and the reforest monitoring framework at regional level is proposed for further implementation.
Introduction
Forest play a key role in global ecological balance. The world's deforested land increases fastest in Asia at 1.4 percent a year. In 1980, only 0.6 percent of the tropical forests were assumed to have disappeared. Today, the decreasing rate has doubled to 1.2 percent, or almost 17 million ha a year, mainly because of population growth and poverty. In developing countries, like Thailand, forest area was reduced from 53.33% in 1961 to 27.95% in 1989 and 26.64% in 1991 (RFD,1992)
Definition Of Land Types
For the purpose of this study, land types are defined into 5 classes according to their vegetation activities, without the separation of natural or planted vegetation, as follow:
-
primary Forest is the forest area that is still very dense and undisturbed. This type of forest includes tropical evergreen forest, tropical rain forest, and some dense mixed deciduous forest.
- Secondary forest is the forest area that is rather cleared, includes some type of mixed deciduous forest, dry dipterocarp forest, and forest area that is abandoned due to deforestation. Normally, it has lower density of tree and also the size of the trees is considerable smaller.
- Active agriculture is the agriculture land that is in the full grown condition such as rice field, or corn field, etc.
- Harvested Land is the agriculture land that is already harvested. This type of land is still have some small vegetation cover.
- Bare Soil is the land type that is covered by no vegetation.
Data Sets
Both low resolution are proposed to be used in this study.
-
Low Resolution Satellite Data (NOAA AVHRR LAC 11 km)
The Cloud-free NOAA AVHRR mosaic National Institute of Environmental Studies (NIES) of Japan is used as the tested data. The mosaic covers an area extending from East 66 degree to East 110 degree, and from North 0 degree to north 35 degree. The mosaic is composed from AVHRR images recorded by NOAA-11 during the year1990.
- High resolution Satellite Data (LANDSAT TM 30 m.)
Because of the proposed multi-resolution classification scheme, high-resolution data have to correspond to the data resolution data. Landsat TM data set according to the NIES's 1990 data is selected.
