Lab 2: ArcGIS Spatial Analyst Tutorial

Lab Overview

In this lab we learned how to use the Spatial Analyst Extension (and respective toolbox) within ESRI’s ArcMap platform (version 10.7 as the 10.8 version is significantly different). The purpose of this interface is for the purpose of performing calculations related to spatial distance and ordination - for example, the distance between two points, the elevation of a specific ordinated point, or other elements of raster math which are important for spatial decision support tools (SDSS) and spatial analyses.

Exercise 1: Preparing for analysis

The data was saved to the student drive on the remote desktop that we accessed via the Cisco SecureVpn. Then we enabled the ARCGIS spatial analyst extension within the ArcMap Interface.

In exercise 1 we learned how to extract and organize data within the same area to pull for use within the ArcMap application. When using ArcMap, we learned how the location of data saves and pull connections is essential between uses of the interface to ensure that all map layers can be reloaded into the map AI without the possibility of a bad connection or bad gateway error. In this way, we saved the ArcMap scene (as a blank map) and our data file to a similar location for the purpose of ease between using imagery and the map AI that interprets the data files.

As this exercise used a specific geodatabase (a proprietary data object for ESRI products), we assigned this geodatabase as the primary workspace within the Map.AI’s environment settings. We enabled the catalogue window and opened up the files within the Stowe.gdb file for the purpose of this lab. We received a data pane that looked like this:

Figure 1. Catalogue Pane of extracted geodatabase in ArcMap 10.7

We then selected all data layers from the catalogue pane and imported them into the MapAI pane, saving the map now with the loaded layers from the downloaded geodatabase.

Exercise 2: Hillshade, Map Symbology

In this exercise we learned how to create a hillshade using an elevation dataset, and how to modify the transparency of other layers to make elevation data visible underneath land use and categorical data. We also learned how to summarize the occurrence of data (as landcovers) using a histogram.

Using the search console on the right side of the application, we searched and selected the hillshade tool from the spatial analyst toolbox. Executing the spatial analyst hillshade function, we created a hillshade data layer for our map that looked like this:

Figure 3: Excecuted Map hillshade layer from slope dataset for Stowe, Vermont.

We learned then how to alter the layer symbology in the layer properties to change the visualization parameters of the layer. We then changed the layer symbology to reflect actual representative colors (i.e., representing the water land use as blue, forests as green) that nearly correspond to a true colour composite in real life as what we would expect land uses to look like. On the display tab, we lowered the transparency of the land use layer so that the hillshade could be seen underneath the land use layer. The resulting image in the map AI looked like this:

Figure 4. Adjusting layer transparency in the Symbology pane for the hillshade layer, and selecting appropriate colors for LULC classifications.

We then opened up the attribute table and learned how we can select different features to emphasize them in the Map.AI through having them selected with a bright blue border. Likewise, we learned how we can identify layer features through using the inspector pane, and specifically the identify tool within the toolbar on the top of the map interface. In selecting wetlands, for instance, the following appeared on the map interface:

Figure 5: Illustration of the "Select by Attribute" Feature, with the respective features visualized in bright blue on the Map.AI interface.

Through selecting the histogram feature in the spatial analyst toolbar, we then created a histogram according the reported frequencies of each land use classification within the study extent. The histogram is as follows:

Figure 6: Histogram of different land cover frequencies in the Map.Ai extent

In this map extent it is clear that forest is the predominant land type, and agricultural land and water comprise the next amount of significant land, with built up are and wetland space as the next visible attributes.

Exercise 3: Toolbox and Executing Functions in Modelbuilder

Unfortunately, the model builder function was not properly downloaded onto the computer and functions had to be run within the catalogue feature.

We learned how within the map environment we can create a toolbox - these are known as the repositories for the new models created within the map file. A model, refers to a process flow for conducting a specific analysis or set of analyses within the ArcGIS toolbox. Within the toolbox, we named this model “Find School” as the function we are using would correspond to creating a new map layer that could be used to identify potential new school locations in Stowe, Vermont. We then adjusted the model environment settings so that the workspace it was running in corresponded to the map AI and the current geodatabase for the downloaded data in this lab.

We then went into the model environment settings and made sure the processing size and the cell size of the original data was preserved. There was an issue running Modelbuilder so I ran the functions independently of the model builder (error message: undetermined)...

Using the model builder I would have dragged the layer files into the interface and then the corresponding functions, and set their parameters. This was done manually. The following layers were an output raster for the slope calculation of the elevation, an output distance raster and direction raster (following a Euclidean distance function) for both the rec_sites file and the school file.

