Help with ArcGis/Spatial Analyst. Anyone?

[hop]

Klar, you sound like me 7-8 months ago.

[Telenater]

I always like trying to help people out with GIS, sometimes I'm even successful.

This is more difficult than it should be, and if anyone has a better solution, I'd love to hear it.

Short of setting up a classification of the raster and then saving the symbology to a layer file (or imporing it to another layer directly from the sybmology), the best way I've found is to use a stretched renderer on the raster, and then in the statistics portion of the Stretched renderer specify my own custom minimum, maximum, mean and standard deviation.

Set your minimum and maximum to the min and max of the range of elevation you want to display, and then play with the mean and standard deviation to get the results that you want. You can end up with some strange results doing this, but'll be able to hone in on what you want. Then once you've found settings you like, you can use the same statistics on another raster and get the same colors for the same elevations.

[klar]

I guess I'm not exactly sure what you mean. If you're using classified data you should be able to manually select each class(and manually set where the breaks are) and choose the color. If you using stretched data then it can be hard to get different datasets with different mins/maxes to line up. But with it manually grouped into classes you should be able to choose what colors you want. But maybe I'm missing something?

well, ideally i would like a way to use one symbology for all my files without having to do it manually for each one, but even the doing it manually bit escapes me so far.

Short of setting up a classification of the raster and then saving the symbology to a layer file (or imporing it to another layer directly from the sybmology), the best way I've found is to use a stretched renderer on the raster, and then in the statistics portion of the Stretched renderer specify my own custom minimum, maximum, mean and standard deviation.

Set your minimum and maximum to the min and max of the range of elevation you want to display, and then play with the mean and standard deviation to get the results that you want. You can end up with some strange results doing this, but'll be able to hone in on what you want. Then once you've found settings you like, you can use the same statistics on another raster and get the same colors for the same elevations.

hm, that sounds interesting but sort of complicated. is there no simple way to do this with a layer file? saving the symbology as a .lyr has always given me not just the same symbology but the exact same data displayed in the end

i'll play around with your way and see what i get, thanks!

Klar, you sound like me 7-8 months ago.

do i get to go on a cruise when i'm done also?

[khakis]

This problem rang a bell for me (had the same issue once) when you posted it yesterday but I couldn't remember exactly how I solved it until last night. Unfortunately I don't have time to work it through today, but maybe I can help you figure it out.

IIRC you can import symbology definitions for rasters, yes? Does one of your rasters contain the entire range of values for all of the rasters (has the max of the group as well as the min value)? If so, set up your breaks and colors for this raster just right, then import the symbology def'n to the rest of the rasters. I don't think it will reclassify for each raster if you do this. If none of your rasters covers the entire range of values, make a copy of one of your rasters (call it dummy_for_symbo or something) and edit a couple of cells so that it does contain the entire range of values. Set up breaks and colors and import as before. Hope this helps. Maybe I'll get a chance to try it out and clarify later today.

[Alpenglo]

I think your problem is common ArcPainInTheAss, I've dealt with it before while trying to make maps using stretched elevation datesets in different regions with different elevation ranges. Unfortunately there's no one-button fix that I know of. As far as I know there's 3 ways to approach this. I can't possibly type enough to describe each process in detail, but here's an overview.


1. Convert the raster data to vector using ArcToolbox - similar elevation values to create discrete features. Then set up the symbology manually to use the same color scheme. I think this is preferred for graphical purposes as you will have nice looking vector shapes and can have lines showing the edge of zones and the boundaries between. The problem with vector in this instance is that the scalable nature of it will imply that you have a much more accurate dataset then you actually do when zoomed close in.

2. Show the raster data as stretched like Telenator describes -Properties>Symbology>Show:Stretched. Ditto what telenator says. Its takes some playing around but will work. However if you have differents mins and maximums that complicates things and will take more experimentation.

3. Show the raster data as classfied - Properties>Symbology>Show:Classified. Then choose either a predetermined scheme of classfications, or choose one manually. I'm a big fan of using the Natural Breaks scheme to figure out where those breaks are, then setting them manually to more commonly used break values. Then click on the little colored boxes in the window and choose the colors manually.


