You have been told, and will be reminded here, that the spherical world is flattened in order to make distance and area calculations easier. So, typically you don't do GIS analysis in spherical coordinates (i.e., in geographic degrees of longitude and latitude). Now, there is no reason that software cannot be programmed to give you distance measurements in meters and area calculations in square meters, regardless of what the native format of the data is. And, to be sure, some of the tools in a robust GIS software program do just that. But, as we will see in this portion of the lesson, it is still a good idea to do GIS in planar coordinates--coordinates based on an appropriate map projection.
We are going to take advantage of an inconsistency in the software to illustrate that converting from spherical to planar coordinates, before attempting any analysis, visual or numerical, is important.
We will further demonstrate that it is additionally important to refine the parameters that govern that conversion from spherical to planar coordinates if we want our results to be as reliable as possible.
Learn more about projection parameters, or projection "metadata" in the Concept Gallery.
See also customizing a map projection in the Concept Gallery.
A. Create a Personal Geodatabase
- In ArcCatalog, navigate to your Lesson7 folder.
- In your Lesson7 folder, create a new Personal Geodatabase called eightymilebuf.mdb
Eventually we will be concerning ourselves with the areas of polygons. Since the feature class attribute table automatically has an area field in it, we will begin by converting the base data that we will use for this part of the exercise into the personal geodatabase format.
B. Convert Shapefile datasets to Feature Classes
- Since you were just using ArcCatalog, open the ArcToolbox pane from the ArcCatalog standard toolbar.
- Expand the Conversion Tools toolbox.
- Expand the To Geodatabase tool set.
- Double-click the Feature Class To Feature Class choice. This will open a dialog window.
- Click the Show Help button in order to display the Help for this conversion process. Note that the phrase "feature class" has been generalized to refer to any spatial vector data layer, be it from a coverage, a shapefile or a geodatabase.
You can hide the Help in order to lessen the clutter on your screen.
- Interact with the dialog window in order to specify the ...\Lesson7\fl_ga.shp shapefile dataset as the the source of the Input Features, your ...\Lesson7\eightymilebuf.mdb personal geodatabase as the Output Location, and fl_ga as the Output Feature Class Name. Leave the Expression slot blank.
Also, note that if desired you have the option to rename the fields prior to the conversion.
- Click OK.
- Verify that there is now an fl_ga feature class in your eightymilebuf.mdb geodatabase.
- Now, convert the fl_gaPlaces.shp shapefile dataset into a feature class called fl_gaPlaces, also in your eightymilebuf.mdb geodatabase.
C. Display the data we have
- Open a new map document in ArcMap.
- Save it as eightymilemap in your Lesson7 folder.
- Add the fl_ga and fl_gaPlaces feature classes from within your eightymilebuf.mdb geodatabase. (Do not add the shapefile versions.)
- If you have not done so, take note of the coordinate system in which these two data layers reside. ___________________________
- Right-click on the fl_gaPlaces layer and select Label Features. Make certain the points are labeled with the values in the AREANAME field.
- Locate Gainesville, FL, in north-central Florida. (There is a Gainesville, GA, but we want the city in Florida.)
- With the Select Features tool, or from within the attribute table, select the Gainesville, FL, point feature.
- Reposition and rescale the map so that the area from approximately Cape Canaveral (Titusville, on the east coast), north to the islands off southern Georgia fills the map display area.
- Save the map document. (And, do save as you go. Do not rely upon the lesson to keep prompting you to do so.)
D. Create a buffer zone with an 80-mile radius
If you do not have the Buffer Wizard available from the Tools toolbar (or another toolbar), go through the following 5 steps:
- Expand the Customize menu and click on Customize Mode...
- In the Customize window, click the Commands tab. It may take some time to become populated.
- From the Categories: list select Tools.
- From the Commands: list click and drag the Buffer Wizard onto the Tools toolbar.
- Close the Customize window.
Now, once the Buffer Wizard is available , make certain the Gainesville point feature is still selected.
- From the Tools toolbar, fire off the Buffer Wizard....
