Add imagery to an ArcGIS Pro project
Explore historic imagery in ArcGIS Pro
Imagery is obtained from many sources, such as satellite, aerial cameras, and scanned historical aerial photos. Modern satellite images and aerial cameras tend to have relatively accurate location information, but scanned historical aerial photos usually do not contain spatial reference information. This means that you could not use them in a GIS system along with other geospatial data, as the software may be unable to correctly place them on the map.
There are different methods to remedy this issue. More advanced workflows involve orthorectification techniques, but they also assume that you have additional information about the images. In this lesson, you'll focus on a simpler georeferencing techniques, where you align the imagery to a map coordinate system.
In the lesson scenario, a commercial development company is required to complete an environmental impact assessment (EIA) before gaining permission to initiate a new project in the Aspen area. The EIA calls for an investigation of historic imagery to determine the original status of the river and vegetation in the project area. The historic imagery available needs to be georeferenced to complete the EIA. In this lesson, you will georeference a historic image, apply a simple transformation to the image, evaluate the result, and compare the initial transformation method to a more complex transformation method using additional control points.
First, you'll set up the project in ArcGIS Pro, and apply the required coordinate system.
- Go to the AspenGeoreferencing item details page.
- Click the Download button and specify an appropriate location for AspenGeoreferencing.ppkx to be saved.
The PPKX file includes a map document and a folder with the image that needs to be georeferenced. For comparison, an additional folder contains an examples of the same image with georeferencing applied that uses a more advanced transformation.
Depending on your web browser, you may have been prompted to choose the file's location before you began the download. Most browsers download to your computer's Downloads folder by default.
- Locate and double click the AspenGeoreferencing.ppkx file to launch ArcGIS Pro and open the AspenGeoreferencing project.
If you don't have ArcGIS Pro or an ArcGIS account, you can sign up for an ArcGIS free trial.
- In ArcGIS Pro, review the AspenGeoref map.
The map displays the World Imagery Hybrid basemap.
- In the map Contents pane, there are two layers, which collectively make up the World Imagery Hybrid basemap. Check each layer on and off to review what it represents on the map.
Note that the basemap displays both imagery and major roads, which will allow you to identify control points while georeferencing. The map is focused on the area of Aspen, Colorado.
Next, you'll add the historical aerial photo.
- In ArcGIS Pro, in the Catalog pane, expand Folders, AspenGeoreferencing, commondata, and raster_data2.
The Catalog pane is usually available when you create or open a project. Once closed, you can open it again by doing the following: Click the View tab on the ribbon, then in the Windows group, click Catalog Pane.
- In raster_data2, right-click AspenSource.tif and choose Add To Current Map.
The image is added to the map and is displayed as a layer in the contents pane, but since it is not georeferenced, it is not displayed on the map and the application warns you that the data source has an unknown coordinate system.
In reality, the image has been added to the map, but displays at latitude and longitude (0,0). Before you remedy this, you'll review the image on the map.
- In the Contents pane, right-click AspenSource.tif and choose Zoom to Layer.
The map now changes extent and displays the historical aerial photo taken in the Aspen area. However, since it is not georeferenced, the application cannot locate and position it in the Aspen area and thus displays it at latitude and longitude (0,0), as you can see indicated in the lower part of the map view.
Unreferenced images are added at the intersection of the Equator and the Greenwich meridian. To verify that current location, you'll zoom out until you can recognize geographic elements.
- Using the Explore tool, zoom out repeatedly with the mouse scroll wheel.
At first the background is completely black, but after a while some land appears, and you can see that you are off the west coast of Africa. This verifies that the unreferenced image was placed at latitude and longitude (0,0)
Next, you'll change extent to the Aspen area and proceed with the georeferencing steps.
- On the ribbon, on the Map tab, in the Navigate group, click Bookmarks, and select Aspen Colorado.
The map extent updates and now covers the same Aspen, Colorado area as represented in the unreferenced image. As a first step of the georeferencing process, you'll now set up the coordinate system in which you want to georeference the historic aerial photo.
All geospatial data and imagery must have a coordinate system or spatial reference defined. There are many different coordinates systems to choose from, but one that is commonly used for data located in Colorado is NAD 1983 UTM Zone 13N. This is the one you'll use in this lesson.
