Описана информационная технология прогнозной оценки нефтегазоносности территории. Предложен метод опорных векторов для классификации векторов наиболее информативных признаков с учетом процедуры валидации. Разработан новый датчик для гравиметрических измерений, основанный на конкурентоспособном адаптивном криогенном чувствительном элементе. Получены оценки вероятности нефтегазоносности территорий по материалам дистанционных исследований. Комплексирование материалов дистанционных исследований с данными геолого-геофизических исследований повышает достоверность и эффективность результатов прогноза.
Описано інформаційну технологію прогнозної оцінки нафтогазоносності території. Запропоновано метод опорних векторів для класифікації векторів найінформативніших ознак з урахуванням процедури валідації. Розроблено новий датчик для гравіметричних вимірювань, що ґрунтується на використанні конкурентоспроможного адаптивного кріогенного чутливого елемента. Отримано оцінки ймовірності нафтогазоносності територій за матеріалами дистанційних досліджень. Комплексування матеріалів дистанційних досліджень з даними геолого-геофізичних досліджень підвищує вірогідність і ефективність результатів прогнозу.
Purpose. Remote sensing (RS) began in the early 60s with the development of image processing of satellite imagery. Wide use of radiometers and hyperspectrometers in RS led to the accumulation of huge volume of experimental data that can now be used for remote detection of minerals. Imaging spectrometry data or hyperspectral imagery acquired by airborne systems have been used in geologic since the early 1980’s and represent a mature technology. The solar spectral range 0,4–2,5 µm provides abundant information about hydroxyl-bearing minerals, sulfates and carbonates common to many geologic units and hydrothermal alteration assemblages. The purpose of this paper is to show the feasibility of hybrid information technology and new sensors for identification of oil and gas.
Design/methodology/approach. We propose to combine hyperspectral, spectral and gravimetric data for oil forecasting using most informative parameters and the SVM-method. Behind the method is the idea of intelligent analysis based on models. Guided by this methodology, we demonstrate some possibilities involving four types of data such as hyperspectral, gravimetric, seismic, and geological data. Our method is also based on Spectral Angle Mapper (SAM) as a tool for matching the separated and members with pure spectra from databases. The analysis of SAM is provided in three spectral regions: VIS, IR and VIS+IR combined. Three kinds of available influences can be analysed: additive noise, constant offset, and slant offset.
Findings. This paper presents an overview of support vector machines (SVM) as one of the most promising intelligent techniques for data analysis, as theoretical approaches and sophisticated applications developed for various research areas and problem domains. It is an attempt to provide a survey of the applications of SVM for oil and gas exploration. The applications of SVM have been grouped and summarized in the different areas of the exploration phase, which can be used as a guide to assess the effectiveness of SVM as against other data mining algorithms. The study introduces an image specific algorithm for oil identification and discusses the implications for geology. Based on the hydrocarbon infiltration theory, gravimetric data, the analysis of crude oil in soil, spectral data of crude oil in sea water, and Hyperion hyperspectral remote sensing images were used to develop the synergetic approach to oil-gas exploration.
Practical value/implications. Our methodology proves to be practical for thorough data analysis in the exploration and production of oil and gas. The results indicate that the area of the oil-gas reservoir could be delimited in two ways: a) directly, by the absorption bands near 1730 nm in Hyperion image; b) indirectly, by using Linear Spectral Unmixing (LSU) and Spectral Angle Matching (SAM) of alteration mineral (e.g. kaolinite, illite). In addition, combined with the optimal bands in the region of visible/near-infrared, SAM can be used to extract the thin oil slick of microseepage.
