I believe you may be concerned about this issue, an industry insider engaged in the development of thematic information systems such as surveying and mapping, geographic information system development (GIS), or more likely a non-professional reader involved in geographic information and website map services in work or life. First of all, why am I concerned about this issue? This may be related to your interests, your worries or your doubts, if this is better. Compared with other map websites, GoogleEarth has an advantage: it can read the three-dimensional coordinates of any point on the ground-longitude, latitude and altitude, without the need for users to register and log in and authorize the use of data. But what is the mathematical accuracy of Google Earth? There are few reports about precision testing and authoritative appraisal in China. Understanding the accuracy of Google Earth is meaningful in at least the following two aspects: (1) It is convenient for users to understand the accuracy and reliability of using Google Earth. (2) It is beneficial to adjust and improve the mathematical accuracy restriction policy of website maps (Baidu map, Gaode map, sky map, etc.). ) In China. Some existing public map websites in China do not provide geographical coordinates (such as Baidu map), and some only provide partial geographical coordinates (such as sky map only provides plane coordinates and does not provide elevation). It is no longer a technical problem for domestic map websites to provide three-dimensional coordinate services. Perhaps the reasons for providing or not providing coordinate information are mostly based on the consideration of evading the restrictions of China's geographic information security policy (for example, China stipulates that the positioning accuracy of public maps should not be higher than 50 meters, and the positioning accuracy of vertical intervals should not be lower than 50 meters). I have been engaged in surveying and mapping for a long time and have worked in map management positions. I made some discussions on this in combination with the problems in my work. Through the analysis and comparison of map websites at home and abroad, as well as the investigation and analysis of the needs and opinions of map users, I think it is unnecessary if the map websites created in our country adopt too high mathematical accuracy restrictions, which may lose many potential users and limit the role of domestic map websites. Combined with the needs of work, the author designed and organized a small-scale but relatively formal field measurement of Google Earth's mathematical accuracy. At that time, the purpose was to provide an objective sub-sample data to infer the accuracy of Google Earth, for the purpose of grasping the relevant affairs at work and for the reference of higher authorities. The field test is completed by Grade A surveying and mapping units. In the aspects of scheme design, determination of test instruments, use of satellite positioning station network, data processing and statistical analysis. , completely follow strict procedures, operating methods and data processing rules to ensure the authenticity and reliability of test data and the objectivity of result analysis. 2 test results first of all, the test results are mentioned above. Most readers are also most concerned about the test results, which can give readers a general understanding. Conclusion According to the statistical analysis of the test data, the longitude, latitude and altitude of the points read from Google Earth are relatively high, and the median errors of plane and elevation of most points are 5m, and the median errors of points and elevations in all survey areas are within 10m. (Note: After Google Earth is turned on, when the mouse moves on the map, the values of longitude, latitude and altitude will dynamically appear below the map. Longitude and latitude are displayed to two decimal places, that is, 0.0 1 sec, and altitude is displayed to the whole meter. For the sake of intuition, the latitude and longitude units in the accuracy calculation results have been converted from degrees, minutes and seconds to meters. (3) Design and implementation of the test The design ideas, field test methods, data analysis and related suggestions of the test scheme listed at the back of this section are mainly provided for relevant professionals to read, and relevant readers can further analyze the scientificity and reliability of this test according to this, or as a reference for in-depth research. 3. 1 Detection area map 1 Select and detect a survey area on Google Earth. It is divided into five measuring areas, and each measuring area is independently observed and counted. Among them, survey area 1 and survey area 2 are located in the east and west of the same medium-sized city, belonging to flat areas, with an average elevation within 10m and an image resolution of 1m on Google Earth. No.3 survey area is located in a small city far away from 1 survey area and No.2 survey area, with flat terrain, average elevation within 10m, and image resolution on Google Earth is1m.. 4 and 5 survey areas are located in mountainous areas, with an altitude of 100m~400m ~ 400 m, and the resolution of Google Earth image is 2.5mm. Among the five survey areas, the distance between the farthest two survey areas is over 300km. The design and selection of survey area mainly consider the following factors and purposes: (1) Including different terrain categories-analyzing the influence of terrain differences in plain, mountainous and alpine areas on accuracy. (2) Including different ground image resolutions —— Analyze the influence of the resolution difference of satellite images on the accuracy of point selection. (3) Including different regions-analyze whether there is regional deviation. 