- In the Contents pane, click AspenSource.tif.
- On the ribbon, on the Imagery tab, in the Alignment group, click Georeference.
- On the ribbon, on the Georefence tab, in the Prepare group, click Set SRS.
SRS stands for Set Spatial Reference System.
The Map Properties and Coordinate Systems tab for the AspenGeoref map displays.
The AspenGeoref map default coordinate system is Web Mercator Auxiliary Sphere which is commonly used in web maps such as a basemap. Web Mercator Auxiliary Sphere it is not acceptable for this project, you'll switch to NAD 1983 UTM Zone 13N which is commonly used in Colorado.
- In the Coordinate System pane, type NAD 1983 UTM Zone 13N in the search box and press Enter.
- In the Coordinate System pane, expand the arrows next to Projected Coordinate System, UTM, and NAD 1983, until NAD 1983 UTM Zone 13N displays. Select NAD 1983 UTM Zone 13N and click OK.
The map now applies the NAD 1983 UTM Zone 13N coordinate system to layers in the map. Georeferencing the historic image will now take place in the same coordinate system and result in an image transformed to NAD 1983 UTM Zone 13N.
Next you'll fit the unreferenced image to the current extent covering the Aspen area, and position it manually to apply a rough alignment with the basemap imagery.
- In the Content pane, verify AspenSource.tif is selected.
- On the ribbon, on the Georeference tab, in the Prepare group, click Fit to Display.
The image is repositioned and is placed within the current map display.
On the map, the historic image is displayed ordered between the imagery basemap and the hybrid reference layer. The hybrid reference layer gives some useful information, such as the position of Highway 82 and other markers that can be used as control points . Next you'll explore the image to review its current position and scale..
- In the map, collapse the Georeferencing: AspenSource.tif information pane to simplify the map display.
- On the ribbon, on the Appearance tab, in the Effects group, click Swipe.
Using the Swipe tool to click and drag reveals content below the selected layer. This helps you evaluate the current image placement and scaling.
- In the Map, use the Swipe tool to review the historic image. Click and drag from top to bottom or side to side to reveal the content below the image. Try to distinguish features that can be used as reference points.
Notice that the image is oriented incorrectly. In particular, Highway 82 runs diagonally in the image. Next, you'll rotate the image.
- On the Map tab, in the Navigate group, click Explore to exit the swipe tool.
- On the ribbon, in the Georeference tab, in the Prepare group, click the Fixed Rotate drop-down arrow, and choose Rotate Right.
The image rotates 90 degrees clockwise.
You can now see that the highway runs in the same direction in the image (in white) and on the basemap.
- Investigate the image using the Explore tool to zoom and check image registration. In the Contents pane, you may also check the historical image on and off to be better review alignment with the basemap.
You'll notice that the image covers a large area of basemap, but represents a much smaller geographic area to the south east of the City of Aspen. You may need to scale the image as a result.
Use easily recognizable river bends to help you orient yourself more precisely.
- If necessary, use the Move and Scale tools to adjust and refine the image location until you are satisfied with the results.
As you use these tools, transparency is applied to the image to assist in identifying the correct position.
While using the move and scale tools, press the C key to momentarily revert back to pan and zoom as needed.
This process is not intended to achieve a perfect registration of the image, you are simply refining the approximate placement to make it easier to visually find matching features between the input image and the basemap.
The process of aligning an image with a reference layer is also called registration. As you move, scale, and rotate, you might notice the location registration is improving, but it is impossible to achieve the same level of registration across the whole image. As you fix registration in one area, it modifies registration in other areas. In the next step of the georeferencing process, you'll use control points and a transformation method to gain a better overall registration across the image.
- On the Map tab, in the Navigate group, click Explore to suspend Move and Scale.
Do not save this image until after the next section when you add control points.
Next you will add control points to the image and apply a transformation.
Create control points
Identify and add control points
When you georeference an image, you define its location using map coordinates and assign the coordinate system of the map frame to the image. You do this by adding control points that identify a location in the image and its corresponding location in the reference basemap. Georeferencing an image allows it to be viewed, queried, and analyzed with your other geographic data. In this section, you will identify and add several control points, and select and apply a transformation method to the image.