- You are going to buffer the fl_gaPlaces layer and check the bot to Use only the selected features. Click Next.
- Set a specified distance of 80 miles. Click Next.
- Let the name of the buffer zone layer default to Buffer_of_fl_gaPlaces, and be certain to save it as a feature class in your eightymilebuf.mdb geodatabase. Click Finish.
- Give the resulting buffer zone polygon a hollow fill. Change the outline color if you wish.
At this point your map display should resemble what you see in Figure 7.1.1, below.
Why is the buffer zone not a circle?
- With the Measure tool, measure the radius/diameter of the buffer polygon. Do so in both north-south and east-west directions.
Be aware of the units in which you are making measurements. Go to Choose Units in the Measure tool's dialog box. This will determine the units in which measurements are reported by the Measure tool.
Eighty miles is approximately 128.75 kilometers. On a great circle, approximately 111.32 km equals 1 degree. So, 80 miles equals about 1.16 degrees. Can you put the display units in Decimal Degrees and verify that the software did a pretty accurate job of making an 80-mile radius buffer zone in the context of spherical data? For that matter, can you put the display units in Miles and measure both the short and the long diameter of the buffer zone and come up with 160 miles (twice 80)? You should be able to. So, we can conclude that, even when the native format of the spatial data is in spherical coordinates, the Measure tool gives us pretty reasonable results in planar coordinates.
- Note the value in the Shape_Area field of the buffer layer attribute table (4.849440). What units is it in? What was the coordinate system you noted above, in step C-4?
E. Select by Location
Imagine a scenario where a pirate radio station in Gainesville knows that its broadcast signal carries a distance of 80 miles. They do not want to just see the cities and towns that fall inside that 80-mile buffer zone, they want to select the places that are within that zone and make a new dataset from that subset of places. That way the maps they make for bulletin boards and underground flyers, and the attribute lists they generate, will depict only those places served by their broadcasts.
So, let's try another of the GIS tools. The point feature for Gainesville should still be selected--make certain that it is.
- Go to the Selection menu, and choose Select by Location....
- Select features from the target layer fl_gaPlaces,
- have the source layer also set to fl_gaPlaces,
- using Target layer(s) features are within a distince of the Source layer feature spatial selection method,
- and set a search distance of 80 miles. Click OK.
Your map should resemble Figure 7.1.2. Well, what do you think of the Select by Location tool?
F. Select an appropriate projection
So, any sense of complacency that the Buffer Wizard had lulled us into should have been shattered by the results of the previous section. Remember, the coordinates of the data we are working with are spherical--decimal degrees. I suggest that we perform the Select by Location task on a dataset that has been projected into a more appropriate coordinate system. Let's choose an equidistant projection, since our concern here is accurate distances. We will perform a virtual reprojection by altering the coordinate system of the data frame and relying on the project-on-the-fly capability of ArcMap.
- Right-click in the map display area and bring up the Data Frame Properties... dialog.
- In the Coordinate System tab expand the Projected Coordinate Systems > Continental > North America and select the USA Contiguous Equidistant Conic choice.
- Take note of the projection parameters displayed in the Current coordinate system window, specifically the central meridian, the origin latitude, and the standard parallels. _______________
- Click the OK button.
The buffer zone polygon should now more closely resemble a circle. That is good...
G. Select by Location, again
Perform the same Select by Location exercise that you did above, in Section E. The software will remember some of your settings, but double-check them. You will need to remember to make it so that Gainesville is the only selected point in the fl_gaPlaces layer.
How did we do this time? Zoom in on Ormand-By-The-Sea. (It's on the east coast.) You should see that it did not get included in the Select by Location process...
What do you think we should do? Recall the projection parameters you noted above. Where is -96 longitude? Let's refine our projection.
H. More specific projection parameters
The area of least distortion incurred by the surface of the globe when it is flattened is controlled by the selection of the projection parameters. Typically it is at the center of a tangent map projection. Additional control over the amount of distortion can be applied by using the secant version of a projection. Secant, in this context, implies that something is cut more than once. In the case of a secant version of a conic projection, the cone intersects the globe at two standard lines, or parallels if the lines of intersection are latitudes. Along those lines of intersection the scale is accurate; it is not distorted; the scale factor is 1.0.
