Lab 10: Raster Analyses - GIS Courses€¦ · GIS Fundamentals Lab10: Introduction to Raster Analysis 1 Lab 10: Raster Analyses What You’ll Learn: Spatial analysis and modeling

GIS Fundamentals Lab10: Introduction to Raster Analysis

1

Lab 10: Raster Analyses What You’ll Learn: Spatial analysis and modeling with raster data. You will estimate the access costs for all points on a landscape, based on slope and distance to roads. You’ll then apply a threshold to this access cost. There is a mix of old and new functions used in this lab. We’ll explain the new, but you are expected to review old labs if needed to apply the old. You should read chapter 10 in the GIS Fundamentals textbook before performing this lab. Data are located in the \Lab10 subdirectory, including Valley3 and Valley9, three- and 9-meter DEMS of a portion of southeastern Minnesota and mardem, a 30 meter resolution raster elevation grid, and Shasta, a 30m DEM in northern California. Also provided is a DEM for the Duluth area and a DuluthNorthRoads layer. All data are in NAD83 UTM coordinates, Z units in meters, the Minnesota files in zone 15, and the California file is in zone 11. What You’ll Produce: Two maps. First, as practice, you’ll create a mosaic of the valley DEMs. Then you’ll fix the Shasta DEM and produce a hillshades corrected elevation map. Then you’ll create the second map, a cost surface with an applied threshold, and some of the intermediate data used for the cost surface; all these 4 parts of the cost surface exercise will be on the same layout.

Project 1: Raster Resampling, Combination, Filtering We often obtain DEM data with various errors, a cell size different that we wish to have, or at different resolutions for different parts of our study areas. We may use raster operations, or functions, to improve our DEM data. In our first project, we will mosaic data of two different resolutions. We often have data from various sources, for example, DEMs at approximately 30m and 10 m for most of the U.S. and 3m resolutions or better for a small subset of areas. Our first task here will be combining two DEMs, valley3, at 3-meter cell size, and valley9, with a 9-meter resolution. We want to use the higher resolution data where we have it but use the lower resolution data elsewhere.


2

Start ArcGIS Pro and add both the valley3 and valley9 DEMs to an empty Map Frame. (If valley3 or valley 9 don’t load; delete your entire Lab10 folder and copy a new, fresh copy of the entire

folder to your L:\home and try again)

Calculate the hillshade for both data sets (using standard default settings and Model shadows) (Analysis→Tools→ArcToolbox→Spatial Analyst Tools –> Surface –> Hillshade). Inspect these hillshades carefully, and note the enhanced detail with the 3 meter DEM, as shown below. The figure on the right is from the hillshade of valley3, on the left the valley9. Note the greater definition of the small streams and streambanks.

Remove the two hillshades from the data view to reduce clutter. We can use the raster calculator to join these two data sets. Note that we want to use the detailed valley3 data where we have it, and the coarser valley9 data everywhere else. Resampling To avoid unforeseen results, we should first convert our data sets to a common resolution, in this case the 9 meter data to a 3-meter cell size. This does not make the 9-meter data better. Basically, we are making a copy of the data at a finer resolution. (Video: Resample)


3

Use Analysis –> Tools–>ToolBox –> Data Management Tools –> Raster –> Raster Processing -> Resample In the resultant window, specify the valley9 as input, an output filename, a 3-meter cell size, and a bilinear resampling (the textbook describes the differences among resampling methods). Name the output raster valley9to_3. After the resampling is done, examine the valley9to_3 and verify that it has a 3-meter resolution, either via the table of contents – properties, then source/raster information, or by zooming on the map and using the measure tool. We can now combine the two data sets. Although there are many ways to do this, perhaps one of the simplest is with clever use of two raster functions – IsNull and Con. The textbook describes these in detail, but in short, the IsNull returns True whenever a cell comparison or value is Null. The con function takes three values, the first is a true/false test, and second is the value to assign to an output grid if the test is true, and the third is the value to assign if the test is false. (Video: Raster Calculator) Using the Raster calculator, we can nest these functions: Analysis –> Tools–>ToolBox → Spatial Analyst Tools → Map Algebra → Raster Calculator Con(IsNull("valley3"),"valley9to_3","valley3") Name the output something like combinedDEM


