DEM and digital geographic base map making

The DEM in the survey area consists of 17 DEM data and 1:50000 map sheet. The topographic map of 17 is scanned, corrected, registered and spliced in ENVI image processing software to form a complete topographic map of1:50,000 survey area, and then the topographic line is vectorized, combined with the 15m grid DEM generated by Japanese satellite ASTER stereo image and the DEM of some areas in China provided by National Geographic Information Center * * on MAPGIS software platform.

(2)1:Digital Geographic Basemap of 50,000 Survey Area

First of all, when vectorizing topographic contours, rivers, roads, peaks, elevation points, residential areas and other elements are also vectorized; Convert the completed 1:50000 DEM into grid data in Surfer format, and then use it in the map? Draw elevation contour map with elevation interval of 100m, 50m or 20m in GIS; Finally, the digital geographic base map of the investigation area is edited. The projection mode of this drawing is Gaussian projection, the central meridian is 8 1 E, and Beijing 54 coordinate system based on Krasovsky ellipsoid is adopted.

(3) 1: 1 DEM of survey area

1. Technical difficulties

High-precision DEM is the basis of quantitative remote sensing investigation and monitoring of 1: 1 10,000 disasters and geological environment. Making high-precision DEM in hilly areas is a technical difficulty at home and abroad. There are two main technical difficulties: first, there are few satellite data for establishing high-precision three-dimensional models; Second, there is a lack of technical methods to generate high-precision DEM in areas where the height difference fluctuates greatly.

2. Technical difficulties and operational procedures

(1) Searching for high-resolution satellite stereo image pairs

This project requires the establishment of 1 ~ 5m grid DEM, but the currently widely used 2.5m stereo image pair of SPOT-5 satellite cannot meet the accuracy requirements. After investigation, except SAR, only the stereo image pair of OrbView satellite in the United States can produce DEM with such high precision. After more than a year's efforts, satellite data were not obtained until June165438+1October 2006. OrbView-3 satellite is one of the earliest commercial satellites providing high-resolution images in the world. The orbit height of the satellite is 470km, the return visit period is less than 3 days, the spectral range of panchromatic band is 450-900nm, and the spatial resolution is1m. In this project, the 12 * * 6 image pair of OrbView satellite image data with 1m resolution is used to build a stereo model and generate DEM.

(2) Software platform

At first, VirtuoZo operation was tried, but the common VirtuoZo digital mapping system software did not support OrbView satellite images. After requesting technical assistance from VirtuoZo suppliers, we obtained the limited right to use VirtuoZoSeri software newly developed for western mapping, which can support OrbView satellite images.

ERDAS, ENVI and PHOTOSHOP are also used in this work.

(3) Three operation flow schemes and their comparison.

Survey area1:50,000 working DEM and digital geographic base map are completed for high-precision DEM. Because it is an exploratory work to make high-precision DEM in mountainous areas, we have designed three sets of workflows: ① Select plane control points from1:50,000 topographic map, and use the elevation determined by1:50,000 DEM to correct the DEM formed by orienting OrbView satellite stereo images with RSAT module; 2 correct Orb? Adjust the orientation by RSAT module through free network. Create DEM by viewing satellite stereo images, and then correct it through control points on topographic map; ③ There is no control point. According to the satellite orbit parameters, through free network adjustment, use RSAT module to orient OrbView satellite stereo image pairs and establish DEM, as shown in figure 1? 2.

Figure 1? 2. Three schemes of establishing 1: 1 10,000 DEM workflow

During the operation of "Scheme 1", the orientation error is very large, and the maximum orientation error is17.852 m. The reason is that the control point itself is too big to control when participating in orientation. The main factors affecting the accuracy of control points are: ① the error of raster topographic map, the control points are read on the corrected1:50,000 raster map, and the pixel scale of1:50,000 raster map is about 4m, so the accuracy is relatively low; 1:50000 map has been corrected one by one, but there will be a big error; As a geographical control, the time interval between map data and image data is more than 20 years. In this strongly weathered area, the terrain will change to some extent, so it is not easy to choose a point with the same name. (2) Terrain variation error. The survey area belongs to mountainous and canyon terrain, so it is difficult to find a relatively fixed reference terrain. Basically, the control points are selected by river. Due to the seasonal change of water surface and strong erosion, the edge or shape of the river has changed greatly in the past 20 years. ③ Conversion error and DEM error of two coordinate systems. Although each picture has its own conversion parameters, there are still conversion errors between different ellipsoid systems. Read the elevation of control points from the DEM provided by National Geographic Information Center. The grid spacing of this DEM is 25m, which is larger than that of 1: 1000.

In the second scheme, firstly, a set of orthographic images is generated by using the free network adjustment method of stereo image pair and digital photogrammetry. Using the latitude and longitude of the image itself, the topographic map is associated with the generated DOM position through coordinate transformation and displacement. Referring to the aster image map of this area, the same name points of the raster map and the image are found, and 54 plane coordinates of the selected control points are read. Then the 54 coordinates of the control point are converted into 80 coordinates, and the 80 coordinates of the control point are nested with the DEM with 1:50000 coordinates to read the elevation data of the control point. Although the control points are determined by this method, the accuracy of the results is still unqualified because of the large time difference between the above topographic map and the image data and the special terrain. The analysis results of control points show that the residual error after control points participate in orientation is much larger than that without control points, and the introduction of control points will increase the internal error in the working area.

Therefore, scheme 3-mainly using the orbit parameters of the satellite for control is finally adopted.

