Dynamic Monitoring Of Oil/Gas Development In The South China Sea Based On Long-Period Time-Series And Multi-Source Remote Sensing Images

With the rapid growth of global population and economy,offshore oil and gas exploration has become a major trend of international energy development and as a primary accelerator for global energy growth.South China Sea(SCS)is not only one of the most abundant area for hydrocarbon energy storage,but also the important energy backup for our country in the future.The critically strategic value of oil/gas resources has prompted the SCS to the core area full of political contradiction and interests tradeoff.As a result,the SCS related disputes have become increasingly intense,complicated and internationalized.Till now,the information on transregional offshore oil/gas production disperse among different marine oil/gas manufacturer and ocean energy administrator.For the reasons of business secrecy and national security,aforementioned departments reluctant to share the oil/gas information or make it available publically.At present,few accessible oil/gas statistics data exists at energy website of the U.S.Energy Information Administration(EIA)and British Petroleum(BP)as the format of statistical yearbook.However,the national-scale statistical yearbook is difficult to spatialize,which impede the understanding on spatio-temporal changes of oil/gas production in the ocean.Few accessible oil/gas spatial databases are sourced from the OSlo and PARis conventions(OSPAR),which provided the geo-locations of offshore installations in the North Sea,and Bureau of Safety and Environmental Enforcement(BSEE),which recorded the geo-location of oil/gas platforms in the North Sea and Gulf of Mexico.But for the vast and controversial ocean like SCS,the island effect of the information on transregional offshore oil/gas production has led to data deficiency in oil/gas resources and understanding margin of the whole process of oil/gas development in the SCS,which has seriously hindered maritime safety management and energy security maintenance.Remote sensing is capable of overcoming the shortcomings such as area inaccessibility,and providing opportunities for information collection of offshore oil/gas development such as the SCS.The ’vast extent of the SCS as well as the long-period duration of oil/gas production makes it hard to monitor the entire SCS in time by mono-source remote sensing images;thus,combination of multi-source remote sensing images becomes an alternative and effective way.However,the following problems we confront when multi-source remote sensing technology are introduced:(ⅰ)How to register multi-source images with different geo-location accuracies on the premise of lacking Ground Control Points?(ⅱ)How to accurately identify offshore oil/gas platforms with tiny size and dim feature under the complex background of vast sea?(ⅲ)How to determine status and attribute information of offshore platforms based on platform geo-location and image series?And(ⅳ)How to effectively correlate the relationship between offshore oil production and feature space of remote sensing?Aiming at the aforementioned difficulties from dynamic monitoring of oil/gas development by remote sensing,this paper combined long-period time-series strategy with multi-source images,including low-medium-high resolutions,optical-radar,and daytime-nighttime data(i.e.,optical images:Landsat-4/5 TM during 1992-2011,Landsat-7 ETM+ during 1999-2013,Landsat-8 OLI during 2013-2016;SAR images:JERS-1 SAR Mosaic during 1993-1998,ALOS-1 PALSAR during 2006-2011;night-time light data:DMSP/OLS during 1992-2013;the total amout is greater than 87200 tiles).A series of methods were developed,including geo-location detection algorithm,status/attribute extraction method and oil production estimation model,to carry out integrated monitoring framework of "geo-loaction identification-status/attribute extraction-production estimation".And the history,status-quo and tendency of oil/gas resources development were analyzed as a dual-scale—the entire SCS and each countries neighboring the SCS.All the results are expected to support China’s marine rights and interests and offshore resources development.The main results are descripted as follows:(1)Created the geo-location identification method for offshore oil/gas platforms.Based on the position-and size-invariant characters of offshore oil/gas platforms,this study propused image time-series strategy,which first combined medium spatial resolution optical images(e.g.,Landsat-7 ETM+ and Landsat-8 OLI)and SAR images(e.g.,ALOS-1 PALSAR)with high geometric accuracy,to automatically identify platforms on a large scale with assistance of Order Statistics Filter(OSF)and Cloud Mask De-noising(CMD)technology.The image time-series strategy solved the problem that how to accurately discriminate tiny targets with weak reflection characterstics under strong false alarm background on remote sensing images.The geo-locations of totally 1143 offshore platforms during 1992-2016 in the SCS were identified by merging time-series platform results from all sensors.With the interpretation validation by high spatial resolution images,the results show that the accuracy for geo-location of offshore platform is 93.5%,the false positive and false negative are merely 4.2%and 2.3%,respectively.