I think the method you choose will depend on how many classes you're dealing with. If your dealing with classes at all. Are you trying to show distinct zones of the same, or similar elavations, or continuous change over an area? If its less than a dozen or a couple dozen zones, then go the vector, or classified raster route. If its many more or continuous change, go the stretched route.

Hope thats even slighly helpful, or understandable. Feel free to post up some screenshots if you need help. I'll try to check back here later. Good luck.

[hop]

do i get to go on a cruise when i'm done also?

...the cruise comes later, after you go to South America, go on a surf trip from CA->BC, and have a few months of skiing.

[taz]

I heart this place. I bet someone here knows how to fix a Flux Capacitor or build a Light Saber as well....

at 34 seconds this guy can help you with the Light Saber thing

[ame="http://www.youtube.com/watch?v=94syGYcdGcU"]YouTube- Boag's Draught: From The Pure Waters Of Tasmania Full Length (HD!)[/ame]

[flyandski365]

Ok, I feel like a total GIS jong here, but I'm running out of patience. All I'm trying to do is add xy data from an excel spreadsheet. It gives me the standard "No ObjectID field" warning which is fine, I can fix that later. My problem is that my points come in properly relative to each other, but there is obviously an issue with the projection because they aren't fitting with the rest of my layers that are all using the same coordinate system. Any thoughts?

[geomorph]

Ok, I feel like a total GIS jong here, but I'm running out of patience. All I'm trying to do is add xy data from an excel spreadsheet. It gives me the standard "No ObjectID field" warning which is fine, I can fix that later. My problem is that my points come in properly relative to each other, but there is obviously an issue with the projection because they aren't fitting with the rest of my layers that are all using the same coordinate system. Any thoughts?

How far off are they? A few meters? Kilometers? Other side of the globe? Dunno for sure, but maybe it's a rounding problem in the original data. How many decimal places do the cells go?

[flyandski365]

They are lat/long to 4 decimal places, and when they are plotted should be a rough outline of FL. Problem is, they come in way out in the GOM and WAY too small. My FL shorelines shapefile zoom full extent is 1:5000000 (showing all of Florida), and for the points I'm bringing in it's something stupid small like 1:80, if that gives you any idea of the scale problem. Is it possible Arcmap is interpreting my DD as something else, like meters or feet? How would I change that?

EDIT: Ok, so yeah, it wants to import the coordinates as UTM (meters of course, doh!), not DD. I guess I have to convert all my lat/long to UTM for it to project properly? I suck at this.

[Art Shirk]

They are lat/long to 4 decimal places, and when they are plotted should be a rough outline of FL. Problem is, they come in way out in the GOM and WAY too small. My FL shorelines shapefile zoom full extent is 1:5000000 (showing all of Florida), and for the points I'm bringing in it's something stupid small like 1:80, if that gives you any idea of the scale problem. Is it possible Arcmap is interpreting my DD as something else, like meters or feet? How would I change that?

[F#*k you cat]

fly and ski

first open a new arcmap document and set the projection in arcmap to wgs 84 decimal degrees.

Create a DBF from the .xls. >>>IN excel - loose the extra excel sheets - Highlight every row and column and save as a DBF.

In Arcmap add the DBF to the table of contents, > right click on the table in the TOC, and calculate points. (I am running this by memory at home). ((((there may be another way to populate points from arctoolbox))))

either way you should have points. save - export to shape by using the data frame as the coordinate system> and you'll be good to go.

I hope I am right

Kev (effucat)

[flyandski365]

fly and ski

first open a new arcmap document and set the projection in arcmap to wgs 84 decimal degrees.

Create a DBF from the .xls. >>>IN excel - Highlight every row and column and save as a DBF.

In Arcmap add the DBF to the table of contents, > right click on the table in the TOC, and calculate points. (I am running this by memory at home). ((((there may be another way to populate points from arctoolbox))))

either way you should have points. save - export to shape by using the data frame as the coordinate system> and you'll be good to go.

I hope I am right

Kev (effucat)

Hey thanks man, I'll give that a shot in the morning. I'm done fucking with it for today, but I'll report back.

Oh and thanks commonlaw, that made me LOL.

[flyandski365]

You da man Kev, worked like a charm. Just one more reason this place is awesome.

