How can I accurately split roads from the TIGER roads dataset, to create proper junctions in a network dataset?

Question

I am trying to create a network dataset in ArcMap using the TIGER roads dataset. A problem with this dataset is that some road segments are not split at intersections. Therefor, network analyst would treat it as though one could not turn onto that road from another (See picture below - the TIGER dataset is like the example of the "poor quality" in some cases)

I have used the "feature to line" tool to split roads where they intersect other roads for everything not classified as a primary or secondary road. However, I cannot do this for the primary or secondary roads, since it would likely be inaccurate to assume that you could come onto/off of the highway at any road that intersects. I have already tried using the Feature to Line tool for Primary and Secondary only where they intersect ramps, but it seems that in many cases one should be able to get onto the highway even where there is not segment classified as a ramp.

Has anyone use TIGER data to create a road network dataset before? Have you had this problem? Any suggestions as how to resolve it?

Answer 1

You can conflate two network datasets. It is time consuming (taking two weeks to get something decent for a large county) and you have to make decisions based on the data attributes to generate a heirarchy. And which attributes are worth keeping.

You can convert the ESRi .sdc to feature class from the tutorial streets network, that comes with arcmap from 2013, buffer out all the primary and secondary roads say maybe 100 feet. You also want to get all the oneway and f_elev and t_eleve attributes as well because you may not have them in your tiger/osm/county gov data. Your heirarchies are net2lcass, rd condition, and frc. You may not need all of them but they are good to keep or combine for reference

Make sure you are working in the same projection between both datasets.

Take newer data from TIGER,county government, OSM --Determine how you want to combine these attribute with your .sdc data (feild names may not be the same, but the data may be similar)--buffer (100 ft)/cut out the primary and secondary rds or rds that correspond to one way, f_elev,t_elev using select by location or clip your .sdc conversion buffer.

Use a combination of Dissolve, Merge, Extend Lines Tool, integrate, Feature to line if necessary (Keep backups in case you mess up your original data). There may be other tools you can use.

Recalculate fields--

Inspect that the 100 foot slivers have the same hierarchy and speed limit across the board between your newly merged dataset. (that the .sdc rd and the newer rds match) Hopefully road names as well. Recalulate heirarchy, feet, and speed limits, Recalulate your minutes attribute, shape_length/5280 *60/speed. No Null values. You have to make some educated decisions on how you want merge these values.

Use Topology Inspector editor if you have a license for it.

Build Your network using an .sdc turn feature class if you think you may need it (similar process as above) and set global turns.

Use the network, and inspect that you are getting the right behavior especially with the f_elev, t_elev and oneway fields.

To edit these is where the time comes in...

Based on the the overpasses that work correctly , you can reverse figure out the pattern on how to change the attributes for t_elev,f_elev and oneway fields in the editor toolbar based on the direction of travel -- from red to green endpoints vs green to red endpoints when editing the vertices for each segment. Approaching an over pass, going over the over pass, and descending the overpass ( and there are four combinations as it relates to oneway rds and overpasses combined).