(4) Methods to improve the accuracy of DEM

The following solutions are adopted in this project: ① Quadratic polynomial correction method is adopted point by point (every grid point participates) to minimize the correction error; (2) In this alpine valley area, it is very difficult to select control points on topographic maps and image maps. Later, the ASTER color image in this area is used as an auxiliary reference point selection. When the control points are combined with DEM to read the elevation information of control points, the DEM corresponding to all control points is maximized as much as possible to reduce the error of manual selection of plane control points. ③ After creating the stereo model, you can see the generated stereo image under the Display Stereo toolbar, but because of the large terrain height difference, you can't display the stereo in the mapping module; In addition, the generated stereo model can't edit DEM, but it can automatically match DEM to generate orthographic images. All these problems were solved together with partners, and finally all the software and hardware problems were solved one by one.

(5) image processing

Image processing of ETM, SPOT, ASTER and CBERS-2 satellite data, including multi-spectral synthesis, data fusion, mosaic, geometric correction and image registration, is mainly carried out on ENVI, PCI and PHOTOSHOP platforms.

Before obtaining high-precision DEM, the correction of high-resolution image with ground resolution ≤ 1m is based on1:50,000 DEM, so its absolute accuracy is only1:50,000. 1: 1000 The accurate registration of high-precision orthophoto images and images of each phase is the basis and guarantee for quantitative interpretation and monitoring of landslides and geological environment. After the qualified 1: 1, DEM is established, the acquired 8 multi-spectral data of QUICKBIRD and Alos * * * from 2004 to 2007 are re-synthesized in the 3, 4 and 2 bands and fused with the panchromatic band, and all the images are corrected and registered with OrbView DOM( 1+0 phase).

(6) Interpretation and verification of human-computer interaction

Man-machine interaction remote sensing interpretation is based on the principle of landslide geology, and on the basis of qualified interpretation, it adopts man-machine interaction method to interpret and obtain the basic information of landslide and geological environment. Interpretation is mainly in MAPGIS, ENVI and PHOTO? Shop platform.

1:50,000 disaster and geological environment interpretation is based on 5m resolution SPOT-5 multispectral orthophoto image, with reference to ASTER, ETM and ALOS images. The geological work in this area is relatively low, and the only detailed data are1:250,000 Zada Sheet and Snow Mountain Sheet. However, according to the interview, due to the complex terrain and bad weather, the surveying and mapping work failed to reach the Parry River basin. For the remote sensing interpretation of this project, firstly, referring to the map and text description, combined with the characteristics of the image, the interpretation marks are established, and then the interpretation is carried out one by one according to the interpretation marks. After the preliminary interpretation was completed, I went to Tibet for on-site verification. Although it is June, it is necessary to cross many passes with an altitude of more than 5000 meters from Zhada to Bali River investigation area. The snow was so thick that although local migrant workers and horses were hired, they still failed to reach the Bali River Basin. Although the climate at the eastern and western ends of the Himalayas is very different, the terrain is basically symmetrical and similar, so we moved to Nangbawa Peak at the eastern end to investigate the topography and environment of glaciers and mudslides there. In addition, through visiting the local water conservancy and geological environment monitoring station of Parry River, we learned about the field situation, collected the field photos of Parry River, and verified the disasters and geological environment in the investigation area through the comparative interpretation of nearby satellite images. After the field verification, the disaster and geological environment in the whole area were further explained and analyzed.

(7) Geographic Information System and Spatial Analysis

Spatial analysis and calculation of the basic information obtained from the above interpretation in GIS system, including disaster types, natural and environmental analysis of key investigation areas, location, shape and scale estimation of disaster bodies; 1:Determination of gravity erosion type and location, scale calculation, risk assessment and analysis of its relationship with environment in 50,000 survey area. This work is mainly in MAPGIS, ARC? View and ENVI platform.

(8) Accuracy of results

1) 1: 1 10,000 remote sensing survey. The overall terrain difficulty of the survey area of this project should belong to the highest third-grade alpine zone, but the local landslide terrain is relatively flat. For multi-temporal landslide monitoring, it is necessary to carry out more rigorous geometric correction and registration of each phase image, so the error is required to be within1m. It should be noted that this is only the relative accuracy within the range of key areas, such as table 1? 2.

Table 1-2 DEM accuracy in key areas of this project: 1: 1 ten thousand.

It should also be noted that in the early stage of the project, 1: 1:50000 can't be used as the data source for establishing high-precision DEM, so 1:50000 DEM can only be established first. Although satellite data with a resolution of 0.6m was purchased in the corresponding key work area, the accuracy of correction and registration is still1:50,000, and the interpretation basis is (positive). It was not until June 5438+February 2006 that the high-precision DEM and interpretation foundation of key areas were re-established.

2)1:50,000 remote sensing survey. The1:50,000 DEM used in this project consists of the above three parts. The domestic part meets the national surveying and mapping standards, and the accuracy of the overseas part is difficult to count.

1:50,000 disaster and geological environment interpretation is based on 5m resolution SPOT-5 multispectral orthophoto image, with reference to ASTER, ETM and ALOS images. As far as the ground resolution is concerned, it is enough to meet the requirements of1:50,000 survey.

In the process of image processing, the domestic DEM which meets the national surveying and mapping standards is mainly used to correct and register the geographical coordinates. SPOT images in the investigation area have different seasons, and PAN data also have different multi-spectral phases. In addition, in the alpine valley area, it is very difficult to correct and integrate. After comparison of various methods, the final correction error of the fused data is less than 10 pixel by finite element calculation. ASTER, ETM and ALOS use the SPOT image after fusion correction for inter-image correction, and the error is controlled within two pixels.