Massachusetts Institute of Technology Department of Urban Studies and Planning 11.520: A Workshop on Geographic Information Systems 11.188: Urban Planning and Social Science Laboratory Lab Exercise 7: Raster Spatial Analysis Distributed: Lab 7 Due: Lab 8 Overview The purpose of this lab exercise is to introduce spatial analysis methods using raster models of geospatial phenomena. Thus far, we have represented spatial phenomena as discrete features modeled in the GIS as points, lines, or polygons--i.e., so-called 'vector' models of geospatial features. Sometimes it is useful to think of spatial phenomena as 'fields' such as temperature, wind velocity, or elevation. The spatial variation of these 'fields' can be modeled in various ways including contour lines and raster grid cells. In this lab exercise, we shall focus on raster models and examine ArcGIS's 'Spatial Analyst' extension. We shall use raster models to create a housing value 'surface' for Cambridge. A housing value 'surface' for Cambridge would show the high- and low-value neighborhoods much like an elevation map shows height. To create the 'surface' we will explore ArcGIS's tools for converting vector data sets into raster data sets--in particular, we will 'rasterize' the 1989 housing sales data for Cambridge and the 1990 Census data for Cambridge block groups. The block group census data and the sales data contain relevant information about housing values, but the block group data may be too coarse and the sales data may be too sparse. One way to generate a smoother housing value surface is to interpolate the housing value at any particular location based on some combination of values observed for proximate housing sales or block groups. To experiment with such methods, we will use a so-called 'raster' data model and some of the ArcGIS Spatial Analyst's capabilities.The computation needed to do such interpolations involve lots of proximity-dependent calculations that are much easier using a so-called 'raster' data model instead of the vector model that we have been using. Thus far, we have represented spatial features--such as Cambridge block group polygons--by the sequence of boundary points that need to be connected to enclose the border of each spatial object--for example, the contiguous collection of city blocks that make up each Census block group. A raster model would overlay a grid (of fixed cell size) over all of Cambridge and then assign a numeric value (such as the block group median housing value) to each grid cell depending upon, say, which block group contained the center of the grid cell. Depending upon the grid cell size that is chosen, such a raster model can be convenient but coarse-grained with jagged boundaries, or fine-grained but overwhelming in the number of cells that must be encoded. In this exercise, we only have time for a few of the many types of spatial analyses that are possible using rasterize data sets. Remember that our immediate goal is to use the cmbbgrp and sales89 data to generate a housing-value 'surface' for the city of Cambridge. We'll do this by 'rasterizing' the block group and sales data and then taking advantage of the regular grid structure in the raster model so that we can easily do the computations that let us smooth out and interpolate the housing values. I. Setting Up Your Work Environment 1. Launch the ArcGIS and add five data layers listed below. • M:\data\cam_lu99.shp Cambridge land use in 1999 per MassGIS Census 1990 block group' centroids for • M:\data\cambbgrp_point.shp Cambridge • M:\data\cambbgrp.shp Census 1990 block group polygons for Cambridge • M:\data\cambtigr (coverage) U.S. Census 1990 TIGER file for Cambridge • M:\data\sales89 Cambridge Housing Sales Data • M:\data\camborder Cambridge polygon 2. Set Display unit = meter. In this exercise you will use "Meter" instead of using "Mile"II. Spatial Analyst Setup ArcGIS's raster manipulation tools are bundled with its Spatial Analyst extension. It's a big bundle so lets open ArcGIS's help system first to find out more about the tools. Open the ArcGIS help page by clicking Help > ArcGIS Desktop help from the menu bar. Click the index tab and type "Spatial analyst. During the exercise, you'll find these online help pages helpful in clarifying the choices and reasoning behind a number of the steps that we will explore. Be sure, at some point, to take a look at the Overview section. The Spatial Analyst module is an extension, so it must be loaded into ArcGIS separately. To load the Spatial Analyst extension: o Click the Tools menu o Click 'Extensions' and check 'Spatial Analyst' o Click 'Close' Fig. 1. Add Extension Although you just activated the "Spatial Analyst" extension, you have to add the "Spatial Analyst tool bar on the menu manually to use the extension (quiteinconvenient!!!). To add "Spatial Analyst" tool bar, go to View > Tool bars from the menu bar and click "Spatial Analyst" Fig. 2. Add Spatial Analyst toolbar Once the Spatial Analyst tool bar loaded, a new main-menu heading called Spatial Analysis will be available whenever you launch the ArcGIS.Setting Analysis Properties: Before building and using raster data sets, we should set the grid cell sizes, the spatial extent of our grid, and the 'no data' regions that we wish to 'mask' off. Let's begin by specifying a grid cell size of 100 meters and an analysis extent covering all of Cambridge. To do this, click 'Spatial Analyst > Option '. When the "Options" window pops up. o In the General tab, set your working directory, select None for the Analysis mask (We will set the mask later), and select the first option for the Analysis Coordinate System. o In the Extent tab, select Same as Layer "camborder polygon" for the Analyst extent. o In the Cell Size tab, select As Specified Below then specify  Cell size--100  Number of rows--57  Number of columns--79 (Number of rows and Number of columns will be automatically computed.) Now that we've set the analysis properties, we are ready to cut up Cambridge into 100-meter raster grid cells. Convert the camborder to a grid layer using these steps and parameter settings: o Select Spatial Analyst > Convert > Feature to Raster. Features to Raster window will show up.  Choose camborder polygon for the Input features.  Choose COUNTY for the Field. (We just want a single value entered into every grid cell at this point. Using the County field will do this since