From blazek@itc.it Mon May 3 12:20:21 2004 Content-Type: text/plain; charset="iso-8859-1" From: Radim Blazek To: Christoph Spoerri Subject: Re: [Qgis-developer] Plugin help Date: Mon, 3 May 2004 12:20:21 +0200 X-Mailer: KMail [version 1.3.2] References: <200404220749.i3M7nav12790@janacek.itc.it.> <200404270830.i3R8UPR20836@janacek.itc.it.> <200404282242.33983.cspoerri@cavtel.net> In-Reply-To: <200404282242.33983.cspoerri@cavtel.net> MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Status: RO X-Status: S On Thursday 29 April 2004 04:42, you wrote: > Hi, > > I followed the ESRI way (the shapefile way and not the coverage way > though). So, each vertex has a measure value. How do you store the 'measure'? PostGIS does not have measure AFAIK. > I searched the web for other models, but wasn't able to find any. Neither did I. > Regarding your two draw-backs of the "ESRI" system: > 1. what do you mean with Mile post not being precise? The system does not > have MP, but you can rather use it to create a layer with MPs. The measures > with the vertices are used to tell you how far along the line your are. Or vice versa, MPs can (must!?) be used to create LRS. In fact I cannot find how ESRI creates LRS. What data are you working with? We have for example the layer of roads (lines) and the layer of mileposts (points). I take these two layer and make LRS from them. My objection is, that if the MP (input point) is not near a vertex on the line, and previous and next segment have different scale, you get an error, which is not necessary. But well, probably not so important. Example: --- line + vertex x1 milepost with km +--------------+------+ x0 x1 x2 The distance x0-x1 in GIS = 0.8 km (real world 1km) The distance x1-x2 in GIS = 1.2 km (real world 1km) Note that of course the distances measured in GIS differ from those in real world and may be both longer and shorter Now we have 3 vertices, the first will have measure 0, the third measure 2. And the second? Which measure? Optimal would be to add a new vertex where the MP is: +--------+-----+------+ x0 x1 x2 > 2. With smooth changes you mean a gradual change in an attribute? > Like the decline of O2 concentration in a polluted stream? I believe this type of > data would be hard to map with any system, since attributes are discrete > entities. No, sorry, I mean a physical changes of mileposts in the real world. If you cannot maintain partial changes in LRS, the only option is to change always all mileposts. > Yes, it would be nice to have a compatible system. Do you have more > documentation on the LRS in GRASS? At the first glance, it seems to have a > bit more overhead than the ESRI (or Oracle spatial) model. I don't have any documentation, as we use it only here. I have to write something, but I don't have much time, let me know if you are interested. For now briefly: All the LRS is stored in one table, each record stores informations about one segment between two mileposts: rsid | integer | id of LRS segment (not so important) lcat | integer | line id, unique id of the line lid | integer | id of 'route' (for example 456 for 'Highway 12') start_map | float8 | the start of segment measured from the beginning of the line end_map | float8 | the start of segment measured from the beginning of the line start_mp | float8 | milepost on the beginning of the segment start_off | float8 | offset on the beginning of the segment end_mp | float8 | milepost on the end of the segment end_off | float8 | offset on the end of the segment Say the you want to georeference a point specified by route id = 456 and km+offset = 15+300. You find a record which has lid = 456 and start_mp < 15 and start_off < 300 and end_mp > 15 and end_off > 300 now it is possible to calculate the distance from the beginning of the line in GIS. In fact, I have here also simple web application which works with this table in Postgres and line layer in PostGIS using a function in Postgres. > Also, what do you do, when you add additional segments between the 3rd and 4th MP (in > addition to the two points)? Whole part between 3 and 4 must be referenced from 3, but such changes are not frequent in real world and to change only the part between 3 and 4 is not so difficult (still talking about changes in real world, traffic signs for example, signs along the river etc.) > Are there any other systems/data formats that > implement this LRS model? I dont know any published, there is certainly one in MGE and I am almost sure that it is different from ESRI. > Christoph > > P.S. it may be worth while to have a chat meeting on this. Let me know if > you are interested to do so. I would prefere mails. Radim