For the historical study I am conducting, I have three layers with roads in 2000, 2008 and 2012 respectively. Only the data for 2000 is complete- i.e. there are additional segments of roads in 2000 that are not present in data for 2008 and 2012. I'm positive that roads did not vanish between 2000 and 2008 😛 I would like to add the length of the additional roads from 2000 to that of 2008 and 2012, so that my data is correct and complete. How do I go about it?
Time is not in my favor and I would really appreciate some help. A BIG thanks to all those who would like to help me out 🙂
For the historical study I am conducting, I have three layers with roads in 2000, 2008 and 2012 respectively. Only the data for 2000 is complete- i.e. there are additional segments of roads in 2000 that are not present in data for 2008 and 2012. I'm positive that roads did not vanish between 2000 and 2008 😛 I would like to add the length of the additional roads from 2000 to that of 2008 and 2012, so that my data is correct and complete. How do I go about it?
Time is not in my favor and I would really appreciate some help. A BIG thanks to all those who would like to help me out 🙂
While Nidhin's suggestion is one approach you can take after you have selected all of the roads that were in 2000 that you want to add to 2008 and 2012 networks, I assume this post is also concerned with how to go about actually making that selection for road networks that have significant overlap with each other which makes the selection difficult to perform manually.
In addition, I want to suggest that all of the sources have to be treated as complete unless you have documentation from the source stating otherwise. The roads from the oldest network may not have disappeared, but they may have conflicted with the purpose of the later networks. If these networks came from different sources or a source that over time decided to change the purpose of the network to better fit its mandate and authoritative documentation, the networks may still all be complete for the original purposes they served. For example, if one of the network represents those roads with a clear legal history dedicating them to the public that Surveyors consider real roads, and another represented the network of roads that were actually maintained by a public jurisdiction, and the oldest represented the network of roads that emergency vehicles will use, they could all be complete for the purpose they were designed to represent yet be very different from each other.
You should make no changes to any of your authoritative source data until you have documented and understood each source network's metadata and purpose. Historical research often has to combine separately developed sources that served different purposes over time and should always preserve the unaltered authoritative version. Changes should only be made to a copy and all alterations made by the historian should be flagged in such a way that the "gaps" filled in by the historian that used data from other sources can be easily filtered out so that the purpose of each source can be taken into consideration when the historian makes interpretations of the data.
To deal with the selection process, if you have ArcGIS 10.2 and an Advanced license I would suggest trying out the Detect Feature Changes tool to evaluate the network differences. The level of comparisons made by this tool appear to cover the bulk of issues that come up when dealing with line network comparisons.
In addition to this tool, Spatial Join can be useful for detecting differences in networks if the common roads in the networks are reasonably close to each other in alignment. I typically extract the centroids of the lines that actually fall on the line of one of the network versions and make them the Join features in a Spatial Join to one of the other line network versions. Detecting where no centroid is present often allows me to select the features that are unmatched, if the segmentation and alignment of the two networks is reasonably similar.
Another technique involves calculating a field to hold the ObjectID of the original lines and then creating a very small buffer around all of the roads in the 2000 network and performing a union of those buffers against itself. Then select only the portion of buffers that had no overlap with any other buffers to get the portion of the roads that clearly exclude all road junctions and use that as the Target of a Spatial Join to one of the other networks with the One to Many option. All buffers that have no Join feature (JoinID = -1) represent segments in the 2000 network untouched by the other network. Select that set of buffers and relate them back to the 2000 network lines on the preserved line ObjectID field to get the set of lines that need to be added to the other network.
After selecting the unmatched lines you can use the method suggested by Nidhin or use the Append tool to append them to the copy of the original source network you should have created.
If these techniques do not work for your particular set of networks please provide some sample screen shots of problem areas.
Add all the three Road Layers in ArcMap. In the editing mode, copy the missing roads from 2000 and paste it on 2008/2012. While pasting, you will be asked to mention the Target. You can select either 2008 or 2012 and click OK.
For the historical study I am conducting, I have three layers with roads in 2000, 2008 and 2012 respectively. Only the data for 2000 is complete- i.e. there are additional segments of roads in 2000 that are not present in data for 2008 and 2012. I'm positive that roads did not vanish between 2000 and 2008 😛 I would like to add the length of the additional roads from 2000 to that of 2008 and 2012, so that my data is correct and complete. How do I go about it?
Time is not in my favor and I would really appreciate some help. A BIG thanks to all those who would like to help me out 🙂
While Nidhin's suggestion is one approach you can take after you have selected all of the roads that were in 2000 that you want to add to 2008 and 2012 networks, I assume this post is also concerned with how to go about actually making that selection for road networks that have significant overlap with each other which makes the selection difficult to perform manually.
In addition, I want to suggest that all of the sources have to be treated as complete unless you have documentation from the source stating otherwise. The roads from the oldest network may not have disappeared, but they may have conflicted with the purpose of the later networks. If these networks came from different sources or a source that over time decided to change the purpose of the network to better fit its mandate and authoritative documentation, the networks may still all be complete for the original purposes they served. For example, if one of the network represents those roads with a clear legal history dedicating them to the public that Surveyors consider real roads, and another represented the network of roads that were actually maintained by a public jurisdiction, and the oldest represented the network of roads that emergency vehicles will use, they could all be complete for the purpose they were designed to represent yet be very different from each other.
You should make no changes to any of your authoritative source data until you have documented and understood each source network's metadata and purpose. Historical research often has to combine separately developed sources that served different purposes over time and should always preserve the unaltered authoritative version. Changes should only be made to a copy and all alterations made by the historian should be flagged in such a way that the "gaps" filled in by the historian that used data from other sources can be easily filtered out so that the purpose of each source can be taken into consideration when the historian makes interpretations of the data.
To deal with the selection process, if you have ArcGIS 10.2 and an Advanced license I would suggest trying out the Detect Feature Changes tool to evaluate the network differences. The level of comparisons made by this tool appear to cover the bulk of issues that come up when dealing with line network comparisons.
In addition to this tool, Spatial Join can be useful for detecting differences in networks if the common roads in the networks are reasonably close to each other in alignment. I typically extract the centroids of the lines that actually fall on the line of one of the network versions and make them the Join features in a Spatial Join to one of the other line network versions. Detecting where no centroid is present often allows me to select the features that are unmatched, if the segmentation and alignment of the two networks is reasonably similar.
Another technique involves calculating a field to hold the ObjectID of the original lines and then creating a very small buffer around all of the roads in the 2000 network and performing a union of those buffers against itself. Then select only the portion of buffers that had no overlap with any other buffers to get the portion of the roads that clearly exclude all road junctions and use that as the Target of a Spatial Join to one of the other networks with the One to Many option. All buffers that have no Join feature (JoinID = -1) represent segments in the 2000 network untouched by the other network. Select that set of buffers and relate them back to the 2000 network lines on the preserved line ObjectID field to get the set of lines that need to be added to the other network.
After selecting the unmatched lines you can use the method suggested by Nidhin or use the Append tool to append them to the copy of the original source network you should have created.
If these techniques do not work for your particular set of networks please provide some sample screen shots of problem areas.
