탄성파속성과 인공신경망을 이용한 울릉분지 가스하이드레이트 포화율 추정

Abstract

Among the unconventional natural resources, gas hydrate has recently received much attention as a promising potential energy source. To produce gas from gas hydrate-bearing sediments, the distribution and the saturation which are directly connected to the volume of gas hydrate should be estimated preferentially at the initial stage of the development for production. In most cases, the distribution of gas hydrate can be identified by using seismic indicators including bottom simulating reflector (BSR) and chimney/column structure which represent indirectly the presentence of gas hydrate. However, these indicators have the restriction on the application because there can be used only when these are displayed on the seismic image section. Meanwhile, since the saturation of gas hydrate is generally calculated by using well logs, the information is limited to the well location. In order to overcome the restriction, the methods of seismic inversion and probabilistic neural network (PNN) are required. Seismic inversion enables the identification of gas hydrate reservoir even if seismic indicators do not exist, and PNN makes it possible to predict the gas hydrate saturation of interest away from wells by combining seismic multi-attributes. In this study, to estimate the distribution and the saturation of gas hydrate broadly distributed in Ulleung basin of East Sea, two types of seismic inversions were first used and then the following seismic attributes showing their own unique features in hydrated sediments were extracted; AI, SI, EI (22.5°), Vp/Vs ratio, , and . Gas hydrate-bearing sediments showed high AI, high SI, high EI (22.5°), low Vp/Vs ratio, high and high compared to those of the surrounding sediments. The sediments containing free gas showed low AI, low SI, low EI (22.5°), high Vp/Vs ratio, low and low due to the phase transition from gas hydrate to gas. Thus, through the investigation of seismic attributes, the distribution of gas hydrate can be estimated even if seismic indicators are not present on the seismic profile. On the other hand, by training these already-extracted seismic attributes, standard seismic attributes and TPBE-derived saturation of gas hydrate which has physically high correlation to seismic attributes, the saturation of gas hydrate away from the wells was estimated based on PNN predictions. For the validation of the predicted saturation, cross-validation method of wells was used. The average correlation coefficient between predicted saturation and actual saturation logs at the UBGH-09 and UBGH2-10 wells was 82.6%. In addition, in the estimated saturation section of gas hydrate, a relative high saturation region of gas hydrate corresponded well to the gas hydrate occurrence zone of each well.