| In recent years,countries and organizations worldwide have successively applied different remote sensing imagings and classification techniques to conduct numerous studies on remote sensing mapping and change characteristics with land use/land cover as the main data type,and numerous land use/land cover datasets have been formed at global and regional scales.However,with the development of relevant research,the shortcomings and limitations of these datasets are gradually highlighted,that is,due to the use of different data sources,classification systems and classification techniques,these land use/land cover datasets have plenty of problems,such as low accuracy,poor consistency,and significant differences from statistical data.Given this,in the context of the coexistence of multi-source data and the increase in fusion research,it is of great significance to carry out research on remote sensing mapping and change characteristics of land use/land cover in the global coastal area based on the multi-source data fusion,which is a very important indicator of coastal environmental change,to reveal coastal ecosystem succession,and to explore the effects of climate change on the coastal areas.According to the geographical characteristics of the global coastal area,this study defines the geographical scope of the global coastal area,establishes one kind of remote sensing classification system of land use/land cover in the global coastal area,and then assesses the consistency of multi-source land use/land cover datasets in the global coastal area.On this basis,the three land use/land cover datasets(MCD12Q1,CCI-LC,and Globe Land30)with relatively higher mapping accuracy are selected as the source datasets,and a data fusion method based on agreement analysis and fuzzy-set theory is adopted to acquire the classified information of land use/land cover in the global coastal area in 2000 and 2010,and to reveal change characteristics and their driving mechanism of land use/land cover in the global coastal area.Then,in combination with the GIMMS NDVI3 g data and several representative secondary types of wetland data at the global scale,the change characteristics of vegetation cover and the remote sensing mapping of potential distribution areas of wetland in the global coastal area are carried out.The main conclusions are as follows:(1)The scope of the global coastal area is defined as follows: In order to adapt to the coastal area survey,the land areas extending to a certain range above the coastline and the sea areas extending to a certain range towards the ocean are taken as the study area(that is,the global coastal area),including the island coasts but excluding the coasts of inland lakes;where the land boundary is defined as the coastline extending 100 km to the land side,and the sea boundary is the union of the-100 m isobath and the 10 km offshore buffer area.(2)The remote sensing classification system for land use/land cover in the global coastal area is established as follows: This research elaborates the progress of land use/land cover classification systems made in regional and global scales,and then expounds the progress of land use/land cover and wetland resource classification systems on the coastal areas.On that basis,this research proposes a land use/land cover remote sensing classification system suitable for the global coastal area.Specifically,the classification system includes six primary types and 20 secondary types as well as 29 tertiary types,which holds a total of 43 basic types;moreover,it covers the land use/land cover types of the coastal area in a more comprehensive way,and also fully emphasizes the wetland resources on the coastal areas.(3)Data source selection and consistency analysis before multi-source data fusion are as follows: the multi-source datasets have a good consistency for description of land use/land cover structure in the global coastal area,but in detail,there are also deviations in area;the correlation coefficient,overall accuracy and Kappa coefficient of each dataset portfolio in 2000 are between 0.8924~0.9974,69.54%~80.18% and 0.6005~0.7337,respectively,and then these values in 2010 are between 0.8814~0.9869,67.46%~81.50% and 0.5748~0.7505,respectively;for the spatial confusion between any two different datasets,grassland,shrubland and wetland have the highest mix-up ratios,followed by farmland and artificial surface,and lastly by forest,unused land and water;in the multisource comparison,78.51% of the land in 2000 and 88.64% of the land in 2010 have a high consistency in the global coastal area.(4)The remote sensing mapping and precision evaluation for land use/land cover in the global coastal area based on the multi-source data fusion are as follows: compared with the above-mentionedthree sets of input data sources,the overall accuracy,Kappa coefficient and average overall consistency of the fusion dataset in 2000 are the highest,which are 86.48%,0.8097 and 82.83%,respectively,and these values in 2010 are also the highest,which are 89.21%,0.8476 and 83.45%,respectively;the producer precision and the user precision of each type in the fusion datasets of 2000 and 2010 are at least higher than those of any two sets of the input data sources,and the average type-specific consistencies in the fusion datasets of 2000 and 2010 improve by 1.23%~29.37% and1.33%~33.13%,respectively.(5)The variation characteristics of vegetation cover in the global coastal area and its correlation with land cover are analyzed as follows: Desert belt is mainly perennial non-vegetation or low vegetation cover areas,tundra belt is primarily moderate or high vegetation cover areas,and forest belt is mostly dense vegetation cover area;the intraannual variations of vegetation cover show a “?” shaped curve in the study area,while inter-annual variations reveal a fluctuating but generally slowly increasing trend during the entire study period;on the monthly,seasonal and annual scales,the overall trend of vegetation cover change is increasing in the study area,while areas with the decreasing trend are relatively few;change trends of vegetation cover in most areas have relatively strong positive persistence in the furture;the increasing trend of relatively high-latitude coastal tundra is extremely significant in the growing season;moreover,the decreasing trend of vegetation cover is relatively significant in the “coastal urban agglomeration”and the “periphery of desert belt”.(6)The remote sensing mapping and precision evaluation of potential distribution areas of global coastal wetlands are as follows: On the basis of in-depth analysis of all kinds of wetland classification system and classification refinement method,a dataset integration method based on category replacement is adopted to realize the integration of several source datasets of wetland secondary types,such as intertidal mudflats,coral reefs,mangroves,rivers and ditches,lakes and pits,freshwater marshes,and seagrass beds.In total,the accuracy evaluation of the integration results shows that: as far as it goes,the accuracy of the mapping results is not ideal enough and needs to be improved.(7)The changes and driving forces of land use/land cover in the global coastal area are analyzed as follows: in 2000~2010,land use/land cover change in the global coastal area is dominated by shrubland expansion and bare land shrinkage,followed by forest shrinkage and permanent ice/snow expansion,lastly by farmland shrinkage and grassland shrinkage,and furthermore,wetland shrinkage and artificial surface expansion are the least;the area of interconversion between 8 types is relatively small,and thereinto,the dominant variation types are the interconversion of farmland,forest and grassland,the interconversion of grassland and shrubland,forest that change into shrubland,and wetland that change into grassland,respectively.Then,although natural driving forces profoundly affect the macro pattern of land use/land cover change,humanistic driving forces are the direct causes of land use/land cover change in the global coastal area from2000 to 2010. |
PDF Full Text Request