3.2 Extraction of sample data The sample data is directly obtained from Google Earth, and the methods are as follows: (1) The requirements for selecting points on Google Earth are: ① the image is clear; ② accurate pointing; ③ The terrain around this point is flat and there are no buildings protruding from the ground; ④ It is convenient for field observation (easy to reach, easy to set up stations and good GPS signal); ⑤ The facula is consistent with Google Earth image. (2) Make a checkpoint sample file, take a screenshot from Google Earth, and number the views; Enlarge the point view image and read the point coordinates (latitude and longitude) and elevation of the target. In order to facilitate on-site inspection, two view files are established for each point, one of which is a small scale selected on site; The other is an enlarged view of the selected points in the field, which can display the latitude, longitude and elevation of the image points on Google Earth. Figure 2 Selection of Flat Points Figure 3 Selection of Mountain Points 3.3 Instruments used for field data acquisition (1): dual-frequency GPS receiver, model "Leica 1230". (2) Observation method: Through the static observation mode of network RTK service mode provided by Zhejiang continuous operation satellite positioning integrated service system, the latitude, longitude and geodetic height (both in WGS-84 coordinate system) are collected on the spot, and each point is observed twice, and the average value is taken if it meets the tolerance. According to the preliminary test of the continuous operation satellite positioning integrated service system in Zhejiang Province, the observation accuracy is equivalent to: plane 5cm(X/Y component) and elevation 10cm. Therefore, compared with Google Earth data, the field measurement results are regarded as "true values" when calculating errors. In order to save the workload of field inspection, the results of 1: 10000 aerial survey field image control points measured in June-May, 2008 were used in 39 ground points located in mountainous areas. The specific method is: call the aerial photogrammetry control point image and the latitude, longitude and elevation measured on the spot, and then read the latitude, longitude and geodetic height (WGS-84) of the same point on Google Earth website, and intercept the view. 4 Data processing and precision statistics Data processing and precision statistics take the survey area as the unit. (1) Convert two groups of latitude and longitude coordinates of the detection point (obtained from Google Earth and field measurement respectively) into Gauschluger plane coordinates, calculate the horizontal and vertical coordinate errors, and count the point median error. (2) Compare the plane coordinates and elevation according to the survey area (WGS-84). (3) Take the field test value as "true value", and calculate the plane coordinates and elevation error of Google Earth. (Note: By comparison, it is found that the median plane error of one survey area located in mountainous area is 10. 13m, and the median elevation error is 22.29m, which is obviously higher than other survey areas. After readjusting the screen shot view of Google Earth and the point photos taken during field survey (actually taken during aerial survey), it is found that some points fall at the elevation conversion or the corner of the room, and there is a great interaction between plane error and elevation error. After analysis, the points at the elevation conversion or the corner of the room were eliminated, some points were added for field investigation, and a few "abnormal points" with unknown reasons were eliminated after analysis. Fig. 4 The accuracy of 99 points in 5 survey areas is easy to be unstable when the elevation changes sharply. The accuracy statistics are shown in the following table: 5 Analysis of test results (1) The accuracy of Google Earth mathematics in various survey areas is not consistent. The plane accuracy and elevation accuracy of survey area 3 are the highest, and the median error is less than 2 m; The plane accuracy of No.4 survey area is the lowest, which is 9.59 m; The median error of plane and elevation of most points is 5m. The point mean error and elevation mean error of all survey areas are within 10m. (2) The difference of terrain types has no obvious influence on the mathematical accuracy of Google Earth, which does not mean that the accuracy of flat land is higher than that of mountain land in general aerial survey. However, the accuracy of measuring points in mountainous areas located on slopes or where the altitude changes sharply is unstable. The reason is that the global DEM made in the United States is used to correct the projection difference in the image processing of Google Earth, and the plane error and elevation error are intertwined at the "foot of the mountain" or "above the ridge", which will lead to unstable accuracy. (3) When the image resolutions of the selected points are 1m and 2.5m respectively, the influence of the image resolution on the mathematical accuracy of Google Earth is not obvious. 6 Conclusions and Suggestions (1) After excluding individual outliers, there are 99 points involved in the calculation, with more effective detection points, standardized operation and even distribution of points, which can be used as samples for Google Earth precision detection in this area. (2) Limited by the number of samples and geographical scope, this test is still not enough to represent the accuracy of Google Earth in China. It is necessary to comprehensively test the accuracy of Google Earth in different terrain categories (plains, hills, mountains and plateaus) in China (east, west, north and south) and make a more comprehensive and objective evaluation. (3) If, after further rigorous argumentation, the maps of Google Earth or other international websites really reach the accuracy of about 10 meter in China, then the existing regulations that the location accuracy of public maps in China should not be higher than 50 meters and the vertical interval should not be less than 50 meters should be adjusted in time to improve the accuracy to the level corresponding to that of Google Earth. (4) This paper has certain reference significance for studying the improvement of China's geographic information security policy. Thanks to Hu Zhengwei, Su Liqian, Feng Litao, Wan Fei and other comrades for taking part in this test. Thanks to Comrade Hu Zhengwei for helping the author complete the data statistics!