- On the Map tab, in the Navigate group, click Bookmarks, and choose Manage Bookmarks.
For this lesson, several suggested locations for control points have been identified and added as bookmarks. You may choose to ignore these and select you own.
- Position or dock the Bookmarks pane below the Catalog pane.
Positioning the Bookmarks pane below the Catalog pane allows easy access to bookmarks from the pane instead of the ribbon.
- On the Georeference tab, in the Adjust group, click and disable Auto Apply.
When, AutoApply is active, it automatically applies a transformation to the source layer and updates the display as control points are added, removed or modified.
- In the Bookmarks pane, click 1st Control Point, and click Zoom To.
The map extent updates to the location for the first control point.
- On the Georeference tab, in the Review group, click Control Point Table. Reposition the table below the Map view.
- On the Georeference tab, in the Adjust group, click Add Control Points.
To add a control point, first click a location on the image you are georeferencing (the source layer), then click the location on the target layer in the map (the reference data), which shows the same area on the ground.
- In the Map, identify and add a control point for the source layer, at a curve along Highway 82.
For a quick view of your underlying reference layer (in this case, the imagery basemap), press the L key to switch the transparency of your source raster on and off.
- Identify the location for the From Point at the same curve along Highway 82.
The result of your prior rough alignment may vary and thus the placement of Highway 82 may appear differently relative to the source image.
- Click to add the From Point on the Source.
- Review the AspenGeoref:AspenSource.tif control point table and notice that the X Map and Y Map values are in reported in meters using the NAD 1983 UTM Zone 13N coordinate system .
Your Source, Map and Residual values may differ as a result of difference in your control point placement.
- Using the 2nd Control Point bookmark, add a 2nd control points following the same workflow as the 1st control point.
Identify and add the target location on the image.
Identify and add the source location.
- Review the AspenGeoref:AspenSource.tif control point table.
- Use the 3rd, and 4th control point bookmarks as a guide and add 2 additional control points along Highway 82.
- On the Georeference tab, in the Adjust group, click Apply.
- Review the map after you have applied adjustment and if necessary refine the adjustment by adding additional control points.
Your display may differ depending on the number of points added and where you placed them.
As you add additional control points, the source raster is adjusted and transformation is applied. Depending on the number of control points added you can choose between several types of transformations. These include polynomial, spline, adjust, projective and similarity transformations, that affect the amount of shifting, rotating and warping applied to the raster.
- Review the control point table.
Your Source, Map and Residual values may differ from these and are affected by the number of points used and the location where they were added.
- In the control point table, verify the transformation method applied is 1st Order Polynomial (Affine).
The first-order polynomial transformation is commonly used to georeference an image. Use a first-order or affine transformation to shift, scale, and rotate a raster dataset. This transformation method ensures that straight lines on the raster dataset will remain mapped as straight lines in the output raster dataset. Squares and rectangles on the raster dataset may be changed to parallelograms of arbitrary scaling and angle orientation.
- In the control point table, click the drop down arrow right of 1st Order Polynomial (Affine). Note the various transformation types available in the list. The choice of a specific transformation types limited by the number of control points created.
- Review the control point table and note the individual residual values associated with each control point. Keep in mind that the RMS error (root mean square) is the sum of all the residuals and is a measure of how accurately the chosen transformation equation is able to fit all of the different control points to their specified location.
Once again your values may differ due to the Source and Map X and Y values added.
- In the Map, expand and review the on-screen image georeferencing details.
Note the Total RMS Errors values and keep in mind that your value may differ.
For each control point, the residual is the difference between where the from point displays in the map, as opposed to the actual location that was specified. The total error is computed by taking the root mean square (RMS) sum of all the residuals to compute the RMS error. The value describes how accurately the chosen transformation method is able to fit all of the different control points to their specified location. When the error is particularly large, you can remove and add control points to adjust the error. However, a large error may also mean that the chosen transformation method is unable to accurately fit the points to their desired location. However, if the points are all properly chosen and a large error is reported, it typically means a different transformation is required.
