Abstract
This paper presents the results of forest discrimination, classification, and volume estimation in two test sites of China using multiparmeters airborne and spaceborne imaging radar data. The SAR data are acquired during SIR-C/X-SAR and GlobeSAR missions. To improve the understanding of radar backscatter to canopy geometric feature, we extracted backscatter coefficient and intensity to analyze the effect of forest type discrimination, and the relationship between forest parameters and radar backscatter. This study shows that it is very efficient and intensity to analyze the effect of forest type discrimination, and the relationship between forest parameters and radar backscatter. This study shows that it is very efficient for multifrequency and multipolarization SAR data to discriminate different types of forest.The intensity of radar backscatter is also quite sensitive to the forest parameters, especially diameter at breast height (dbh) and tree mean height. Based on this sensitivity, the forest volume of the test site was mapped into five classes.

Introduction
Remote Sensing for forest monitoring is important for understanding phenomena such as the global carbon cycle, hydrologic cycle, and energy balance. Many environmental problems of increasing importance are related to forest extension and condition. In 1983 NASA estimated that forest cover 33% of the total land surface, contain 90% of the standing biomass, and yield 65% of the net primary production. An increase in both global population growth and deforestation around the world requires accurate and timely information about the distribution and rate of change of global vegetation. This information has traditionally been derived from a variety of sources, the accuracy of which is difficult to ascertain. Remote sensing is a potential method for gathering this information, as has been demonstrated from AVHRR data. However, pervasive fog and cloud as well as limitations in solar illumination in the tropics and subtropics and high latitudes severely limit the utility of visible and infrared sensors. Synthetic aperture radar (SAR), with its near all weather, day and night capabilities, is a powerful tool for acquiring biophysical data that will improve understanding of global climate, the hydrologic cycle, the carbon-nitrogen cycle, and the global energy balance.

Many airborne and spaceborne SAR system have been used to carry out a large amount of experiments for investigating the forest ecosystems. The airborne systems, such as the NASA/JPL AIRSAR system, operating at P,L and C band, has flown over many forest sites inUSA and Europe (Zebker et al., 1991; Le Toan et al, 1992; Beaudoin et al., 1994; Rignot et al.; 1994; Skriver et al., 1994; Ranson et al., 1996). The experiments of the Canadian CV-580, as well as the European airborne system, mainly operating at C and X band also have been carried out in North America and Europe (Drieman et al., 1989; Hoekman, 1990). The spaceborne systems, such as the Seasat SAR, SIR-B, SIR-C/X-SAR and ERS-1, SAR, etc., were used for investigations of boreal, temperature and sub-tropical forestry test sites (Ford et al., 1988; Dobson et al., 1992; Ranson et al., 1995; Stofan et al., 1995; Rignotet al., 1995). These experiments and studies have shown that radar is sensitive to forest structureal parameters including above-ground biomass (Dobson et al., 1992; Pulliainen et al., 1994; Skriver et al., 1994; Ferrazzoli et al., 1995; Ranson et al., 1996). The study has demonstrated the use of SAR data to retrieve forest mapping and monitoring. Some models have been proposed for understanding and interpreting the electromagnetic interaction effects producing this sensitivity (Durden et al., 1989; Richards et al., 1990; Ulably et al., 1990; Sun et al;, 1991; Chauhan et al., 1991; Wang et al., 1993). So it is available for carrying out forest monitoring and developing the algorithms to estimate forest parameters based on the models and the sensitivity.

In this paper, we firstly describe briefly the test sites and data resources. The methodologyand result of forest discrimination and parameter estimation are then presented. The relationship between radar backscatter features and forest parameters are studied and the effect of SAR data on forest type discrimination are also discussed based on the extraction of radar backscatter coefficients.

Test Site and Data Sources
The distribution of forest has large difference in the northern and southern China due to its vast territory. Hence we selected two test sites in the south and north to explore the forest applications. A test site for three north shelter forests was selected west of Yichuan country, Shanxi Province, with the Yellow River flowing south on the eastern side of the image. This site has an average elevation of 1300 to 1400m, with hilly landforms of loess characterized by gullies and ravines due to surface erosion. It is a temperature, dry to semi-dry region with little annual rainfall. The forest in the site is part of three north shelter forests in China. Forest types are mainly deciduous, coniferous, and mixed forest. Deciduous trees are mainly popular and oak, coniferous trees are pine stands, and mixed forest are composed of deciduous mixed forests are composed of deciduous mixed forest and coniferous-deciduous forest. Another test site is Zhaoqing area, which is located in Guangdong Province of South China. It is a topic to subtropic regions, has heavy precipitation and moist climate, where soil moisture contents may vary widely within a short period. Soils consist of clay and silt, and have medium roughness. The ground surface has less vegetation cover because pine stands are artificial regeneration. Topgraphically, the region varies from flat to gently rolling, with a maximum elevation changes of less than 100m, and an average slope of 15 degree. The forest study is concentrated in the northeast of the test site. This area mainly consists of forest lands, agricultural fields, orchards, and water bodied, and with an area of more than 600 hectare. The forest stands are predominately coniferous(>60%), while deciduous and mixed trees are few in numbers (Liao et al., 1996). Pine dominated coniferous forests are plantation stands with average age of 10-15 years so that they are all young and have a crown closure of less than 20%. The height of pine trees usually ranges from 2 to 4 m with the tallest reaching 6.5 m. The density of the trees is from 850 to 5500 trees per hectare. The pine trees have 5-8cm diameter at breast height (dbh).