The impact of spatial sampling and signal-to-noise ratio on image quality in onshore seismics

Abstract

An onshore seismic survey is best conducted symmetrically, due to the reciprocity theorem of the wave field. Within the family of symmetric geometries the cross spread is most used. Recent developments show a marked increase of the number of available channels, nowadays 100,000+. This enables the use of point receivers, meaning that every geophone has its output recorded, instead of being in an array which is summed before sending the data.Geophones are ever increasing in capability, seeing an increase to lower corner frequencies and the use of batteries and GPS systems to make them cableless and suitable for point-receiver recording. MEMS are new on the market, with a noise floor of < 15 ng-per-square-root-of-Herz being available. The two main quantities which are not specified in the manufacturers’ data sheets of geophones, but which are highly relevant to the user are the noise floor and the dynamic operating range.The case study shows that most surveys nowadays are 3D cross-spread surveys with a source and receiver spacing of approximately 40m each.The most used data format in the geophysical industry is SEG-Y.A synthetic subsurface model has been generated, through which shots have been modelled. Using real field data from Saudi Aramco tests have been conducted to both increase the shot and receiver interval, and to add various amounts of noise to this data. The shots have subsequently been migrated to generate a subsurface image, for which two methods of quality control have been devised: the correlation of the image with respect to the full, no-noise model, and the number of lines as detected through the Canny edge method. The quantity spatial interval over SNR has been used to create a function for both QC methods; Corr = 117.862/( SI/SNR + 114.650), and Nl proportional to (SI/SNR)^1/4 for the number of lines. The fit for the number of lines has a very low R-squared, hence a new method must be sought for quality control.For the given model SI/SNR = 75 has been selected as a maximum for this ratio to obtain a migrated result of sufficient quality for it to be able to be interpreted.