Comparative Study On Fusion And Void Repair Methods Of Typical Open DEM Data

The development of DEM data on a global scale is of guiding significance for the study of global terrain and landforms.At present,the classic open DEM data sets with large coverage and high resolution are mainly ASTER GDEM series and AW3D30 series data.The data set does not contain data vacancies between 60°N-60°S.In areas outside 60°N-60°S,ASTER GDEM v3 data is continuously distributed,and AW3D30 v2.2 data contains more data vacancies.Therefore,in order to obtain a high-quality and continuously distributed DEM data set on a global scale,in this study,60°N north is used as the study area,AW3D30 v2.2 data is the main data source,and ASTER GDEM v3 data is the auxiliary data source.Two kinds of data sources are proposed.Different data production methods.1)Using Cristovalja and surrounding areas in Russia as the research sample area,the data fusion of ASTER GDEM v3 and AW3D30 v2.2 data at 60°N position was performed.In the vicinity of 60°N,different types of inverse distance weighting methods are used to fuse the two data to obtain a transition zone,which connects AW3D30 v2.2 to the south and ASTER GDEM v3 to the north;2)The area north of 60°N,to as an example of the Iceland region,the direct mosaic,indirect mosaic,and Gaussian smoothing are used to repair the void area existing in the AW3D30 v2.2 data.Among them,the indirect mosaic adopts the average elevation difference of ASTER GDEM v3 and AW3D30 v2.2 in different elevation and slope grades to establish the elevation difference model and repair the data hole existing in the AW3D30 v2.2 data.In the calculated curve edges,during the 60°N position data fusion process,the average value of the transition zone span obtained by the two data fusions is 107.5 pixels,and the average span is 3052.55 m.It can be obtained by visual interpretation that the direct fusion of ASTER GDEM v3 and AW3D30 v2.2 will produce a clear boundary,and each inverse distance weighted result does not show a clear boundary at 60°N;profile analysis is available,using Gaussian inverse distance For the transition data obtained by weighted interpolation,the average annual elevation difference between the transition data and ASTER GDEM v3 is 1.3 m at the northern boundary,and the average error between the transition data and AW3D30 v2.2 is basically 0 at the southern boundary;The information entropy is available.The Gaussian inverse distance weighted average surface error information entropy is 3.22,and the information entropy in gentle slopes,slopes,steep slopes,and steep slopes is 2.99,3.05,3.70,and 3.16,respectively.the trend of.In the AW3D30 v2.2 data repair,visual interpretation can be obtained that direct mosaic,indirect mosaic,and Gaussian smoothing all achieve the repair of the cavity area data;from the section elevation of the cavity area boundary,the cavity area data obtained by direct mosaic There is a large elevation difference with AW3D30 v2.2.Indirect mosaic and Gaussian smoothing do not show a large elevation difference at the boundary of the cavity area;from the perspective of the profile elevation inside the cavity area,direct mosaic and indirect mosaic and Gaussian smoothing The elevation difference inside the cavity area is basically consistent with the elevation difference of the boundary;from the perspective of overall smoothness,the overall smoothness of Gaussian smoothing is better than that of direct mosaic and indirect mosaic;in the cavity area,the average of direct mosaic,indirect mosaic,and Gaussian smoothing is used.The errors are 1.0 m,1.0 m,and 13.9 m,respectively,and the root mean square error of Gaussian smoothness is the lowest,which is 14.7 m.It can be obtained that the Gaussian inverse distance weighting method is used at 60°N to fuse ASTER GDEM v3 and AW3D30 v2.2 data to generate data with a wide coverage and high data quality.It is a better choice to use Gaussian at 60°N.The result of smoothing the repair of the holes in the AW3D30 v2.2 data is better.Therefore,it is feasible to obtain DEM data with wider coverage and higher resolution through data fusion and hole repair.