At the end of my post about defining the projection of AutoCAD files, I promised to write about Spatial Adjustment. Spatial adjustment is much like georeferencing an image, wherein you place control points at known common locations between the target layer and a reference layer, then run a process to align the two. ArcMap’s spatial adjustment toolbar allows you to do the same thing to a vector dataset using a slightly different, but familiar workflow. Please note that spatial adjustment should only be used when it is impossible to determine the projection of your vector data. Defining the correct projection will result in a much more accurate result.

I am returning to the city of Bloomington, Indiana and their AutoCAD files for this example. I’ve downloaded the landuse zoning dxf and the municipal boundary shp. I then projected the shapefile to a different coordinate system and cleared the projection definition so ArcMap is putting it in the wrong place. We’ll pretend you’ve received this shapefile and nobody has told you its projection. You need it to line up with the zoning AutoCAD layer.

It is more likely you would be doing things the other way around, that is, that you would have received AutoCAD data with an unknown projection. However, spatial adjustment only works while inside an edit session. You cannot “start editing” AutoCAD data. You would need to convert it to a shapefile or feature class before you could spatially adjust it.

That brings us to step one. Start an edit session. Then, go to View –> Toolbars and check on the Spatial Adjustment toolbar if it is not visible already. Use the main dropdown menu to Set Adjust Data and chose your target layer (in this case the municipal boundary shape). You will have the option to adjust all the features or only the selected features. We will do all of them.

Then, press the button that looks just like Georeferencing’s “Add Control Points” to add your first Displacement Link. The lingo is different; the concept is the same.

Click on a distinguishing feature on the municipal boundary and then on that same feature in the zoning layer. Remember, you always start with the layer that you want to move and end by telling it where to go.

One difference that takes some getting used to is that there is no Shift or Fit to Display ability like there is in georeferencing. So, you can’t get the target layer and the reference layer into the same window before you start adding displacement links. You will have to keep zooming back and forth between them. If you can’t place your links as precisely as you’d like due to this, don’t worry, you can get them in the general vicinity and then press Modify Link to edit them.

Another difference between this and georeferencing is that placing links doesn’t automatically cause the layer to move. That only happens when you select Adjust from the dropdown menu. You will not be able to select Adjust until you have placed three links. I recommend doing so as soon as you can just to get things into the same universe. (There’s no limit to how many times you can re-adjust later.) Just like with georeferencing, place your three links as far apart from each other as possible.

Once you have adjusted, your links will disappear, and you will be able to refine your results by adding more links.

(After first adjustment)

At this point you may also want to experiment with different adjustment methods. The first three in the list are coordinate system transformations. You should stick with these when the only reason your data isn’t lining up is that you don’t know the projection.

• Affine (the default) will scale, skew, rotate, and translate the data.
• Similarity will scale, rotate, and translate but it will not skew. It will not change the aspect ratio or the shape of features– only their size, position and orientation.
• Project uses a more complex formula that is tailored to features that have been captured from aerial photography.

The last two should only be used if you know there are errors within your target layer that make it less accurate than your reference layer.

• Rubbersheeting behaves like it sounds– pulling and stretching your layer in a piecewise fashion. The areas that are closer to displacement links will move more dramatically. Identity links (the black and red square crosshair button) can be used to nail down features that should not move. If rubbersheeting is used it should be as a second step after transformation.
• Edge Matching is used to merge adjacent layers that have been split up. It makes sure features line up along one edge only.

You can preview the results of different adjustment methods in the preview window. You can also use the preview window to see the effects different displacement links will have.

Another way to check your work is by looking at the link table. If you are using a transformation adjustment method, get an idea of your RMS error. If you place a link that increases the RMS and Residual errors substantially, delete the link and try something else. I recommend keeping both the Preview Window and the Link Table open while you are working.

When you are happy with what you see, make a final adjustment and then save your edits.

A multipart polygon is a group of polygons that have gaps between them, but only one attribute table entry (and therefore only one object ID). They might be used in a case where there are multiple buildings on the same parcel, multiple zones of influence for the same actor, or multiple lilypads on the same pond. If all the attributes are the same, might as well use the same record for all, instead of a long list of repeating values. You will know you are dealing with multipart features if you click on one polygon to select or identify it, and other polygons light up too. But how do you create one?

I know of two ways: one I have been doing for a long time and another I just learned. I will start with the new way, although I think I still prefer the old. Both sets of instructions assume you have already digitized the first polygon.

Tacking on a multipart, Way #1:

1. With the editor tool (black arrow), select your polygon.



2. Change the edit Task to Modify Feature.



3. Switch to the sketch tool (yellow pencil). Lines will extend out between your starting and ending vertices.



4. Hover over one of these vertices and right click to Finish Part.



5. Digitize your next polygon like normal. When you have created the final vertex, right click to Finish Sketch.

Or, select Finish Part if you want to create additional polygons in the multipart feature. Finish Sketch when you’re done creating the last one.

You have to follow these steps exactly as I have written them or it won’t work. I find this procedure a little nit-picky but it may just be because I am not used to it yet. The only advantage I can think of is that you never end up of creating a second attribute record this way, which may help in keeping object IDs straight.

Tacking on a multipart, Way #2:

1. Digitize your second polygon in the normal way.
2. With the Edit tool, hold down the Shift Key to select both polygons.
   (Or, click and drag to draw a box around them)
3. Select Merge from the Editor menu.

Try them both and pick your favorite!

