The assignment is to:
The basic strategy for screen digitizing with ArcMap is to: (1) georeference (in the ArcMap GeoReferenceing Tool) (if it is not already georeferenced) the document that you want to digitize from (map, photograph, remotely-sensed image, etc), (2) create an empty shapefile (in Arc Catalog, though you may prefer to use a new feature class in a geodatabae), (3) display the georeferenced base image and add the empty shapefile/feature class, and (4) digitize the desired content into the (once empty) shapefile/feature class.
You need ArcMap's GeoReferencing Tool.
With the image and the street lines (or GPS points) loaded in your display, you will note that they probably don't overlap at all before we georeference the image.
Examine the photo to determine approximately where it fits on the street map (or GPS points).
Zoom the street map (or GPS points) to show approximatly the same area as the photo's extent.
Make sure the photo name is in the georeference toolbar
use Georeference->Fit_to_Display to get the photo and street (GPS) data displayed about on top of each other.
Select an initial tie point to bring one point of the image into correspondence with the street map. Perhaps a prominent intersection will be available.
Select a second tie point to improve the fit. (Perhaps another intersection.) This should let the image be rotated and scaled closer to a fit. Select a third point and you will probably see a pretty good fit.
Selecting more points, and perhaps messing with the geometric model, (the order of the polynomial, the number of tie points) may get your fit better, (but remember that the allroad3 street data are from the census bureau and their concern was consistent topology of census blocks rather than with exact location and shape of features... your GPS points may be better). Once you are satisfied with the image's position relative to your base data, save the now georeferenced image with its coordinate system info.
When you are selecting tie points, click the image first and the map second; like you are stretching the image to fit the map.
You can access the table of tie points, determine which are contributing the greatest error and change them.
You can set the transparency of the image in the layer's property dialogs. This is particularly helpful when georeferencing image to image, but can make it easier to see the street map's linework too.
Make sure that the editing tool is available under the right and left click on the Tool menu item.
In ArcCatalog, create a new shapefiles of the appropriate type, (point, arc/line, polygon) with the appropriate coordinate system. In ArcCatalog, connnect to your U: directory, Then use File->New->Shapefile. Name the file, select the type of features it will contain, edit/select/copy a spatial reference system for the file (import the one for the allroad3 data).
Back in ArcMap, add the empty shapefiles to your display.
You need the editing toolbar panel, so click the editor button (pencil with four nodes and three line segments) if you don't see the panel.
Editor -> start editing. Select which layer to edit (your new shapefile). Select the task to perform (Create New Feature).
Select the editing pencil tool and click on, along, or around the things you want to digitize. You can zoom and pan to make this easier.
Turn editing "off" and save your work when you are done or need to take a break.
With editing "off" you can add fields to the tables as you did in previous labs with the options button on the table display.
Read the help on all of the tools to help you cut, reshape, etc the coordinates.
There is a snapping function in the editor tool to help you, but I don't think that it works with images (yet). Within a shapefile, and when you digitize over another shapefile use "snap" to make points coincide with existing points. You can also use it to snap within the layer you are digitizing. If it seems appropriate, snapping can produce cleaner (looking?) data sets, at least they will match each other.
Having the Editing Tool "on" also permits editing the data in the associated table interactively.
Click "clockwise" around polygons.
Double-click to end lines or polygons.
The task is to estimate the area of intersection between two sets of polygons. This is a classic case of overlay analysis: you have two sets of polygons and you want to see where they are coincident. Arc has several options in it's toolbox, so...
Click on the red toolbox icon to open the toolbox panel.
Then, chase the hierarchy down to the intersect(ion?) tool:
Analysis -> Overlay -> Intersect.
Choose the shapefiles you want to intersect (your trees and your buildings) and indicate where to write the resulting set of polygons (shapefile). Hopefully the tool runs withuot errors and adds the new shapefile to the display.
Once you have this new shapefile of polygons of intersection, add a new field (call it area and make it a double) to the shapefile's table, then calculate the areas by right click at top of column, like when you used the "field calculate" function, but note that current versions of the software have a "calculate geometry" choice. Use it to calculate the polygon areas. (NB it assumes that the data or the display frame are in projected coordinates (like UTM rather than WGS84 plate carre). If the dialog tells you "Area - disabled" set the display frame properties to be a projected coordiante system.)
Back to the task... It might make sense to report the area of intersection as a percentage. But... of what? Total tree area? Total building area? Total block area? And, if the latter, consider how to count road areas fairly. Should a "block" include out to the street centerlines? Using the near or far edge of the road would give you a different estimate.
You might want to refresh your digitizing skill by quickly digitizing a "total area" polygon for the blocks and calculating its area as above.
(In older versions of the software, you used a bit of VBA (that you could cut and paste from the "help!" files) to calculate 'area'. In "field calculator, advanced".)
Often, images have an associated "world" file that gives locatonal information about the image data. These often are named like the image but with a "w" in the extension, imagefile.tif and imagefile.tfw, for example. The contents of the world file are a list six numbers that tell where the image fits in a cartesian coordinate system. The six numbers define an affine transformation from one cartesian coordinate system to another. It looks like this (but without the annotations):
1.95 - raster cell x 0 - rotation paramter (defined as always 0 in ArcMap) 0 - rotation parameter (defined as always 0 in ArdMap) -1.95 - raster cell y 622097.18 - upper left x 2361213.32 - upper left yThe order is important. The last two numbers translate the image to the desired location in the coordinate system. The first and fourth scale the raster cells to the coordinate system. Some software uses the second and third numbers to rotate the coordinates, but ArcMap seems to assume that the two systems are not rotated from each other (north is north in both) and so, two 0s are commonly here.