EXERCISE 4 – Keeping your feet dry

Environmental Resources 372:362
Intermediate Environmental Geomatics

Prologue - Data Scale

As an example of how the scale at which data is produced affects its accuracy and utility, let's take a quick look at the New Jersey coast.

Now load Z:\databank\njdep2\admin\state into the data frame. Next load Y:\intgeo\avdata\usa\states and restrict this data layer just to display New Jersey. Make it transparent. Compare the outline of the coasts between the two data sets. You'd get quite a different measure of the length of the coast from each of these data sets. Think a second about the coast of New Jersey. What's missing from both of these data sets? Add Z:\databank\njdep2\coast_shore\coastply. Ah, that's right. The state outline from the countrywide layer may be good for continental applications, but you wouldn't want to use it to determine sampling sites based on distance from Barnegat Bay, for instance.


Water, Water . . .

One of the most frequent uses of GIS is to find areas on the earth’s surface that meet certain conditions. For example, you want to buy a house – where would be the best place to look? Well, you might start with a few simple criteria to start narrowing down the possibilities. Such things as distance to work, location of good schools and avoiding congested roads seem a reasonable place to begin. You can make the treasure hunt for the ‘right house’ as complex as you want as long as you have access to the data. (Remember acquisition of data can be the most time-consuming part of any GIS endeavor).

Today you are going to do a similar hunt – right in your back yard. We are going to work with the Lawrence Brook watershed to see what types of landuse tend to be found in floodprone areas. We might be interested in this type of question for a number of reasons – you want a house with a dry basement (good luck), you want to reduce your flood insurance payments or you are a city manager who is interested in long term management of flood prone areas to minimize flood damage downstream.

Oh, Where to Begin…..

Saving the Selected:

Add landuse and flood from \\ad-rsc\data\teach\intgeo\Classwork\avdata\lbwpuse and become familiar with these layers and their attributes. We want to create a separate layer of water and floodprone areas in the Lawrence Brook watershed. Select those polygons that are water or flood prone.

Once you have created your selection, export this as a new layer (right-click on layer name, Data->Export Data) - it will contain just the portion of the flood layer that meets the conditions you specified. To keep things straight, rename the new layer to something you will remember.

Buffer for safety:

Generally, one would be interested in completely avoiding flood prone areas which can be accomplished by having a buffer around these areas (say 100 feet). To buffer the coverage we've just created we'll venture into ArcToolBox and use the Buffer Tool (under the Analysis Tools/Proximity Menu). Fill in the appropriate boxes, saving the output to your own directory (fill in the appropriate buffer distance, set the dissolve type to "all" and keep the rest of the defaults).

From the original flood layer, we now have not only flood prone areas but also buffers around them. It might be interesting to know what is the current landuse of those areas inside the buffer. Again to ArcTool Box. Under Analysis Tools/Extract select Clip and then clip the landuse by the buffered flood prone areas you created – careful, be sure you have the right order of the layers in the wizard. Again to keep things tidy and so as not to get confused, be sure to give your new layer a logical name. Clipping extracts features from one layer using the features from another layer as a cookie cutter.

After you clip, open up your output table. Sort by area, then select the polygon with the largest area. Now examine the selected feature closely. Notice something odd? There's more than one polygon associated with that feature. How can this be? Somewhat confusingly, a single feature (represented by a single record in the attribute table) can be made up more than one polygon. The clip creates one feature in the output layer for each clipped feature in the original layer. Depending on precisely how the layer to be clipped overlaps with our cookie cutter layer, there might one or there might be many polygons for each feature in the output. This isn't a problem for us, but it is something we should be aware of.

What is a problem for us is that the areas in that you see in that column are from the polygons in the original landuse coverage. You can verify this by checking the area of the corresponding polygon in the original landuse layer.

What to do? You’re lucky we’re using ArcGIS 9.2, this used to be a more annoying to fix.  Clear any selected features. Add a field called Acres.  Make it type double with a precision and scale of 0.  Then right-click on the field name and select Calculate Geometry.  Set the property you wish to calculate to Area.  Use the coordinate system of the data source.  Set your units to acres.  Hit Ok.  Now you’ve got the real area of your polygons measured in acres.  If you don’t believe me, repeat the process, but calculate the units as square feet.  By comparing the original area field in your clipped output to the new sq. footage field, you can then see which polygons in the clipped output differ in area from the original land use layer, and get some idea of how much ArcGIS would have overestimated the area if you didn’t fix this significant problem.

 

Creating a tally:

There is a lot of information just in this new layer and while you can get a general sense of what the predominant landuse is, let’s be a little more specific. In the attribute table of landuse in floodprone areas, right-click on the field LUCODE and summarize the different landuses by the sum of the areas. This will create a table that shows the cumulative acreage for each unique value in LUCode.

Homework:

a. Create a map of the above (land use in the buffered floodprone areas) following the rules of map-making.  Use the LUCODE for your map, but replace the numbers in the legend with appropriate verbal descriptions of the land use class. Use the following descriptions (it's up to you to match them with the appropriate LUCODE values): barren, water, forest/shrub, wetland, developed, agriculture. Be sure to use these descriptions instead of the numerical code on the maps and questions that follow as well.

b. Which landuse (using LUCODE) has the most features in floodprone areas? Which landuse (again using LUCODE) has the largest area inside the buffer? Answer in acres, please.

c. Create a separate map using the same information, but this time let’s assume we really don’t care about buffers and we are only interested in what the landuse is in the original, unbuffered floodprone areas. Answer the questions in (c) above with this new condition. In addition, which landuse changed the most when we did not consider a buffer? Remember to convert your areas to the units you used in (b).

d. Do another map, but this time let's be really floodprone sensitive and include a buffer three times the original buffer. Which landuses are now most affected compared to no buffer and the 300 foot buffer?

Homework is due before start of class on Monday, February 25th.