Study On Monitoring Of Territory Development Using Remote Sensing Time Series

Oriented to the demand for territory development monitoring from the viewpoint of integration of sea and land,and to provide the technical examples for territory development monitoring under different situations,the remote sensing time series technology was used to monitor two hot topics in land area and sea area respectively,aiming at their unique features of time series.The technology for extracting the construction land taking up farmland was implemented in the research of land area,to serve the monitoring and evaluation of geographical conditions.The technology for change detection of the construction conditions in islands and reefs was implemented in the research of sea area,to serve the real-time monitoring and analysis of the islands and reefs in the South China Sea.The pixel level remote sensing time series were constructed and reconstructed for follow-up researches.The information of construction land taking up farmland from 2000 to 2015 in the four urban agglomeration study areas were extracted by using 250 m MODIS NDVI time series.And the construction condition in Meiji,Zhubi and Nanwei islands were monitored using Landsat 8 images.The main research contents and conclusions are listed as follow:(1)The construction and reconstruction of pixel-level remote sensing time series.After comparing several methods of time series reconstruction,the discrete Fourier transform method was selected to reduce the noise.The time series after reconstruction obtained obvious smoothing effect,at the same time,the effective time series characteristic of different land use types were retained.The time series of the same land use types were highly similar,while the time series of different land use types have their distinct characteristics,which provided a good data base for the follow-up researches.(2)The technology of extraction of construction land taking up farmland based on the spatio-temporal similarity of time series.A spatio-temporal similarity analysis method was proposed,which combined the similarity of time series and the spatial neighborhood similarity of pixels,considering both the time series and spatial characteristics.The improved dynamic time warping algorithm was used to calculate the spatio-temporal similarity between the time series of to-be-extracted pixel and the standard reference time series of construction land taking up farmland.The threshold segmentation of dynamic time warping value was used to extract the construction land taking up farmland from the MODIS NDVI time series.The four study areas including the Yangtze River Delta Urban Agglomerations,the Pearl River Delta Urban Agglomerations,Beijing-Tianj in-Hebei Urban Agglomerations,and Chang-Zhu-Tan Urban Agglomerations,whose natural,social and economic backgrounds differ from each other.Their Hellden accuracy of extraction reached 88.68%,85.38%,90.83%and 92.57%respectively,which has achieved a high accuracy,and proved the applicability of this method.(3)Monitoring the construction conditions of islands and reefs in the South China Sea,based on the morphological and structural characteristics of time series.Differs from the research that aim at a certain land use and land cover change type,a method for monitoring the construction conditions of islands and reefs was put forward,which involved a variety of land use types.The method extracted the change nodes of time series,according to the changing characteristics of pixel-level remote sensing time series,based on the morphological and structural characteristics of time series.At the same time,the distinction between noise and suspected change pixels were also considered.In addition,spatial neighborhood relations were used to constrain the results,so as to achieve the purpose of extracting the changing time nodes.Meiji,Zhubi,and Nanwei islands were selected as study areas,whose Hellden accuracy of results reached 92.75%,91.85%,and 95.60%,respectively.The results showed that this method obtained good results in different study areas with different construction speed and scale,proved the feasibility of this method.To sum up,this study explored the application potential of remote sensing time series in the field of monitoring of territory development.To meet the demand for monitoring of both land area and sea area,the spatio-temporal similarity analysis of NDVI time series was proposed to extract the construction land taking up farmland,and the construction conditions of islands and reefs in the South China Sea was monitored by extracting the change nodes of time series,based on the morphological and structural characteristics of time series.The technology of remote sensing time series had achived good results in different study areas,which indicated the feasibility and high applicability of these methods in the field of monitoring of territory development.The remote sensing time series technology could meet the demand for monitoring in different situations,and provide technical support for the monitoring of territory development.