Study On Interpretation Of Rangeland Resources Types Based On Multi-source Remote Sensing Data

Rangeland resources play an important role in social development and human existence. It has received people's extensive attention because of its multiple value in ecology, economy, society, etc. As a high-tech, RS technology provides an efficient, in-time scientific way to investigate the special distributing and utilization situation of rangeland resources. To meet the needs of country for renewal data of rangeland resources and production programming, The research choose three kinds of different satellite data based on applicability, economy and can easy-got to apply to investigation and classification to discuss their applied characteristic in rangeland resources classification on " 3S " technical information platform. The studying area is set up in the northern slope of Tianshan Mountains, the whole area is about 5443km2. This area contains the abundant rangeland resources types, which establish the basis for the utilization of remote sensing technology in classifying. First of all, the pretreatment was carried on to multi-source remote sensing image, which including geometry correcting, sharpness and to choose optimum wave bands; Secondly, the classification system of rangeland resources and types catalog based on it were introduced. The research mainly used supervised classification method (MLC) combined with GPS data and unscramble symbol of RS image to classify the rangeland resources in studying area. Topographic vectorgraph and DEM also were built in order to assistant classification and check-up of types, and the 3-D image picture enable the physiognomy and special distributing of rangeland resources to reappearance. Analysed from the classification images and check-up result of multi-source RS data, the conclusion are as follows: Whether in plain, partly middle-mountain, sub-alp or alp region, ETM+ can discern and classification well, the accurate rate of classified types can reach 73.79%~86.43%, but in lower-mountain and mostly middle-mountain it is difficult to classify the rangeland resources type. MODIS can only classify in the first unit because of its lower special resolution ratio, so, it is not suitable for the application in classifying. However, it can be used for carrying out dynamic monitoring and biomass estimating in rangeland resources because of its high time resolution ratio and lower price. For this reason, the study chose NDVI and RVI vegetation index as well as combined the biomass to make a quantitative description to the types obtained from CBERS data. Further, the biomass estimating model, attribute database and graphic database of the studying area have set up. This research will offers a set of beneficial technological means for rangeland resources investigation, managing and monitoring, it can also provide the scientific data and practice basis for the production programming of studying area.