4

Be careful to type the expression as shown, the operators are case sensitive, meaning “con” is not the same thing as “Con”. The Raster Calculator is a rather rigid parser. It is easy to add extra spaces and other minor syntax errors that will keep your expression from executing. Pick the operators from the Tools or Raster windows, when possible and enter data via the keyboard only when necessary. This function first tests if the cell of Valley3 is Null. This is true everywhere in our study area outside the data region of Valley3. When the value is Null, the con function assigns the value found in Valley9_to3 to the cell in the output data set. When the value is not Null, the con function assigns the value found in Valley3. Examine the combineDEM raster. Compute and inspect a hillshade and verify that it has the higher detail contributed by the Valley3 data set. Now we’ll look at another way to fix DEMs with errors. Add a new empty Map Frame and remove the Map Frame that contains the valley data. (Practice is over, this is start of the 1st turn-in exercise) Raster Filters We’d now like to introduce filtering as a tool to fix “noisy” data. This is often used with interpolated surfaces, particularly LiDAR data, and similar tools are used near edges for mosaiced DEMs and other continuous surfaces. Filters are described in Chapter 10 of the GIS Fundamentals textbook. Load the DEM named Shasta, and calculate a hillshade surface for the Shasta DEM. Leave the Azimuth as is, set the Altitude to 25, and check the model shadows box. Inspect the output hillshade layer and notice the funny artifacts. These are both data “spikes,” white points with a long thin shadow trailing to the southeast, or “pits,” dark areas on the northwest with a white “edge” on the southeast. How do we remove these? As described in the text, we may use a low-pass filter to identify and get rid of this speckle. We may apply a low-pass filter to the SHASTA dem (not the Hillshade) with the Analysis –> Tools–>ToolBox –> Spatial Analyst Tools –>Neighborhood –> Filter command.


5

Specify the Shasta DEM as the input, and a filter type of Low pass, and specify an output name, like Shasta_lo. Run the filter, calculate a hillshade for this new layer, using the default parameters, and do not Model shadows. Inspect the output, looking at the areas that have previously had spikes and pits (zoom way in). The before and after figure below left and right, show the same corresponding area above, it shows the reduction in the size of the spikes and pits, although they are still visible. We could stop here, and just accept the filtered data layer. But if we look carefully at the filtered and unfiltered hillshade, we’ll see we pay a cost for filtering. We lose some of the fine detail, apparent in the image of the unfiltered hillshade to the left, below, compared to the filtered hillshade to the right.

We’d like to keep both, and we can. First, we should subtract the filtered DEM layer from the original Shasta layer. (remember not to

accidentally use a hillshade layer; Hillshades are used for visual reference, seldom for any analysis) Analysis –> Tools–>ToolBox –> Spatial Analyst Tools →Map Algebra→Raster Calculator Select the following in the Raster Calculator "shasta" - "shasta_lo" Name the output difference.

(see figure at right); then Run. Calculate the hillshade for the difference layer you just created.


6

The figure at right shows a zoomed in version of the hillshade of the difference output, with the largest differences at and around the spikes and pits, seen as squares of contrasting shades. (Shown for

your information; you do not need to create a hillshade of your difference layer) We can now then replace the cells were the difference is large. Here, large is a relative term, but after a few trials, and looking at the difference histogram, a threshold of about 15 works fairly well. We want to replace the cells in the shasta image when they are more than 15 meters different from the filtered surface. Otherwise, we leave the Shasta surface alone, and hence, we don’t get any of the degradation in detail in otherwise good data. We can open the Analysis –> Tools–>ToolBox –> Spatial Analyst Tools →Map Algebra→Raster Calculator, and apply the following function, also shown within the raster calculator in the figure following: Con(Abs(“difference”) > 15, “shasta_lo”,”shasta”) If the absolute value of the difference is greater than 15, we write the filtered value to the output. Otherwise, we write the original data value. Name the output something like smoothed. We can verify that this is helpful by viewing the hillshade of the output raster (again using default parameters and not modeling shadows), and comparing it to the original, and the filtered surfaces. Note that we have removed much of the speckle but maintained the detail.


