• 地理教学 •

### 基于Landsat遥感影像的海上油气平台提取与监测

1. （1．a．宁波市阿拉图数字科技中心，b．宁波市测绘设计研究院，浙江 宁波 315040；2．南京大学 中国南海研究协同创新中心，南京 210023；3．国家测绘地理信息局第一航测遥感院，陕西 西安 710054）
• 出版日期:2017-01-05 发布日期:2017-01-05
• 作者简介:赵赛帅（1990―），女，浙江人，助理工程师，硕士，研究方向为遥感与GIS，（E-mail）tata-0613@163.com
• 基金资助:
国家重点研发项目（2016YFB0501502）

### Extraction and Monitoring of Offshore Oil and Gas Platforms Based on Landsat Imagery

ZHAO Saishuai1a,2，SUN Chao2，WANG Haijiang1b，CHENG Wangyu3

1. （1．a．Ningbo Alatu Digital Science and Technology Center；b．Ningbo Institute of Surveying and Mapping，Ningbo 315040，China；2．Collaborative Innovation Center for South China Sea Research，Nanjing University，Nanjing 210023，China；3．The First Institute of Aero-Photogrammetry and Remote Sensing，NASMG，Xi’an 710054，China）
• Online:2017-01-05 Published:2017-01-05

Abstract: The monitoring of the development of offshore oil and gas platforms has great economic and military significance. According to the characteristics of remote sensing image data, a method for extracting and monitoring offshore oil and gas platforms was proposed based on the Landsat series optical images with medium resolution. Firstly, series pretreatments including geometric correction, radiometric correction, image enhancement, and area mask were performed on the TM/ETM+/OLI images. Secondly, images were traversed using multiple sliding windows with dynamic thresholds to recognize suspected oil and gas platforms. Finally, oil and gas platforms were screened by laminating 3 pixel-level images of consecutive periods to get rid of false alarm interference of thin clouds and the ships. In this work, the South China Sea at the junction of Malaysia and Brunei was studied with a total of 21 Landsat images from 1991 to 2016. The historical development process of the oil and gas platforms in the area in recent 25 years was dynamically monitored, and a total of 66 oil and gas platforms were detected using the proposed method. To evaluate the accuracy of the results, the oil and gas platforms in ALOS and PALSAR images were manually identified. The results showed that the platform extraction rate was 82.67% and the correction rate was 86.11%. Based on the features of highlight and invariant position of targets, the method was able to detect oil and gas platforms from complex backgrounds. It was found that distribution of oil and gas platforms could be determined by superimposing 3 images of good quality and of consecutive time series. As Landsat satellite images of a high time resolution for a long time span are available in the archives, it is possible to conduct dynamic monitoring of the historical development process of the oil and gas platforms. This effectively compensates for that SAR images cannot be used in long-term dynamic monitoring. However, as in this method, false alarm caused by clouds is mainly removed by image overlaying, when there are many clouds in the optical images, not much area can be used for monitoring, making image utilization rate low and the correction rate of oil and gas platform recognition reduced. Therefore, there are some deficiencies with the monitoring method as it is influenced by the image cloud. Images with fewer clouds should be chosen as far as possible to monitor oil and gas platforms. Monitoring results in the South China Sea at the junction of Malaysia and Brunei showed that: offshore oil and gas platforms were mainly developed in 1990s; after that, not many offshore oil and gas platforms were developed, but the deep sea far away from the land was focused on for oil and gas platform development.