[Telenater]

FWIW, I prefer importing excel spreadsheets to access tables rather than saving to .dbf because the import tools in access do a reasonable job of cleaning the data and letting you assign field types.

edit: and the save as, with data frame's projection works, but only if you've got all the projections right. It can sometimes be hard to tell if you've got a datum shift problem. I prefer assigning a projection during the "Display XY" process rather than later. It avoids getting tangled up in the morass of having coordinate systems that may or may not be responding correctly in the data frame.

[grubbers]

i literally just ran into this issue yesterday. if you're using decimal degrees, define the projection as wgs_1984, which should be under geographic coordinate systems. arcmap will automatically project the points to their correct locations even if the two layers have different coordinate systems. you can then convert the projection to that of the shapefile using arctoolbox (the tool is under data management i believe) or just leave it if all you have to do is display the points on the map.

[khakis]

probably worth noting here that you *should* be generating these points in the projection/coordinate system as the data were collected. If it was with a consumer GPS and the user didn't mess with anything, WGS84 is a safe bet.

I tend to use F#*k You Cat's method of generating points in a new ArcMap project...keeps things from getting jacked up, as they often do when working with .dbf files. Once I have my points on a blank map I drag over a roads or landgrid layer to make sure they're in the right place then export.

[Telenater]

If I get tabular data in decimal degrees I generally start with the guess that it's WGS84, but it isn't always. I've had several data sets handed to me that turned out to actually be NAD83, and was only able to tell because some of the points were known to have been collected with very high accuracy GPS units and they overlapped some high precision USGS survey markers. I haven't seen it in the past 5 years or so, but you might get data handed to you in NAD27 from legacy data sets, which could produce a significant error if you assume WGS84.

And I want to reiterate (myself and Khakis), assign your coordinate system as you're doing the Display XY process. ArcGIS can do strange things when you combine an unknown projection with a known. And for actual analysis it's a bad idea to rely on the on-the-fly projection capabilities.

[stapes]

Bookmarking this helpful thread.

I'm off to work on my first GIS final project. Good times at the computer lab

[flyandski365]

BAMP

Ok guys I'm back for more. Small problem but very annoying. ArcGIS keeps cutting 1-4 characters off of cells contents in various and seemingly random fields in the attribute table when I reopen a map. At first I thought it was auto-shortening my entries, but that shouldn't be the case because all my fields are set properly for the number and types of characters I need in them. So is this this thing just fucking with me or what?

EDIT: Ok so going back through the field properties shows me that it is a problem with the maximum allowable characters, but this is not the way I originally set up the attribute table - everything was fine before, but then I upgraded to ArcGIS10. Could it have something to the upgrade? Any way to change field properties without deleting a field and re-creating it?

[Random Task]

Hmmm... Depends on what format the attribute table is in. If it's in a personal gdb, just open it up in MS Access and change the table's properties in design view. If it's a .dbf or file gdb I think you'll need to add a new field and calculate it. Fortunately, this only takes a couple of seconds for one field - but if you have multiple fields with the problem, it's a pain in the ass.

We've had a handful of users switch to 10 at my office, this is the first I've heard of your problem - but there have been numerous other bugs, especially geoprocessing tools blowing up. Personally, I'm waiting until SP 2 or so to upgrade...

[bagtagley]

I'm not sure what in 10 would cause the change. Did you change the way you're storing the data?

If the data isn't in a PGDB, one option is to export the features back into their existing location with a different name. In the export dialogue, you can go through the fields and make the necessary changes. Then, delete the original feature and rename the new feature to the original name. Since it's a geoprocessing tool, you can do the same thing in a batch with Python.

RT, I've been on 10 since August. It's typical of any ESRI overhaul. Lots of features, minimal attention to detail. Though, it's worlds better than 9.0. Our web server is still on 9.3.1 (not for long), so we update all the projects behind our web services in ArcMap 9. I don't really like it anymore. It looks old and it's clunky to use. I don't regret upgrading at all.

[flyandski365]

I like 10 so far, a few hiccups here and there but overall it's gravy.

I just went ahead and created new fields and calculated them, didn't take too long as there was only a handful of fields with that problem. A solution so simple I never would have thought of it. Thanks again guys, beats the hell out of ESRI tech support.

Yours truly,

ArcGIS Jong