7

We could apply the filter successively, and average the local points further, and apply the Con function to a difference layer again. It would lead to an improved surface, and we would do this if we were using the data for a project where the noise reduced our accuracies. But since you won’t learn much new with repetition here, we’ll just produce a map and move on to project 2. For display, calculate a hillshade of the smoothed raster, name it something like HillSmooth. Use an altitude of 35 degrees and model shadows. Add the shaded relief (that is the HillSmooth), making it semi-transparent (50-60%) over the smoothed elevation data. (See box at right for a “check” of the values of your layers to this

point; check the value ranges with your results”) Make sure you model shadows in the shaded relief surface (points will be deducted if you do not – we need it to evaluate how well you’ve applied the filtering). Create a Layout as shown below, add a reasonable legend, name, north arrow, description, and scale bar, and export the map as a .pdf. In the figure below, the symbology stretch type is set to Standard Deviations, n=1.5


8

Project 2: A Cost Surface We will create a cost surface for locating a building. Our cost surface will depend on slope and distance to existing roads. In our problem, we will assign a road construction cost of $25 per meter of road required. We have a vector data layer of roads, digitized from USGS maps, and we will use grid functions to convert this to a cost data layer. Slope also affects access costs, because roads on steeper terrain are more expensive. The cost is nonlinear, increasing slowly at first for low slopes, then more rapidly at steeper slopes. The relationship is not linear, in that it is more than 10 times as expensive to build a road on a 20% slope than a 2% slope, and impossible to build over a steep threshold. We provide both a roads data layer and a distance data layer. You will calculate a slope cost using the raster calculator. We use a few formulas for a very crude m3odel of costs, a more accurate analysis would require much more complicated processing, but our approach gives you the general idea while gaining practice with raster calculations. We will then calculate a distance cost, and then total cost by adding both layers. We will then add these two cost layers. Finally, we wish to apply an upper threshold of $25,000 to consider only those areas that are within our budget.

Start ArcGIS Pro. Create a new project, add a map, add the raster DulNorthDEM, and calculate the percent slope, Analysis→Tools→Toolbox→Spatial Analyst Tools→Surface→Slope

Specify and remember the name you assign for the slope layer, here I use Slope_tif1


9

Add the output slope surface to the map and inspect it. Use the cursor and the TOC slope layer Properties –Source tab. What are the slopes? What are the highest and lowest values? Do they appear reasonable?

Next apply a formula that determines the cost of building on slopes: Left click on Analysis→Tools→Toolbox→Spatial Analyst Tools→Map Algebra→Raster Calculator Use the mouse to select the equation Exp(“Slope_tif1” * 0.2) as shown in the window at right, where “Slope_tif1” is the name I chose for my slope layer in the previous calculations. You should substitute your slope layer name if it is different. This applies an exponential function to each cell value and saves it in the output raster slope_exp. If you inspect the output, you’ll note it has a lowest value of 1, and a very large highest value, something like 2.38 x 1022 (note image will appear almost solid black) We need to shave off the highest end, as it can cause functions to calculate too large a value later. We want to place a ceiling on values over 300.


10

We use the Con function in the Raster Calculator, applying the condition as shown below, Con((“slope_exp”) > 300, 300, “slope_exp”) Verify for yourself that this makes sense – remember it is Con (condition, value if condition is true, value if condition is false) The thumbnail below shows a green to red color ramp applied to the symbology, with the red areas showing steepest slopes.

We now calculate slope costs. Our formula states that is $1000 times the slope factor, starting at 0 extra dollars for a flat (0 slope) surface. Using the raster calculator as shown above, we can derive a slopecost raster. Apply the formula “slope_exp300” * 1000 – 1000 Any idea why the final -1000 at the end of the formula? What is the exp(0)?


11

Assign the symbology as shown at right. Note we’ve specified the Min-Max stretch type, with custom statistics, and specified 0 Min, 25000 Max in the Options (Video: Custom Ramp)

Compare your Map to the thumbnail image below to verify your slopecost function is calculated correctly. You should have something that looks like this, when applying a max-min symbology over the 0 - $25,000 range to your slope cost.


12

Add the DuluthNorthRoads vector layer to your map. Save your Project before going any further! Calculate the distance from roads raster, via Analysis→Tools→Toolbox→Spatial Analyst

Tools→Distance →Euclidean Distance.

• Use DuluthNorthRoads as the input

• Name the output raster distance.

