A small analysis of ArcGIS earth data

ArcGIS Earth, a lightweight, three-dimensional terrestrial application. Because the working relationship download trial for half a day, just take this opportunity to briefly study the general idea of ArcGIS Earth, especially the composition of terrain data and the loading of image data, in this summary. The following ArcGIS Earth is referred to as AE.

OK, this article first talk about words, then on the hard vegetables, and finally on some seasoning side dishes. No more nonsense, just walk away.

I use the latest version of the official website, just released a week, the interface style and the spirit of VS, I believe that developers will have a sense of déjà vu. However, the installation process is not smooth, prompted me to install the. Net Framework4.5.2 version and above, small e-June, I installed a VS2012 operating system, although not C #, but you this version of the requirements is a bit too ah, and you need to manually install the framework. The most sad thing is to upgrade the framework after the discovery of the VS can not be used, the outweigh, had to uninstall vs after reinstall, some twists and turns, can not endure ah.

After the installation is complete, the first simple to browse all the functions, the overall operation is relatively smooth, but the mouse operation has a larger difference, even with the right mouse button to control the elevation of the earth, the individual feel that this is somewhat inappropriate. The common application of the Earth class is to use the mouse wheel to control the operation, so that the right button can be used to expand some menu choices.

Menu bar design is also very simple and beautiful, the product designer really very attentively, non-(because) often (for) capacity (work) easy (CAN) on (not) hand (more). For example, adding a layer, combined with ArcGIS's powerful online service, can only make the grief home a sigh of empty mournful. Spit Groove, the release of a client product, how to allow users to experience the first time, rather than to find the URL of the data service, and even need users to configure these data services, seemingly simple thing, in fact, need multiple parts and coordination between the team, but become unable to solve, the time cost is too big, After the coordination has to do, the effect is also greatly discounted. Personal self-sweeping in front of the snow, which is also helpless. Look at the small e home, try to search the keyword population, below is a bunch of population map service let you choose.

Open the first online map service, a thematic map of the United States population density, click Load to add to the three-dimensional ball.

So the most basic map data loading can be very simple implementation, looked at the map for the Mercator projection, I found a latitude and longitude map service USA MAPSERVER,EPSG for 4326, found can also be superimposed on the earth. Oh, yes, well, both coordinate systems can be supported, which requires some special handling. However, a look at the URL of the request map slice found that the original client request included dynamic projection information, the end of the server will convert the 4326 map service to 3857 of the Mercator Slice returned to AE. This shows that AE will first request the JSON information of the map service, so that it can know the projection information of the map service, and then unify the request and load the Mercator projection corresponding map slices. So, it is very likely that the earth grid under AE is in accordance with Mercator rather than latitude and longitude, and AE itself does not necessarily support the conversion of map slices under latitude and longitude, so all slices are directly using the service side of the tile processing good. As for this kind of service to do dynamic projection of the way, which is better than the inferior is not explored.

Since the earth grid, then look at the effect of the north and South Pole, found that the effect of AE is good, smooth effect, and North and south because it will be reduced to a point, texture effect is usually not ideal, is AE to the north and South Pole did a special treatment, another to fill a? Finally found quite interesting, small e is also scheming for a prostitute, in order to avoid this problem, the Antarctic image is all white, so just can't see it. After overlaying a non-solid Antarctic map Service, the problem becomes leaky.

Although the map overlay function is a bit of a drawback, through product design to avoid technical problems, but in general, but also embodies the rich map of ArcGIS services, but also overlay kml and other data, has a good extensibility, but the KML is also flawed, does not support Chinese, the effect is as follows:

OK, Get started AE, the overall feeling is good, the basic functions can be very good use, even if there is no sun and light, but still have the atmosphere and stars. Provides this more practical function, also provides the printing of this boring chicken.

I think the most heart is basemap this control, at a glance, and the following terrain box is very humane to open the terrain function, in addition, if you proxy into the United States IP, it will be based on your current network environment to provide the corresponding U.S. map services, or very thoughtful design.

Well, since I have already highlighted the terrain in red, you should know the following is the core of this article, AE terrain data in the end what is the structure? Well, first of all, the terrain is very general, but the two techniques of MRF and LERC have been learned.