The forward residual shows you the error in the same units as the data frame spatial reference. The inverse residual shows you the error in the pixel units. The forward-inverse residual is a measure of how close your accuracy is, measured in pixels. Residuals closer to zero are considered more accurate.
- In the Map, zoom out to display the full image.
Keep in mind you may have more and differently located control points and as such, your display may differ.
When using the georeferencing tools, it is important to understand that the RMS error reported with the transformation refers to the ability of the current mathematical model to fit the existing control points but does not report the accuracy of the image. If your project requires an accuracy report, you must measure independent points (not used as control points).
- On the Georeference tab, in the Save group, click Save.
Georeferencing modifies the shape of the original pixels and performs a process of resampling of the image. Saving the image updates and maintains its georeferencing information in an associated auxiliary file. The auxiliary file contains transformation parameters for the image.
It is recommended that you use the Save button when georeferencing using control points, not the Save As New button, which creates a new version of the image. Using Save displays the original image in its georeferenced location any time it is loaded into ArcGIS. Save generates an *.aux.xml file with the transformation method and control points, which can be edited later if needed by adding more control points (or removing a bad control point).
If a new image file is required, use the Save As New option. This may be required if you need to provide the georeferenced image to a user of another software package, but note that using Save As New duplicates the total data volume. Further, if areas of poor registration are noticed and new control points are added in the future, another copy of the file must be created. For these reasons, if you're working only in ArcGIS, Save is recommended rather than Save As to write out a new file.
- On the map zoom to the Aspen Colorado area.
Review the transformed image and notice that for relatively flat areas, the first-order transformation provides adequate results. However, in areas that are noticeable hilly, a different transformation type will provide better results.
- Save the project.
In this section, you applied a first-order polynomial transformation to the image. In the next section, you will compare the results of a first-order polynomial transformation to the results of a spline transformation.
Compare transformation methods
Evaluate control points and transformation methods
Georeferencing is the process of transforming a scanned map or aerial photograph so it appears spatially correct in a map. By associating features on the image with real-world x,y coordinates, you can progressively warp the image so it fits to other spatial datasets. Depending on your ability to find control points, you may choose various transformation methods. For example, to use the spline transformation to fit an image to a terrain, there must be 10 or more control points distributed around the image that includes terrain extremes such as ridges and valleys. In this section, you will compare the results of a first-order polynomial transformation to a spline transformation.
- Close the AspenGeoref map.
Your first-order polynomial transformation completed in previous steps may differ based on the number and the location of control points selected. As a result, you will use an existing example of a first-order polynomial transformation with 10 control points for comparison with a spline transformation consisting of 32 control points. Both comparison images are in NAD 1983 UTM Zone 13N and are used in this step for comparison only.
Avoid using Web Mercator when georeferencing. Web Mercator is the de facto standard for web map applications and simplifies sharing across the web, however when using it for data transformation and editing in localised areas far away from the equator, it introduces pronounced deviations that greatly affect accuracy of results.
- In the Catalog pane, expand Maps, right-click FirstOrder, and click Open.
The FirstOrder map contains an example of a first-order polynomial applied to the same historic source image you georeferenced previously, the difference being that this image has 10 control points where you added the minimum of 4 control points.
- Right-click the Spline map and click Open.
This map displays the same historic source image, transformed using a spline transformation and has in excess of 30 control points applied. In this map, the image appears to be warped and transformed more than in the FirstOrder map since a spline transformation applies localised stretching within the area of a specific control point.
Note the alignment of the forested slope in the southeast (lower right) corner of the image. Compared to the base imagery, you see that this region is still not properly aligned with the orthorectified imagery in the basemap.
- In the map Contents pane, click AspenSpline.tif.
- On the ribbon, on the Imagery tab, in the Alignment group, click Georeference.
The map updates to display the image and current control points.
In the map, note that the red X appears at all of the control points in the image, but the green X for the to positions are not visible. This is because the mathematics of the spline transformation perfectly fit every control point to its desired location, so the from points are directly on top of the green to points. Note that this is true at all control points but does not ensure proper placement of portions of the imagery away from the control points.
