Hi,
That was very nice of Amanda to consolidate the things. Thanks for that. Been observing people not replying all. Suggest if every body can mark the idea/suggestion/answer to the list instead to a person.
Regards, Sreeram.
On 12/22/05, Hargis, Amanda <ahargis@co.boulder.co.us> wrote: > > Thank you, everyone who emailed and phoned with ideas and thoughts about > getting started in linear referencing. The GIS community is truly > filled with some very generous people. > > Here is a summary, with my original questions first, the answers to my > questions listed next, concluded by some additional suggestions. The > most frequent answer to my questions was, "It depends!" We'll > definitely be following up on your suggestions about project planning, > further research and people to contact. > > Original Questions: > 1. How do we choose route lengths? Should these be long or short? As > an example potential route, we have a highway that traverses the county > from north to south, passing through two incorporated cities as well as > unincorporated county. Should each stretch be a route, or should the > entire length be one big long route? > > 2. How do we choose a route ID? Should this be a randomly generated > number, or a value that matches a key field in a transportation > maintenance table we have? > > Answers about Length of Route: > 1. It depends on the applications you will use it for. > 2. It depends on how the existing roads are numbered - by mile > marker? If so, where do the miles start? > 3. Shorter is better, so that interpolation of distance will be > more accurate. > 4. Longer is better, with the data determining the length - e.g., > perhaps split the segment when the speed limit changes > 5. Segments need to be broken at junctions with other segments for > routing (snow plow routes, etc.) > 6. Set up length based on "character" or "utility" rather than > "long" or "short" - it should be an identifiable thing > 7. Break the segments at intersections > 8. It depends on who will use the data, and how > 9. Longer is generally better > 10. It depends on how the routes will be used - find a model that > allows you to use multiple road names (e.g., Hwy. 287 is also Main > Street) > 11. Base the length on the usual method of identification for that > road and break the segments where the road name / address scheme > changes. > > Answers about Choice of RouteID: > 1. It depends on the correspondence of the tables to the features. > What if your database ID changes? Then the correspondence to the route > will be lost. > 2. RouteID has to match a key field in an event table > 3. RouteID should be automatically generated by the GIS > 4. A DBMS guideline is never to associate a record ID with any > field in the database > 5. It depends on what defines the routes as unique things, and how > you use them > 6. A randomly generated number, so performance is fastest through > indexing > 7. Random to enforce uniqueness > 8. Perhaps match it to County Road ID > 9. It depends on how the system is used. Design them to > accommodate the road data you will encounter. > 10. User-ids link the routes to external tables, and internal ids > link the routes to sections, etc. > 11. Make it a long enough number that there won't be conflicts (10 - > 12 digits) and base it on the addressing grid / milepost system > > Suggestions on who to chat with: > 1. City of Boulder > 2. Adams County > 3. Various consultants > 4. Neighboring counties - coordinate! > 5. Colorado DOT - coordinate! Use their routes / IDs? > > Suggestions on reading material: > > http://support.esri.com/index.cfm?fa=downloads.dataModels.filteredGatewa > y&dmid=14 > http://www.dot.state.ia.us/gis/lrs_project.htm > http://www.esri.com/news/arcuser/index.html : July-Sept. 2002 > and Oct. - Dec. 2002 issues > > Suggestions on Event Tables: > 1. Make one big event table, with each record holding an Event Type > and an Event Value (e.g., event type = Name: event value = Arapahoe) > 2. Make lots of event tables, each with a single focus (e.g., Road > Name events, Pavement Type events, etc.) > 3. Make two event tables: one for static data and one for dynamic > data, since in databases, you model each type of data differently > > Thanks, > -Amanda > ......................................... > Amanda Hargis > GIS Coordinator, Boulder County, Colorado > 303-441-3958 (voice) 303-441-3983 (fax) >
|