When you choose to load the terrain, AE will first request to obtain the terrain of the JSON information, the same terrain data is also the tile under the Mercator projection, generally the topographic data are the latitude and longitude of the slice, if cut into the Mercator grid, it is also very hard, after all, is the global data. This way we also know the range of xyz corresponding to the URL of the terrain data.

Download a topographic data to see what's inside, and who's going to call us. It is stated here that the slice should be in accordance with the MRF specification. MRF is a cloud-oriented data format provided by NASA, which preserves the compactness of traditional image data and the flexibility to absorb slicing data in a networked environment, and I am not familiar with this data format. For the data of a MRF specification, there will be a header that records the attribute information of the data, such as data size, length and width, etc. In this way, even without data, the slice is the header information of 1KB, but there is no data content, or the data content is considered to be zero. Download a larger topographic data to study and find that all terrain data has a Cntzimage logo on the head, which is also the only clue. Here naturally a little relieved, since small e such large companies do not have encryption and key design, it seems that the product manager to treat the program ape can not be too harsh, otherwise people will change the diamond pattern.

Cntzimage What does this logo mean? Check a bit around, but the principle is relatively simple, considering this is the terrain data, personal guess should be contrast Z image meaning, provide an absolute value m,z is the meaning of elevation, and the image of each point is a difference from the absolute value of M. This idea is similar to Google's high water mark, which can improve compression efficiency. But conjecture is a guess, and you have to prove it yourself.

The final discovery is that the terrain slices are 257*257, and should be designed for the boundary gap, using ESRI's own patented but open lossy compression algorithm Lerc (Limited Error Raster Compression). The advantage of the algorithm is not in this said, we can understand. And LERC has been supported in NASA's MRF specification and Gdal, the algorithm in the compression and decompression of the CPU consumption is low, although it may be lossy, but the compression efficiency and JPEG and so on, the personal opinion is more suitable for the network environment, especially to the JS end of the performance consumption of decompression

Suspense, ESRI provides LERC C + + source on GitHub, and after compiling the terrain data, you get the tile's property information, such as the range, followed by the image information, which can be rgba, or float and other types, The elevation values are naturally float types, and the corresponding values are as follows:

Well, the terrain data has been parsed, essentially a DEM slice data, and there is no triangle network information, so the terrain of AE is probably the way of using height map, and is not based on feature point structure network, real-time construction of a simple grid, so the terrain effect is general.

I think small easy may also never thought, programmer a provides LERC source (but must be provided, because it is open source), programmer B uses LERC to generate terrain data, programmer C accidentally know the secret between the two, so AE terrain data There is the possibility of leakage and resolution.

The introduction of MRF and LERC is not detailed, which mainly describes the position and realization of the two in topographic data. Interested in this can look at this paper "MRF as a Cloud Optimized Raster format and LERC Compression", written very simple, can have a simple understanding of the MRF format, In fact, my understanding is a 3DTiles of ideas, overnight this idea of the format on the rotten Street. If you debug your own LERC program to parse the terrain file, you should have a good understanding of this paper.

So the terrain data of AE cracked, and the image data is a common map service, here is an interesting place, the different image data may be offset and non-offset, so there will be some position on the earth is not allowed, can only say Ma De in China.

OK, here is a brief introduction of the other aspects of AE, poi retrieval function general, a lot of Chinese features support is not sufficient, contrary to English words should also be good, find a few can.

Acreage is also good, will consider the terrain situation of the height factor, and not just a simple two-point distance. But the best is the vector paste effect, but also no sawtooth, the effect is good.

Finally, the layer manager of AE is also more mature. Last saved in Arcgisearth.workspace, the JSON format facilitates the next open content and status reading.

This trip to AE to the end of this, simply said, AE ball in technology and not too many advanced and distinctive places, or I have limited ability to see. But on the product, with online strong support, in the product system to improve and positioning, Lazi is very in place, providing such a lightweight online 3D client. The functionality provided by itself is not much, but the consideration is more thoughtful, despite the shortcomings, are also technical components, and in the product design I am very envious, a kind of can not speak and admire.

A small analysis of ArcGIS earth data