The Global Land Cover Classification Experiment Research

In recent years, the international community has been devoted to researching on globalsurface cover and its changes through the space remote sensing technology. Only the UnitedStates completed the spatial resolution of1km surface coverage data products around theworld, the eu completed the spatial resolution of300meters so far,but its overall accuracy islow, classification system and the temporal and spatial resolution are difficult to effectivelymeet the needs of global change and earth system research.It has not yet been tried applicationsatellites to carry out the global surface meteorological or land cover mapping work in ourcourtry and need to develop autonomous high-precision global surface coverage effectiveremote sensing dynamic monitoring technique and data products urgently for our country’sglobal change research and the development of earth system mode.It aims to design anddevelop a new generation of global surface coverage data series products for global changeand earth system model study provides reliable basic data support to meet the global changeresearch in our country and the urgent needs of the earth system model in this paper.Shaanxi province was studies as the experimental area in this paper,It is easy to implement sampling and precision of the field test and verify because of Complex and varied topography and climate characteristics and complete type of surface coverage in Shaanxi province.It establish global surface cover classification technology processes and methods for the two phase of the Landsat data in2000and2010through data collection first, and then geometric correction, radiation correction, terrain correction, and then classification, finally the whole process ofproduct quality inspection test.It focus on he level4test classification strategy including division, image characteristics extraction, Multiple classifiers ensemble and manual editing based on expert knowledge.In order to establish the global surface coverage classification technologyprocesses and methods. It is identified a set of in line with the classification system ultimately of this paper bystudying and researching the existing six sets of global surface covering the data products. Itmakes surface coverage type decomposite and simplify step by step by level4classificationstrategy.At the same time,It can improve the classification of the target, reduce degree ofdifficulty classification and improve the precision of the single factor extraction andefficiency.Finally it formed the process named single class elements hierarchical extract maskclassification.