Reliability Analysis And Controlling Methods For National Geographic State Monitoring Data

The development of theory and technology in National Geographic State Monitoring(NGSM)provides a strong technical support for obtaining a large number of geographical information.However,there are not enough researches on reliability problems in the data,models and methods of NGSM.This problem has drawn wide attentions in terms of large-scale applications and decision-makings based on NGSM data.The reliability theory of NGSM is the extension of the spatial data uncertainty theory and reliability theory.In addition,NGSM reliability analysis is a new application direction of reliability theory and technology.Towards the complexity of the NGSM system,this paper puts forward the reliability analysis methods of land cover data and geographic features from the aspects of geographical data,data processing and semantics.In addition,reliability controlling methods of data processing are proposed.The research can promote the development of the theory and technology of NGSM.Specific research work includes:(1)This paper proposes the analysis method of the NGSM data.There is a lack of effective NGSM data reliability.evaluation models and methods currently resulting inthat the degree of reliability of the data is unknown for data users and decision makers,affecting the application of data-based decision-making.Based on the uncertainty sources of spatial data,the spatial reliability evaluation indicator system of the NGSM is developed.The proposed indicator system realizes the embodiment and extension of reliability indicators.The evaluation model for the land cover data and the geographic national features in the NGSM is established.The effectiveness and practicability of the proposed method are proved by the reliability evaluation experiments of a wide range of NGSM data.The evaluations can master the data reliability level,provide targeted decision support for different fields.(2)This paper proposes a fuzzy reliability reasoning model of NGSM data processing based on the Fault Tree Analysis(FTA).The process of NGSM is a series of data analysis and processing activities to transform the observation data into the final results.The NGSM process contains a large number of variables and uncertain factors.Based on the reliability influence factors and their mutual relations in the process of data acquisition and processing,a fuzzy reliability reasoning model based on FTA is proposed.Based on the analysis of the reliability factors,the randomness,the correlation and the fuzzy uncertainty in the propagation process are obtained and the propagation law of the monitoring process is revealed.The experiments are conducted on the reliability analysis of NGSM data acquisition process.The results show that the process reliability model can reflect the trend of data reliability to a certain extent,and reduce the dependence on the reference data,which provides a new way to evaluate of NGSM data.(3)This paper conducts a semantic similarity metric model based on graph theory and local semantic density.Semantic similarity describes the reliability of the semantic level in the process of geospatial data interaction between different systems,which affects the applicability and usability of the NGSM data in the target system.Taking the national standard of Land Use Classification(GB/T 21010-2007)as an example,this paper analyzes the influence of the semantic heterogeneity between the classification systems on the data reliability.The semantic analysis results describe the applicability of NGSM data in the Land Use Classification system and provide references for data interaction and knowledge sharing between two systems.(4)Currently,the main data processing ways of NGSM include:(a)Data acquisition based on artificial interpretation and machine automatic interpretation;(b)change detection methods based on multi-temporal remote sensing images.In this paper,the reliability controlling methods are proposed for the two data processing ways.1)In this paper,the reliability controlling mechanism of data processing process based on fuzzy importance is proposed,and the influence degree of uncertainty events on reliability is calculated.The uncertainty feedback mechanism can provide basis for the regulation of the reliability influence factors in the data processing process to modify the data processing and improve reliability.2)Uncertainty exists in the remote sensing image change detection and in the continuous transmission and accumulation,affecting the reliability of the results of changes detection.This paper introduces the idea of reliability controlling in the change detection algorithm based on active contour model,and proposes an active-contour-based change detection algorithm considering the local uncertainty.By measuring the uncertainty in the change detection and controlling the influence of the uncertainty factors on the results,the reliability controlling of the change detection process can be carried out to improve the reliability of change detection algorithm.The experiment proves the validity and practicability of the proposed reliable change detection method through the applications of forest fire monitoring and urban land cover change monitoring.