This comes up a lot. Perhaps you’re digitizing a lake, and it has an island in the middle of it. Or there’s an area of lawn surrounding a building. If you’re starting from scratch, I recommend drawing lines and converting them to polygons. Similarly, if you need to make a lot of holes, I recommend converting your polygons into lines, editing the lines, then converting back to polygons. That method is simpler for large scale changes.

But sometimes you just need a quick hole or two right now. For instance, you might be working in the Planning Office for the city of Lansing, Michigan. You might be in charge of maintaining their Existing Land Use Map, shown below.

In the southwest quadrant of the map, there is one region of Institutional land that has a Single Residence in the middle. Let’s pretend that house was just built, and you’ve been asked to update the map.

Here’s how you could do it:

1. Make your Land Use polygon layer partially transparent so you can see what you’re doing underneath.

Howto: Right click on the layer in the Table of Contents to bring up Layer Properties. On the Display tab, enter a number greater than 0 and less than 100 in the Transparent % box. 50% is good.

2. Enable editing and double click on the Institutional polygon to show the sketch vertices. Note: Double clicking brings you into Modify Feature mode, allowing you to move vertices in the existing polygon. A single click is Create Feature mode, which draws a new polygon.

3. Right click on the edge of the sketch, near where you want to create the hole, to Insert Vertex. Then create two more vertices next to it.

4. Pull the center vertex in towards one of the outer corners of your hole.

5. Create as many new vertices as needed to drag out to all of the other corners.

6. Prepare to close the empty space and the filled in space by placing two vertices at each location close to another. You’ll end up with a narrow rectangle connecting the edge of the filled space with the edge of the empty space.

7. The idea is to turn this rectangle into a line by merging the vertices at the top and bottom. To do this, turn on snapping to sketch vertices.

Howto: Select snapping from the Editor drop down menu and check the Edit sketch vertices box.

8. Pull the vertices together.

When you click away, that connecting line will disappear, giving you the two distinct areas you want. Now, it’s a simple matter to draw a new polygon for the Residential area within that hole. Just snap to its four corners.

Make sure that snapping to the editable layer, and not just the sketch, is turned on for this.

Note: If you are just creating a visual map, you can draw the smaller polygon on top of the larger, and then set Symbol Levels to ensure they display correctly. (Symbol levels are available from the Advanced drop down in the Symbology tab of the Layer Properties dialog). This is a quick fix if you’re on a tight deadline. It is not recommend if you’re going to be doing analysis on the layer or distributing it to others.

Another Note: There are other ways to do this, such as using the Cut Polygon Features task (for a hole), or the Clip task (for a polygon inside a polygon). ArcMap’s Help describes both of these methods. I like following the steps I have outlined because I can accomplish everything by clicking and dragging without switching between editing tasks. Within GIS, there are many ways to do most things, and you will develop your own style.

This isn’t as straightforward as it initially appears. If you need to digitize some points of interest, you create an empty point layer and start clicking away. If you need to digitize some river or road lines, you create an empty line layer and start tracing. But if you follow the same logic to create polygons, unless those polygons don’t touch, you’re going to run into trouble.

Polygons that are separate from one another, such as building footprints or lakes, no problem. But polygons that form a continuous surface, such as landcover, administrative areas (e.g. counties, zip codes), soil types, or flood plains, big problem.

Why? Well, let’s take the example of a soil map. I downloaded this one from the Antrim County Community Center. It’s a 1928 Soil Survey of Antrim County, Michigan.

I zoomed in on an area of Antrim gravelly sandy loam, and drew my first polygon. Then I created a second polygon for the Lupton muck next to it.

Looks fine, except when we zoom in. The polygons don’t line up. Those empty spaces in between are called slivers. They are to be avoided at all costs. They will mess up feature counts and calculations of areas. They don’t represent the real world, and they look bad. However, they are virtually impossible to avoid if you are digitizing in this manner. You can try to by zooming really far in while you are drawing, and being really careful, but what a headache! Fortunately, the cost is not that high.

One way to avoid slivers is by the use of snapping. The snapping toolbar is available from the Editor dropdown menu in ArcMap. Here, I have turned on snapping to vertices in both the existing polygons and the new one I’m drawing (the sketch). Now, when I hover close to a vertex, the mouse will jump right on top of it. Snapping makes it impossible for you to create a vertex in your new polygon that is close to a vertex in any nearby polygon. It forces all vertices to be in the exact same place, or far away.

Note: This will make creating very tiny polygons difficult. If you run into trouble with this you will need to change your snapping tolerance by selecting Options in the Editor dropdown.

Snapping is enough to fix the problem if you have just a few polygons or a lot of straight lines (like rectangular buildings). But in the case of this curvy soil map, you’re still going to have to zoom in and be careful to “hit” every vertex. It only offers a small help.

A much better way: Don’t digitize the polygons. Digitize the lines between the polygons.

Create an empty line file, and don’t worry about attributes yet. Zoom in to the scale you plan to digitize at, and don’t deviate from that zoom. Turn snapping on so that intersections meet. Then, just pan around and draw.

Now

1. Create an empty polygon layer and add it to your map
2. Start editing and make the polygon layer your target
3. Select all the features in the line layer
4. Turn on the Topology menu, and use the Construct Features tool. This will create a polygon from every distinct closed area.

Note: Snapping is important for the tool to recognize these areas as closed. If your result has holes, it is because you missed a spot.

Congratulations! You have just about halved the number of vertices you had to draw, and created truly contiguous polygons. Now, add your attribute fields. It’s a good idea to save the line file in case you need to go back and make changes later.