Study On Dynamic Monitoring Of Wetland Changes In Daqing Oilfield

The contradictory relationship between resource exploitation and environmental protection has always been a hot research topic.With the continuous development of social economy,the requirements of ecological environment construction are gradually put on the agenda.Daqing area is a typical area where the contradiction exists.Through high-resolution images,we find that some oil Wells in Daqing area are built around and inside the wetland,So it is very important to carry out effective dynamic monitoring of wetlands in resource exploitation activities.In this paper,Daqing area is taken as the research area to carry out wetland dynamic monitoring work based on time series.On the platform of Google Earth engine,we make full use of abundant satellite data resources to carry out wetland change monitoring in Daqing area from 1990 to 2019.By comparing and analyzing the threshold method and the wetland classification method of machine learning,Otsu’s method is suitable for binary classification task of long time series wetland information extraction,and Random Forest algorithm is suitable for land use type classification in the year of abnormal change.In this paper,the wetland information of Daqing area from 1990 to 2019 is extracted by threshold method.And through the year-on-year iteration and statistical analysis to explore the spatial-temporal change of wetlands in the study area,and through the mutual verification of landtrendr algorithm,it finds out that the change segmentation point is in 2001.Finally,the Random Forest and Support Vector Machine algorithm are used to classify the land use in Daqing in 1990,2001 and 2019,and compare the three stages of land use classification map to explore the factors affecting wetland change.The results of this paper are as follows:(1)The results show that SWI and MNDWI can extract wetland accurately in visual interpretation analysis.The AUC values of ROC curve in threshold analysis accuracy evaluation are 0.998 and 0.983,respectively.The analysis process is based on the conclusion of comparison between different data sets and different data quality on Google Earth Engine platform,Therefore,a wetland extraction index system suitable for Google Earth Engine is constructed.Finally,wetland information extraction in the study area over 30 years is completed based on SWI index.(2)In the past 30 years,the wetland area in Daqing has changed periodically,with the first cycle from 1990 to 2001 and the second cycle from 2002 to 2019,The peak value of wetland area appears in each cycle,which is 1998 and 2005 respectively.The wetland area value in these two years is far greater than that in other years in the cycle.(3)Through the iterative analysis of time series wetland information,a map of wetland information change for many years is formed.The wetland change process and active area of wetland change can be effectively expressed by visualization technology 。 According to the statistics,Zhaoyuan County,salbert Mongolian Autonomous County,Datong District and Zhaozhou County in the study area are divided into the first phase of wetland change cycle(the period of wetland change is more frequent from 1990 to 2001),Lindian County,Ranghulu District,Sartu District,Longfeng District and Honggang district are classified as the second phase of wetland change cycle(the period of wetland change is more frequent from 2002 to 2019).According to the administrative boundaries,the wetland change activity in the study area can be divided into 9 categories.Sartu district is the ninth category,which is the most active administrative region in the past 30 years.The next descending order is as follows: Longfeng District,Honggang District,Zhaoyuan County,Lindian District,salbert Mongolian Autonomous County,Ranghulu District,Datong District,Zhaozhou county.(4)Support Vector Machine and Random Forest machine learning algorithm are used to carry out multi-classification tasks of land use types in Daqing area.Comparing the differences of land use type change among three time nodes in 1990,2001 and 2019,and analyzes the wetland change factors in Daqing area with the help of historical statistical yearbook data of the research area,and finds that the wetland in the oilfield exploitation area decreases with the increase of oil well density..