(2)Proposed the status/attribute extraction method for offshore oil/gas platforms.To overcome the difficulty that how to register multi-source images(e.g.,Landsat-4/5 TM,and JERS-1 SAR Mosaic)with low geometric accuracy on the condition of a vast area without/with less GCPs,this study propused a corss geo-correction strategy using platforms derived from high-geometric accuracy images as GCPs to rectify low-geometric accuracy image.Afterwards,stable statistical features of long-period time-series,integrated with methods of sequential pattern recognition and empirical correction function,were adopted to extract multi-dimensional status/attribute information for offshore oil/gas platforms such as operation lifetime and status,platform size and type,working depth.The results show that the uncertainty of status/attribute extraction can be effectively offset by introducing the stable statistical features of image time-series instead of the mono-phased features:operation status error of 80.5%platforms is less than 1 year and operation status error of 87.5%platforms is less than 2 years;the absolute error percent of platform size simulation is less than 30%;the overall accuracy of platform type determination is 89.6%,with the Kappa coefficient of 0.791.(3)Explored the oil production estimation model for offshore oil/gas platforms.Highlighting the differences of spatial light intensity and temporal light stability on Night-Time Light(NTL)DMSP/OLS images,this study accurately picked out target lights related to offshore oil production under the feature space constructed by spatio-temporal statistics of NTL.This approach solved the problem that how to discriminate the night lights confusion sourced from different work modes(e.g.,offshore oil/gas production,fishery working,and ephemeral noises)on NTL data.Using the DMSP/OLS data with global relative calibration as a link and the sum of light intensity corresponding offshore oil production as a measure,an Oil Production Estimation Model(OPEM)of the North Sea was subsequently established and transplated to the SCS.The results show that the data range of the OPEM covers 0.1-2×10~5 ksm~3 and its correlation coefficient R2 reaches 0.946.Currently,the OPEM is favorable to estimate offshore production for the entire SCS,evidenced by the absolute error percent merely 2.92%,rather than apart neighboring countries,with the average absolute error percent up to 28.93%.(4)Clarified the whole process of offshore oil/gas development in the SCS.The general trends for offshore oil/gas development presented rapid establishment,platform miniaturization,deep-sea exploration,and confict intensification:(ⅰ)Rapid establishment.The total number of offshore platforms linearly increased from 153 to 996 during 1992-2016,with an annual increment of 35.The offshore oil production raised from 0.383 X 10~5 ksm~3 of 1992 to 1.066×10~5 ksm~3 of 2013.(ⅱ)Platform miniaturization.The proportion of small platforms(area below 1600 m~2)increased from 37.2%to 63.5%during 1992-2016,which led to the obvious decline in a’verage platform size from 6852 m~2 to 2930 m~2.(ⅲ)Deep-sea exploration.The average distance between platforms to shoreline increased from 71.9 km to 101.5 km during 1992-2016;the average operational depth of platforms raised from 42.6 in to 59.1 m.with the remarkable increment in maximum operational depth from 122 m(shallow water)to 1471 m(almost ultra-deep water).(ⅳ)Confict intensification.For the disputed regions of the SCS,the total number of offshore platforms increased tremendously from 1 to 89 during 1992-2016,with the mean annual growth rate increasing from 1,1 before 2005 to 7.2 after 2005.Up to 2013,the annual oil production capacity of these regions approximately reached 5475 ksm~3.In summary,integrating multi-source images with long-period time-series analysis,this paper overcame a series of challenges such as geometric registration,geo-location identification,status/attribute acquisition,and oil production estimation.Furthermore,the spatio-temporal dataset fully covering the whole area of the SCS(Gulf of Thailand included)was established to lift the veil on spatial expansion and illegal exploitation of offshore oil/gas resource among the neighboring countries of the SCS.All the methods and results are expected to support decision-making on offshore energy development,maritime security management and sovereignty maintenance in the SCS.