• Set the cell size to 3

• Don’t specify an output direction raster

• Distance Method Planar (Video:Distance)

(Note: if this operation runs but never seems to end; i.e. stays at 2%. You can try to Cancel the operation but that might not work. You may have to Exit ArcGIS Pro or select the top Citrix Dropdown and open Task Manager and end the ArcGIS Pro Task. You can also restart Citrix. When you get your Project back open you can try: Select the DuluthNorthRoads in the Table of Contents, Data, Export Features, use the default output name, Run; then repeat the Euclidean Distance on the new copy of the roads. IF ALL ELSE FAILS use the distance layer we created for emergency L:\FNRM3131 or 5131 \Lab10\distance_if_yours_doesnt_work.tif. This layer is already complete, just use it for the next step)

You should get something that looks like the above figure, depending on your color ramp: (used in this image “classify with 5 Natural Breaks (Jenks))


13

Now: Calculate distance cost, as $25 per meter:

(used in this image “classify with 5 Natural Breaks (Jenks

))

Now add the distance and slope costs, to calculate a totcost surface.


14

You should get a layer with range of values approximately from 0 to 350,000. The high end is from the steepest slopes, and it is a long-tailed distribution, meaning there are few, very high values which skew the symbology stretch, and obscure much of the variation. If we cut off the stretch at 10,000 (see figure at right), then we show more detail. The detail is shown in the thumbnail of the layer below. Note the drop-off near the roads at this stretch, with a threshold above 10,000 reached due to the distance function.


15

We now introduce permanent reclassification. This is a conversion from one set of numbers to another, into a new raster. We do this in a raster GIS through a reclass table. This table has a column for input values (Old values in the figure below) and a column for output values (New values in the figure below). Each cell value is examined, and input value matched to an entry in the table, and the corresponding output value reassigned according to the table. For example, the table at right specifies that all Old values between 0 and 200 are assigned a new value of 0, values between 200 and 400 are assigned a value of 1 and values from 400 to 10,000 are assigned a value of NoData. In a permanent reclassification, output values are saved to a new raster. This is different from reclassification when we specify the symbology. Up until now you’ve been specifying lists of values for coloring a display. No data are changed in the source file, nor are new files saved. In previous lessons we have only performed reclassifications for display. Today we will perform a permanent reclassification. It is easy to get confused, because the classify menus are similar these two classifications. We wish to consider only areas where the total costs are below $25,000 We need to create a mask, with 1 where the total cost is below $25,000, and NoData where the total cost is above $25,000 Review the video Reclass for instructions on how to use the Reclassify tool Start the tool by ArcToolbox→Spatial Analyst Tools→Reclass→Reclassify.

This is an example

This data not is used in this lab!


16

In the reclassification window (not shown), set the

− Use totcost as your input layer,

− Add entries by filling in the start, end, and new fields. When you finish a row, a new blank one should appear below.

− The first row should Old values of 1 to 25000 to New values of 1,

− The second row should assign values above 25000 to NoData. Here we specify the largest value, 350000, but any number larger than the largest value in the data set will.

− Add a row that assigns NoData to NoData

− Place the output in a raster named classmask

This should result in a reclassification table as shown at right: Click Run to execute the function


17

This should result in a mask layer called “classmask” Use the raster calculator to multiply the mask (“classmask”) by the total cost (“totcost”) layer (instructions not shown, refer to earlier parts if you are unsure how; (hint: see the geoprocessing tool we introduced at the bottom of page 3 in these instructions)); call the output something like Final_Cost. It should look something like this: Create a 4-panel pdf map of your work: Create a composite layout with four separate maps on the same layout. You’ll have to create maps with 1) the distance cost layer, 2) another map with the slope costs layer, 3) another map with the mask layer, and 4) a map with the final cost layer.


18

Add the roads layer to each map. Use a contrasting color for the mask, and color the distance and slopes costs and the final cost as graduated colors, Include the appropriate legend for the layout, with a title, name, north arrow, legend, and scalebar

Note that you should make all of the maps the same size in the layout view. You can do this by right clicking on a frame, and then selecting Properties then Display Options in the formatting pane at the right of the main window,

then modifying the scale. Experiment with scales to find one appropriate for the maps, then use the same one for all maps shown in the layout.


19

(Video: MapSize&Scale) You should turn in a map that looks something like: Note: your Distance and Slope surfaces should be stretched over the same range as the Final Cost surface (use Symbology, Stretch, Stretch Type =Standard Deviation 2,

Statistics = Custom, Max =25000, and also change the Label value to $25,000 )

Documents

Lab 10: Raster Analyses - GIS Courses€¦ · GIS Fundamentals Lab10: Introduction to Raster Analysis 1 Lab 10: Raster Analyses What You’ll Learn: Spatial analysis and modeling