Keyword: Forest cover, Normalized Vegetation Index, Total Reflected Radiance Index, Remote Sensing, Change Analysis

Abstract
Forest cover is an important indicator for environmental assessment of a territory. Land cover maps that also shows forest cover can be established by different methods, with different levels of accuracy. However, in environmental assessment, time and accuracy are requirements which should be fulfilled. Fast analysis provides the ability to accomplish land cover mapping of large areas in a short time. Appropriate accuracy ensures that the thematic content of the map is on an adequate level. In this paper the authors report on the application of the normalized vegetation index (NDVI) and the total reflected radiance index (TRRI) for land cover mapping with an emphasis on forest cover in the area of Tanh Linh District, Binh Thuan province, Vietnam. Recently drastic clear cutting of natural forest has occurred in this area for different purposes: logging, agricultural cultivation etc. The research has been performed based on multi-temporal LANDSAT TM data (1989, 1992, 1996 and 1998) and topographical maps compiled in 1965. Change analysis has been executed based on digital interpretation results. The authors further report on a land cover classification algorithm with NDVI and TRRI indices, which enables the removal of some inaccuracies of vegetation classification by the NDVI method in the case of high resolution data.

Introduction
Vegetation cover in general and forest cover in particular are indicators for the quality of the environment. The methods of vegetation and forest cover classification has progressed in recent years. However, speed of analysis and accuracy of results are problems under discussion and issues to be improved. In this paper the authors have developed a method of automated classification of forest cover which can be used for any remote sensing data set which has at least two spectral channels in the red and infrared regions. The proposed method has been used to asses forest cover change of an area that belongs to Dong Nai and Binh Thuan provinces The study area covers Tan Phu, Ham Tan and Tanh Linh districts where recently drastic forest clear cutting has occurred that has led to serious environment problems such as drought, flooding etc. The flush flood which occurred on July 30, 1999 is the biggest flood in over 20 years in this area and has caused very serious damage for local people. During the period from 1965 to 1998 about 52% of natural forest cover has been cut in the study area. When comparing forest cover of 1998 to 1989, 33% of natural forest cover has been cleared. Apparently if the forest cover had not been destroyed in such a scale, maybe many disasters would not have occurred as seriously as they have done recently. Moreover, when comparing the official statistical data from the National Statistical Yearbook for Vietnam with data with data derived from satellite remote sensing data one finds out very large discrepancies. According to the Statistical Yearbook from 1990 to 1995 only 11,922 ha of natural forest had been cut for the two provinces Dong Nai and Binh Thuan, but from remote sensing data the depletion of natural forest for only two districts is as much as 22,259 ha. Certainly if we use remote sensing techniques to detect forest cover change for all of the two provinces, the depleted forest area will be much greater than the official figure. Apparently the development of remote sensing techniques to monitor forest cover is urgently needed. The data derived from remote sensing image is objective and reliable for evaluation of the status of the environment and also very useful for the selection of proper environmental management schemes.

Materials and Method Materials and Method
The following information sources have been used for study:

• Topographical maps, scale 1/50,000, published in 1967
• LANDSAT TM images from 1989, 1992, 1996 and 1998
• Ground truth data in the study area from 1997, 1998 and 1999

The topographical maps in UTM projection were compiled by the U.S. Army in 1965 for military purposes. The maps were established by aerial photos and other information sources in UTM projection in 1965. Therefore content of the maps reflect land cover status of the study area some time in 1965. The land cover was classified as natural forest, shrub, forest plantation, rice field and bare land. The topographical maps of study area were digitized. The contour lines were input to the database to support image interpretation. A land cover map of the study area based on topographical maps is shown in Figure 1.


Figure 1. Forest cover extracted from topographical maps UTM 1965