Development And Application Of Grassland Classification And Biomass Monitoring Remote Sensing System For Qinghai Lake Basin Based On IDL

Considering the demands of RS data’s characteristic and image visualization, IDL was selected as a developing language for “grassland classification and biomass monitoring remote sensing system for Qinghai Lake basin”. In allusion to software’s function requirement and realistic developing key-point, the text describes detailed how to deal with grassland monitoring RS data, data’s visualization analysis and application by IDL, and discussed particularly for the design idea, technical rout, and implement methods of the system. By the means of developing living example, system user interaction, visual affect, terrace portability and expansibility are attempted with IDL development platform software design technology. Finally, the results achieved are as follows:(1) The article analyzed a variety of programming languages and platforms. The economic and convenient ENVI+IDL secondary development technology was choose to set up the system. The article studied the characteristics and kinds of key technologies of IDL language. The main control module, remote sensing classification module, the grass vegetation index calculation module and the grassland biomass calculation module were designed respectively. The rapid classification and the calculations of grassland biomass, coverage and leaf area index were achieved by the system and the goal of the grassland fast monitoring was achieved.(2) Using the mahalanobisdistance, euclidean distance, spectral angle mapping, maximum likelihood classification, decision tree for remote sensing classification of grassland. The decision tree and maximum likelihood classification results were superior to other methods. The maximum likelihood was the highest classification accuracy, up to 77.8%.Analyzed the six vegetation index NDVI, RVI, DVI, EVI, MSAVI and SAVI with aboveground biomass, there exists different degree of correlation between them. Among them, the highest correlation between RVI and biomass(the correlation coefficient is 0.776). Take various spectral index established as the independent variable to set up the linear, logarithms, quadratic curve and cubic curve regression models. Through analysis and comparison, finalized cubic curve model(y = 3.9852x3- 17.661x2 + 70.785 x + 65.624) whose independent variable was RVI had the highest precision and the R2 was 0.687. It was the best grassland biomass monitoring vegetation index model in the region Qinghai Lake basin.(3) After the system is completed, the grassland classification and grassland biomass calculation of the Qinghai Lake basin was realized and the results were achieved rapid visualization. The system has a friendly human-machine interactive interface and the simple operation. The system integrates the algorithm commonly used in grassland research and it has a large number of messages to point. Therefore, researchers can quickly deal with remote sensing images of grassland without sophisticated professional software. All system modules can be achieved quickly and well. Each module of classification, vegetation index, grassland biomass is all achieved. The visualization system is ideal. It makes overall tone of the image moderate in line with the human visual requirements, by making two percent linear stretch of the image. It is convenient to browse the image by image translation, image scaling, Eagle Eye and so on. System is good at scalability. Each module is independently developed, reducing the dependence between each other, so that the module can be added or deleted at any time. The optimization of system is reasonable. The slow loops are replaced by functions. In order to optimize memory’s usage, releasing the failure variables in memory in time. System run speed is satisfactory.