Research On Local Marine Geomagnetic Field Model And Detecting Of Geomagnetic Abnormal Datas

Geomagnetic field model and geomagnetic diagram have important meaning and researchvalue such as in the geoscience, resources exploration, the earthquake forecast, geomagneticassistant navigation etc. This paper mainly studied to the problems that make use of the marinegeomagnetic data to build local marine geomagnetic field model. It is analysis and discuss in thispaper that the data quality control of geomagnetic abnormal data detecting method, the commonways to build local marine geomagnetic field model, the method of geomagnetic map gridding,the BP net work based on genetic algorithms to build local marine geomagnetic field model.Main contents as follows:1. Introduced the method of build geomagnetic field model and present research conditionof domestic and international. It is introduced and analysised in turns that the ways to set uplocal geomagnetic field model, geomagnetic survey data processing and geomagnetic abnormaldata detecting, geomagnetic map grading etc.2. Improve the disadvantage of common least square tend surface method in geomagneticabnormal data detecting, aim at the problem that lack of ability in robust of traditional marinegeomagnetic data processing quality control. Robust estimate is introduced, so the method thatbased on the robust least square tend surface method is used in abnormal data detecting. Theexperiment used geomagnetic survey data of some sea region, the result shows that, the methodbased on robust least square tend surface is better than common least square tend surface indetecting, more efficiency and stabilization.3. To the problem of quality control and node choice in local marine geomagnetic fieldmodel based on multi-surface function, a combining fitting with robust trend surface andorthogonal multiquadrics is proposed. An adaptive node choosing method is proposed based onthe effect of every node to the curve fitting calculated by using the orthogonal least squares.4. Introduced the BP network to build local marine geomagnetic field model, the BPnetwork based on genetic algorithms is used for the problem that part better result. The differentof this two methods is analyzed in simulate data and real data. Considering gross error and nodechoice to build the geomagnetic field model, result shows that, the method based on geneticalgorithms is better than the other in precision and stability.5. The problem of high precision geomagnetic navigation mapping and gridding, if the highspatial resolution of geomagnetic grid data is not adapt the geomagnetic navigation. Coonsmodel is used in reconstruct local marine geomagnetic map based on some sea area geomagnetic survey data. Taylor and Kriging model as the compare experiment to build geomagnetic map, theresult shows that Coons is better than other models in precision, provide a reference method togeomagnetic navigation mapping and gridding.