So, let's position the center of our projection actually at the location of Gainesville. And let's specify the locations of the standard lines such that distortion in the vicinity of Gainesville is minimized.
- Bring up the Data Frame Properties.
- Select the Coordinate System tab, if necessary. USA Contiguous Equidistant Conic should be listed under Current coordinate system.
We need to modify the parameters you see listed in the Projection area of this dialog box. But, in order to do so we need to determine the geographic location of the center of our soon-to-be-redefined coordinate system; We need to determine the longitude and latitude of Gainesville. From that pair of geographic coordinates we will determine the remaining projection parameters. And--look at the values in the list--we need those projection parameter values to be in decimal degrees.
First, let's find out the geographic coordinates of Gainesville:
- Put the coordinate system of the map display area back into so-called Geographic coordinates: GCS NAD83. Can you do it without my help?
Just in case: Geographic Coordinate Systems < North America < NAD 1983.
- We want a lon-lat coordinate pair for Gainesville in decimal degrees, remember. What are the units in the coordinate readout window in the lower-right of your ArcMap window? If they are not in decimal degrees, make it so:
Properties... < General tab, make the Display: Units show Decimal Degrees.
- Position the point of your mouse cursor over the point feature representing the location of Gainesville, and read off the location in the coordinate display window. Round the lon-lat values to the nearest tenth and record them. ______________
The rounded values for the longitude and latitude of Gainesville that I determined are: -82.3 longitude, 29.7 latitude. Do you agree?
If you look in the HELP at the description of the Albers Equal Area Conic projection (under the Uses and applications bullet), you will find a rule of thumb by which Standard Parallels are positioned within the north-south extent of the area of concern, and relative to the center of the projection. All conic projections are similar in this regard, so we can extend this rule-of-thumb to our equidistant conic. Without going through the math, if we put our Standard Parallels one degree to the north and south of Gainesville, we will be only 10 km away from the rule-of-thumb locations. And since 1 degree is easy to add/subtract to/from the latitude value we have, we will go with it.
Therefore, given the longitude and latitude that we determined for Gainesville, we can now specify the following projection parameters for the Contiguous Equidistant Conic projection:
- Central Meridian: -82.3
- Standard Parallel 1: 28.7
- Standard Parallel 2: 30.7
- Latitude of Origin: 29.7
- (Leave the False Easting and False Northing equal to 0.)
- So, change the coordinate system of the map display area back to the Equidistant Conic projection, but with our new parameters. If you are using ArcGIS version 10.2, select and then right-click on the USA Contiguous Equidistant Conic projection to select Copy and Modify... which will bring up the Projected Coordinate System Properties dialog box (or just double-click on the USA Contiguous Equidistant Conic projection).
If you are using version 10.1, click the Modify... button to access the Projected Coordinate System Properties dialog box.
- Interact with the parameter values, supplying the values specified in step 5 above. See Figure 7.1.3.
- Click Ok twice to close out the Data Frame Properties windows.
I. Select by Location, one more time
Once again, perform the same Select by Location exercise that you did above.
How did we do this time? Is Ormand-By-The-Sea included this time?
J. Save the results of the analysis
With the 158 places selected, let's save the results as a new feature class.
- Right-click on the fl_gaPlaces layer in the Table of Contents, and select Data | Export Data...
- Export the Selected features, and
- choose the second radio button in order to Use the same Coordinate System as the data frame.
- Save the selected places as a feature class in your eightymilebuf.mdb Name it pirateRadioPlaces.
- Go ahead and add it to the map session.
K. Save the map document
That is it for Part II
You have just completed Part II of this project, which involved heightening your awareness of the importance of an appropriately chosen coordinate system/map projection. Not only for visual analysis, but for overlay analysis.
The deliverable from this Part of the lesson will be your eightymilebuf.mdb geodatabase file.