The following data layers look like this when computed for the landscape:

Figure 7. Euclidean Distance function outputs for i) distance from Recreation sites overtop of the Slope output

Figure 8: Euclidean Distance output for distance from existing School Sites.

We then learned how to reclassify datasets for the suitability layer which creates a distance weighted layer for suitability for new locations.

These are part of the MCDA tools that were integrated into the ESRI suite in the early 2000s as a means to support decision support tools. The weighted overlay tool requires the use of normalized data, so using raster math we rescaled the values of distances to 1 to 10; indicating least to most preferable respectively.

The raster math field in this instance was included in the reclassify feature as part of the spatial analyst toolbox with the Map.AI Catalogue. Using this function all values were reclassified between ranges of 1 and 10. We had to go into the classification parameters and change the amount of classes within the visualization parameters to 1-10 from 1-9. Once reclassified the map visualization changes. For example, the reclassified layer from slope looks like the following as a raw output - this will (like the other outputs) need to be changed to a colour ramp as opposed to a random visualization procedure.

Figure 9: Reclassified Slope layer for Input into the Suitability Layer.

After reclassification, data layers were weighted and combined to make the suitability layer for new school locations. The following weights include the following:

Reclassed distance to recreation sites: 50%

Reclassed distance to schools: 25%

Reclassed slope: 13%

Land use: 12%

In the Weighted Overlay tool, input the raster datasets and weight them respectively. The sum of weights is a good check to ensure your total adds up to 100% cumulative weight in the overlay layer.

Figure 10: Weighted Overlay and Combination tool used to make the Suitability Layer

The scale values were then optimized for each layer. For example, wetlands were restricted as those are areas that cannot be built upon. The resulting map layer was computed, renamed as suitable areas, and then visualized within the Map.AI. The resulting suitability layer was colour ramped from red to green to represent suitability better than the random colour ramp that is generically assigned.

Figure 11: Suitability Layer for the composite of variables

Optimal sites were selected following an aggregation procedure, where the conditional toolset was used and a majority filter for best areas. The procedure here is similar to creating a heat map or following a Getis-Geo-Ordination for optimal site selection, following a clustering procedure or “hot-spot, cold-spot” ideology. When loaded back into the Map.AI the most preferable areas are accentuated as discrete map features (in bright red below):

Figure 12: Results of the majority filter for possible and buildable land.

The best site for development was selected through first stipulating the average size necessary for a school, and then removing optimal area sites that did not fit that criterion. The size of each area was calculated through converting the raster cells into the polygon dataset. Using the select by location tool within the visualization toolbox, we put in the optimal area layer and selected locations that intersected with roads. This isolated potential sites to those in the southern extent of the map. Using SQL parameters in the "select layer by attribute" feature, we refined our search by using the subset selection and specifying parcels larger than the minimum size for a school with a value of “40469”.

Using the copy features tool, the selected layers were then isolated from the overall optimal areas layer. In this instance, the lab resolve did not give the pictured result from the tutorial even when trying to run the entire lab again. This is assumed that since the model builder package did not properly download, this probably means other features in ArcMap were not properly installed. However, the resulting final site selected included the following:

Figure 13: Resulting Final site after filtering out surrounding land using the majority filter and the select layer by attribute features.

Exercise 4: Finding a Least Cost Path

In exercise number four, we then learned about optimizing the isolation procedure for site selection using a best route approach, otherwise known as a “least cost path”. In this model the weighted overlay was created just using a reclassified slope and land use. The idea of this is to limit those in which steep slopes are used. Representing a higher cost, greater slope values (towards 10) were associated with higher cost to build. In this case, the result of this weighted overlay, the lower the value the better as opposed to optimizing the highest score for selection. The resulting cost-surface map is presented below:

Figure 14: Interpolated cost surface for the least-cost path tool

Running the cost distance tool, the cost of building between the site selected and the destination was computed as an output cost path. This was then converted from a raster to a polyline to optimize where a new road would be laid. The following map is the result over the raster convert to polyline function. The polyline symbology was accentuated to make the line visible over the land use and hillshade layer.

Figure 15: resulting line for the least cost path tool, converted to a polyline (in black) between existing site and future destination/location.

Conclusion

Overall, in this lab we learned about the basics of setting up an ArcMap processing environment, including toolboxes, Modelbuilder and the several different functions and visualization parameters that can be used to conduct functions, such as selecting a new school district in Stowe, Vermont. In addition to visualization, we learned about the beginnings of SQL code and how to program specific select operation tools as part of the advanced analytics add-on in ArcMap and ESRI products.

Google Sites

Report abuse