Study On Conductivity Imaging And Inversion For Airborne Time-domain Electromagnetics

Time-domain airborne electromagnetic method (AEM), the transient electromagneticdetection devices are placed in the high-speed aircraft to detect the earth, which is speed, lowcost, high efficiency, can be widely used in geological mapping, mineral exploration,hydrographic surveys,environmental monitoring and so on. In recent years, with the country’sgreat importance to the field of geophysical, time-domain airborne electromagnetic method inour country has developed rapidly.This article rely on the National Natural Science Foundation of the key technical studiesmagnifying liked to inverse the time domain airborne electromagnetic and863major projects"aviation geophysical exploration technology system" sub topics "the podded time-domainhelicopter airborne electromagnetic survey of theoretical research and system the actual design,we study the helicopter time-domain airborne electromagnetic data imaging and inversion,major findings and content are as follows：This article describes the basic principles of time-domain airborne electromagnetic methodand introduced its detection methods, the basic structure, transmitting and receiving device. Tolaunch a step pulse, after emission current is shut off, we computing in the inducedelectromotive force changed over time generated in the receiving coil at the center of thetransmitter coil, using the calculations method, which transform the frequency-domain totime-domain. On the basis of Hankel representation in the frequency domain electromagneticfield, firstly calculate the transient response of the time-domain electromagnetic field in theLaplace transform domain, and then do the inverse Laplace transform with the GS transform, tocalculate the induced electromotive force in the receiving coil. In actual airborneelectromagnetic survey system, the emission waveforms have different types. On the basis ofoff-time response known which launched by step pulse, according to the convolution nature andsignal processing theory, we obtain the off-time response corresponding to other emissionwaveform, to lay the foundation for subsequent research. In the condition of same energy of theemission current constant, we calculate the half-space earth model off-time responsecorresponding to different emission waveforms. The results show that the trapezoidal waveresponse has the biggest amplitude.The study results showed that when the conductivity of the good conductor increases to acertain extent, the off-time response is no longer expression good attenuation trend,when theon-time data amplitude is still large,we studied the algorithm of calculation time-domainairborne electromagnetic data on-time response, refer to Yin (2008), based on the known stepresponse, we make current derivative to do the convolution with the transmitted waveform to getthe corresponding on-time response, and calculate the on-time response corresponding todifferent emission waveform.Second, we studied the conductivity-depth imaging (CDI) of time-domain airborneelectromagnetic data. The CDI which using the interpolation table is simple, fast, and has been widely applied inthe airborne data interpretation. First we introduced the method how to get conductivity-depthresults from interpolation tables, we calculate the electromagnetic response of a variety ofhalf-space earth model on specific flight altitude as the interpolation table, the using theobserved response to look-up tableto get the corresponding apparent conductivity from the lastchannel, last we calculate the apparent depth with the apparent conductivity and center time ofeach channel, thus, CDI results is got. Study the first-order derivative of conductivity-depthimaging based on the results of interpolation table, the method define the first order derivativeof the conductivity acorrding to the skin depth, the method highlights the trend of the earth’sformation, enhance the results of interpolation table CDI, improves the conductivity depthimaging resolution. The calculation for typical model shows that the last one is more effective.Thinking the look-up table method based on the pseudo-layer half-space conductivity-depthimaging, we introduction the pseudo-layer whose conductivity is zero to remove the errorcaused by the bird height, terrain, etc. In order to avoid the two-value problem caused bysingle-channel inversion, we convert two channel data to the time constantτ iand amplitude ofresponse αi, and to enhance the relation between channels when looking-up table, reducing theimpact of two-value problem. This method calculate different half-space earth model of theelectromagnetic response under some flight altitude as an interpolation table to obtain theapparent conductivity, and then calculate the depth depending on flight altitude and pseudo-layerthickness. Calculations for typical model show that the method is more accurate.Third, the1D time-domain airborne electromagnetic data inversion technique wasreaearched.We fullly described the principle of the Zohdy inversion, the Zohdy inversion process aregiven combined with model examples. On the basis of the CDI results form the interpolationtable, we firstly adjust the depth at the same time maintain the conductivity, and then keep thedepth constant, iteratively adjust the conductivity until the fitting error of the observed responseand calculate response reach to the specified accuracy. Typical model examples show that themethod is high accuracy, good resolution to good conductivity anomalies.The traditional inversion methods are firstly linearized the complex mapping between theearth model and the response and the iteratively inversion, this process need for the Jacobianmatrix calculation, and which is easy to fall into local minima. The BP neural network has theability to approximate any nonlinear mapping, which is suitable for non-linear inversion. In thispaper, the neural network is introduced into the time domain airborne electromagnetic datainversion. A magnifying glass liked inversion algorithm was illustrated. We decide the initialmodel according to the results of conductivity-depth imaging, and then use the fitting error todetermine the direction to adjust model parameters, then we build a model group to train theneural network, after the network is mature we input observer response to the net, whose outputwill be the new initial model, the process proceed accurately until the fitting error between thenew model response and observed response achieve to the specified accuracy. Typical modelexamples show that this method can achieve better results.Fourth, this paper presents the interpretation of measured data got form airborneelectromagnetic experiments.The data were measured in the airborne experiment in Changchunsuburb using the JLATEM helicopter time-domain airborne electromagnetic survey systemdeveloped by the Instrument Science and Electrical Engineering of Jilin University. After thesimple pretreatment data process, the interpolation table method and the pseudo-layer half-space look-up table method CDI results were given, and do a simple analysis.