8 replies to this topic

[ELeFevre]

I'm working on a general reference thematic map showing various utilities (gas, electric, etc) across four regions of the US. I'm trying to figure out which rivers to show on the map. The river layer I'm using is much too detailed... so I need to pull out the semi-majors and major rivers. What's the best way to do this? By length? By flow? The rivers layer I'm using has a classification field which includes "stream, intermitent stream, et cetera" but even still, there are too many features to use this field. I can also query the feature by length in miles and kilometers.

I was thinking that a rivers length might be a good indication of its flow (stream power?). Maybe it's time to dig out the old Fluvial Geomorphology textbook. Thanks in advance. Erin.

[CHART]

Erin,

Maybe you want to look for major washershed polygons of your areas. I think, by overlaying these polygons over your rivers it might be a starting point for generalization. Then you can remove the level of tributaries you want depending on scale of representation. Removing by lenght might not always be a good idea.
I would think there should be software out there to help you remove the level of hydrography you want. e.g. selecting the major river in the watershed and having it select upwards to the required level...
Or one could write a script to do it via geographic-intersections

[Martin Gamache]

You can also try to grab Tom Patterson's river dataset which he used on his recent US map. I think the river layer is available seperately as a vector file. If not you can always ask him.

[Charlie Frye]

Erin,

Most (actually I don't know of any) river data (I'm presuming you're working with lines) does not have the attribution to support what you need, which I think is to remove any rivers that aren't absolutely necessary and abstract/simplify the remaining rivers until they become comfortably part of the background. After all, how could the people who made that data know you were going to try to make your map today with their data?

My suggestion, with my best practices in data modeling hat on, is to start adding attributes to your river data so that the next time you find yourself needing to reduce the effect of rivers on your map, it will be a bit easier to deal with. Basically, evaluate the decisions you make about whether a river is kept in your map or not, and determine why or how you made that decision, then encode those decisions into your data.

Personally, I don't find that stream order is useful for this case. I liked the idea about using watersheds, and identifying the primary-, secondary-, and tertiary-level rivers within each watershed. But, I think that would only works for small scale maps and for some, but not all locales at mid-scales (as soon as some rivers become polygons or coastlines of major water bodies get involved the problem becomes more complicated).

For large scale maps, the problem is further exacerbated by local density, which means that when there?s lots of water in an area, more detail is expected,and as one moves away from that hydro-intensive area, the expectation for details and richness of information diminishes. An approach I?ve experimented with in this arena with some success, is to use an attribute that contains the largest map scale I?ve used that feature on (that way a simple greater/less than query will get you all the features you?ve used at a given scale of map, which would be the starting point for making a map of a different scale). I like this approach too, because it cuts across representations (lines vs. polygons), and across different feature types; for instance there are different rules for when you might keep perennial streams (minimum width) versus intermittent streams (minimum length). Basically it?s easier to model the use of your rules for mapmaking than it is to model your rules when the rules are too many or too variable, or the software lacks the tools to automate what you need.

Charlie

[merft]

I'm working on a general reference thematic map showing various utilities (gas, electric, etc) across four regions of the US. I'm trying to figure out which rivers to show on the map. The river layer I'm using is much too detailed... so I need to pull out the semi-majors and major rivers. What's the best way to do this? By length? By flow? The rivers layer I'm using has a classification field which includes "stream, intermitent stream, et cetera" but even still, there are too many features to use this field. I can also query the feature by length in miles and kilometers.

I was thinking that a rivers length might be a good indication of its flow (stream power?). Maybe it's time to dig out the old Fluvial Geomorphology textbook. Thanks in advance. Erin.

If you hydrology source uses CFCC codes (from the Census), you can use the attached Access database I created to classify the waterways a little easier. Load your hydrology dataset, and join to the appropriate table. Just be sure that the fields are identical. I usually classify be perennial waterways.

-Tom

Attached Files

•  CFCC_Tables.zip   31.59K   76 downloads

[MapMedia]

Erin - I usually start with watersheds. But seeing how watersheds are nested (River Basins, etc...), it is important to decide which scale of watersheds best fits your map. For small scale maps, I might start at the River Basin scale (this would include the entire Sacramento River for instance) and plot only the main stem of the River Basin's name sake.

Sometimes it does help to see level 1 hydrology plotted, to start building your visual feel of the terrain. Flow is proportional to stream levels, so if you do not need flow (cfs) to set a threshold or something, then stream level or order classifications work fine.

Rarely do I see a small scale map of any area I know, that, for me, nails all of the pertinent hydrology (either over-reaches or misses a few key drainages) - While you can't please everybody, getting the key prominent drainages is very important, this is why I believe using river basins and their "watershed children" is a great method.


Chris

[ELeFevre]

Thanks for all of the great replies, everyone. I apologize for not responding, but it's been a busy week and going to remain so until next week. Anyways, thanks for your help...it's invaluable and greatly appreciated!

[CHART]

I think Charlie is correct, a fully automated solution would probably not work 100%.

To that effect I remembered the method used to generalize 1: 50 000 scale maps down to 1: 250 000 for the Canadian NTS system for hand drafted maps.

I will try to put it in a modern digital work flow.

1. Plot out your water network (with watershed polygon outlines if you want) at the required final scale. Make sure you have a reference system (grid ticks). Don't use black ink.
2. Using a black marker trace out the water network in a generalized way directly on your plot.
3. Scan your work.
4. Geo-reference your new raster map using the grid.
5. Remove the all colors (e.g. using blackart software)
6. Auto-Vectorize and build topology.

You now have a data set for the desired scale...

Sounds absurd?

Anyway just sharing...

[Dennis McClendon]

You could always stand on the shoulders of giants, and just borrow the rivers from Tom Patterson's work, either for the National Parks map http://www.nps.gov/c.../NPSmap2.ai.zip
or his new physical map of the US, for which he spent a lot of time thinning the hydrography.