| Surface water is an important part of the hydrosphere,it distribut on the earth surface and mainly forms of rivers,lakes,wetlands.Affected by different natural conditions and human factors,surface water changing its reserves and distribution all the time.The traditional surface water information monitoring mainly based on investigation,which has low efficiency,long cycle and high cost,and it is difficult to collect surface water information rapidly in large scale.At present the popularization and application of remote sensing technology provides a wide range of real-time image data,which make it possible to extract and monitor surface water automaticly.The terrain of Fujian is dominated by mountains and hills,and the mountains are surrounded by a wide shadow,which makes the surface water,especially the small water body,difficult to be extracted accurately and completely.Therefore,based on the multi-temporal remote sensing data,we developed a water index suitable for the mountainous terrain and a index enhance narrow water features,researched on automatic extraction and updating of surface water information in Fujian province,the main research contents and conclusions are as follows:(1)The slope adjusted water index(SAWI).A water index with remarkable characteristics and high precision is the premise and foundation of automatic extraction and updating of the surface water body information,therefore,we present a new water index in this paper.The experimental results showed that SAWI can not only reflect the characteristics of surface water,but also suppress the shadow effect of mountain terrain effectively,moreover,it was suitable for plain area and has strong adaptability to complex terrain.(2)Automatic extraction of surface water.Based on the morphology and distribution characteristics of surface water in Fujian province,the surface water information can be divided into open water and narrow water.Then,these two types of water information were extracted from center to boundary.To assist in extracting narrow water,we used the method of morphological White Top-Hat to establish a narrow water index(MNWI),highlighted linear features and enhanced narrow water features.The experimental results demonstrated that the surface water information could be extracted in a high accuracy by our method,in open areas,the boundary information of open water was extracted exactly;in narrow areas,the MNWI makes the narrow water easier to distinguish.In addition,our method also have a good applicability in large areas,the overall accuracy of surface water information extraction can reach 96.61%in Fujian province.(3)Automatic updating of the surface water information.Based on the surface water variation characteristics of Fujian,a method of automatic updating surface water information was studied in this part.With the reference of correct water information at one temporal,we transfer learning the sample information from different temporal images in order to achieve a higher accuracy.The experimental results showed that,the surface water of Fujian was able to be updated accurately by our method,the total time costs is only 2 hours and 38 minutes,the overall accuracy and Kappa coefficient can reach 96.61%and 0.932,which showed prominent advantage on efficiency and precision.Studies in this paper mainly focused on the automatic extraction and updating of surface water information in Fujian.Our method extracte surface water information in large area rapidly,accurately and completely,improve the efficiency and precision of surface water extraction,dynamic monitor the change of surface water information,thus be able to provide timely and reliable data for disaster emergency management and research on regional water environment. |
PDF Full Text Request