The Genetic Algorithm’s Application In The Sea Water Acoustic Impedance Inversion

Since the21st century, the rapid development of the earthquake oceanographylets us see the huge potential of the disciplines in the exploration, developments andprotections of Marine resources. And earthquake oceanography is the bridge whichconnects all kinds of the advantages of offshore exploration seismology, including thehigh lateral resolution and continuity of gathering information, with inner relationshipbetween temperature salinity and density in physical oceanography to study themicrostructure characteristics of sea water. Acoustic impedance inversion of marinewater has played a vital role in the study of earthquake oceanography. The accuracy ofacoustic impedance inversion of sea water not only determines the accuracy ofacoustic velocity and density of sea water, but also even affects the inversionprecision of micro-structure of temperature salinity and density of sea water. But theinversion approaches of acoustic impedance about ocean water are few, this paperattempts to apply nonlinear genetic algorithm acoustic impedance inversion of seawater, and study how to further improve its accuracy.Genetic algorithm is a kind of optimization search algorithm of intelligenceglobal and having the adaptive function, which simulates creatural genetic andevolutionary process in nature. The biological law of "natural selection" and "survivalof the fittest" widely exists in the nature, which can be divided into three evolutionstages of selection, crossover and mutation by genetic algorithm. Due to theparallelism and adaptability and robustness of genetic algorithm itself, as well asdisplaying superior performance in solving some complex problem, the algorithm hasbeen used in more and more science research field. Acoustic impedance inversion inseismic exploration is a problem of multi-parameter, multiple solution and complexnonlinear, but the characteristics of genetic algorithm make the algorithm widely usedin the seismic acoustic impedance inversion and acquire good performance. However,randomness, slow convergence of the genetic algorithm limit the application of the algorithm in seismic acoustic impedance inversion.For changing some shortcomings of the genetic algorithms showed in seismicimpedance inversion, the paper does several improvements to the traditional geneticalgorithms according to the adjacent sea floor acoustic impedance is small and the seawater reflection coefficients is similar. According to the sea water acoustic impedancereflection coefficients changes in a little range [-0.001,0.001], the paper uses hot bathmethod to reduce the solution space and improve the genetic algorithms search speed;Using reciprocal conversion method to calculate the fitness function values of thechromosomes to increase individual differences and ensure the future generationpopulation have a number of outstanding individuals; Using multi-populations elitesprotection strategy and population migration strategy to increase the individualdiversity of each population and protect the best individual in the population; Usingimproved crossover makes the crossover display its full neighborhood searchcapabilities; Designing a table for a set of crossover and mutation probability toprotect the individual diversity of the population; And using two selection operators ofdifferent characteristics to ensure the individual diversity of the population.To compare the inversion effects of the genetic algorithms and improved geneticalgorithm, the paper designs20layers sea water model and100layers sea watermodel based on the XCTD data to validate results of the two inversion methods. Andtwo sea water models inversed results demonstrate the improved genetic algorithmcan inverse more efficiency and accuracy sea water acoustic impedance than thetraditional genetic algorithm. Finally, the improved genetic algorithm is applied to theinversion of a stacked two-dimensional seismic data. The paper extracts sea waterimpedance of the studied sea area and derives the two-dimensional distributions of thetemperature, the salinity, the velocity and the density of the sea water. The contrastiveresults between the real XCTD data and the inversed values verify the improvedgenetic algorithm in the paper is validity.