The number of links used in the image determine the complexity of the transformation that can be used to transform the raster dataset to map coordinates. However, adding more links will not necessarily yield a better registration. It is recommended that you distribute the control points around the entire raster dataset rather than concentrating them in one area.
In general, the greater the distribution of accurate control points between the raster dataset and the target data, the better the alignment results, because you'll have more widely spaced points with which to georeference the raster dataset.
Keep in mind that your georeferenced data is only as accurate as the data to which it is aligned. To minimize errors, georeference to data that is at the highest accuracy and resolution available to meet your project needs.
- On the Georeference tab, in the Review group, click Control Point Table.
- If necessary, reposition the table below the map.
- Review the control point table and note that there are 32 points with associated Source X and Source Y values and XMap and YMap values.
- On the map, review the on-screen image georeferencing details.
Note the RMS Errors values.
Although the RMS error is an assessment of the transformation's ability to fit the desired control points, a low RMS error does not ensure an accurate registration. For example, the transformation may still contain significant errors if you have a poorly placed control point. The more control points of equal quality used, the more accurately the mathematical transformation can convert the input data to output coordinates. Typically, the Adjust and Spline transformation methods give an RMS of nearly zero; however, this does not mean that the image will be perfectly georeferenced.
- Close the control point table.
in addition, you may want to remove the spline control points from the map display. This can be done clicking Close Georeference on the Georeference tab, in the Close group.
- Dock and position the FirstOrder and Spline maps side by side for comparison.
Compare the alignment of both images in their respective maps to the basemap, paying special attention to mountainous slopes on the east and west of the images.
Next you will link the maps for simpler navigation.
- In the Catalog pane, expand Maps.
- Select both the FirstOrder and Spline maps by holding down the Ctrl key.
- On the ribbon, on the View tab, in the Link group, click Link Views, and choose Center And Scale.
Both maps are now linked. As you zoom in on one map, the second map updates to the same extent.
- In the FirstOrder map, pan and zoom to the upper left of the image and locate Highway 82. Review the image registration along the Highway 82. in this area.
As you changed focus and pan in the FirstOrder map, the Spline map updates to match and allows you to do a visual comparison of the difference in image registration between the maps.
The FirstOrder map image has a first-order polynomial applied. This is one of the first transformations available to you as you begin georeferencing the image. It requires no set amount of control points and allows for a consistent but diminishing movement of the image you are working on. However, the more control points you add, the less effective this transformation becomes. If adding for example six control points, your image will shift greatly and begin to line up with your reference map, but after that threshold (6), the shifts become minimal and the number of control points you have to add to gain the same effect rapidly increases, which leads to the use of the next transformation: Spline.
- In the FirstOrder map, pan and zoom to the bend in the river and road near the center of the image.
Note the difference in image transformation in this area. A Spline transformation benefits from the number of control points used and requires at least 10 control points before you can select it.
A spline transformation requires the larger number of control points because it stretches the image, essentially moving a part of your image exactly to the control point location. Therefore, if you want to use this transformation, you must place control points everywhere on your image. It is one of few transformations that can fit terrain and uses warping or rubber sheeting to move portions of the image in opposite directions. The advantage is terrain fitting, with the requirement for more tie points.
- On your own, in the FirstOrder map, identify and zoom to an additional location along a valley or slope. For comparison, select an area in the FirstOrder map that appears to have poorer image registration with the basemap.
How does image transformation differ in valleys and slopes. Does the spline image show a higher level of accuracy ?
- When you are finished exploring the maps, save the project.
When you're georeferencing, start with a first-order transformation, but depending on your ability to find ad-hoc control points, use higher level transformations such as spline, in which a minimum of 10 control points are required to enable the transformation. For reasonable accuracy, distribute the control points around the image and on terrain extremes. Use a first-order transformation if you want satisfactory, but not perfect, georeferencing and you don't have a lot of time. Use the spline transformation if you want a perfectly georeferenced image and have a lot of time. However, when georeferencing urban imagery and applying Spline, bending of linear features such as roads and angular features such as street blocks may occur.
The developer completing the EIA in the Aspen area can now use your georeferenced image for a comparative analysis of the historical vegetation and river flow to gain permission for the proposed new project.