Skip main navigation

Georeferencing Corona imagery in QGIS

A practical demonstration of how to work with Corona imagery in QGIS.

Now that we have downloaded our Corona image, we need to add it to QGIS!

Extracting our Corona imagery

Before we can work with our imagery, we need to get it to the right place and extract it.

  • 1) Move your Corona download into the folder you are using for this course – it might be a good idea to make a “Corona” folder for it.

Corona file screenshot Our downloaded Corona image.

Now we need to extract it.

This is an identical process to what we did with the Landsat imagery in Week 3.5.
  • 2) Right-click the image and select “7-Zip” > “Extract Here”.
  • 3) Repeat this with the new “tar” file.
This time you will see four new tif image files!
Extracted Corona screenshot The four extracted tif images.
Corona films are very long and thin, so they split the scanned images into four overlapping sections. It is possible to stick them all back together, but often we are only interested in part of the image, so we can skip this step and only work with one of the four sections.
  • 4) Double-click each image in turn to select which one you want to work with in QGIS.
  • 5) Your image may be upside down, if it is, you can rotate it by right-clicking it and selecting “rotate right”, then repeating this again.
Rotate image screenshot You can rotate your image if you need to.
We are now ready to begin working with our image in QGIS!

Georeferencing in QGIS

Unlike the modern satellite imagery, Corona digital imagery is created by scanning photographic film, so it has no built-in location information and does not ‘know’ where it belongs on Earth. We need to give it this information – this is called “georeferencing”.
We do this by comparing the old Corona image with a modern satellite image and finding shared points that match. These are called “ground control points” – the more of these we can provide, the more accurate the georeferencing will be.

Adding a modern satellite image and starting the georeferencer

We will need a modern satellite image that covers our area of interest – so before you start make sure you have a Sentinel-2 or Landsat image to work with.
You can look back at Week 2 or Week 3 if you need a reminder on how to do this.
  • 1) Start QGIS – you can either continue with a previous project or create a new one.
  • 2) Add the modern satellite image of your area using “Layer” > “Add Layer” > “Add Raster Layer”.
Reference image screenshot You need to add some modern satellite imagery of your area to georeference the Corona.
  • 3) Open the georeferencer by navigating to “Raster” > “Georeferencer…” on the main menu.
  • 4) In the georeferencer window navigate to “File” > “Open Raster…” and open your selected Corona image.
Your section of Corona imagery will appear in the georeferencer window – now we can start georeferencing!
Georeferencer screenshot The georeferencer with a section of Corona imagery added.

Setting up the georeferencer

Before we can start finding and adding ground control points (GCPs) we need to setup the georeferencer.
  • 1) Navigate to “Settings” > “Transformation Settings”.
  • 2) Change “Transformation type” to “Polynomial 1”.
  • 3) Make sure “Target SRS” is set to “EPSG:4326 – WGS84”, this should be the default.
  • 4) In “Output raster” replace “modified” with “GR”– this just makes it clear that the new Corona image we produced has been georeferenced.
  • 5) Change “Compression” to “LZW” – this will make our final file as small as possible.
  • 6) Tick “Save GCP points” – this helps backup our hard work in case anything goes wrong.
  • 7) Tick “Load in QGIS when done” – this just means that we won’t have to add our new image manually at the end of the process!
  • 8) Click OK to return to the georeferencer.
Transformation settings screenshot Adjusting the transformation settings.

Adding GCPs

Now we can start adding our GCPs – ground control points.
  • 1) Zoom in to the top-left of your Corona image and try to find a distinctive geographical feature.
  • 2) Switch to the main QGIS window and try to locate this feature on your Sentinel-2 or Landsat image.
This can be a very tricky process! You can use natural or anthropomorphic features, but they must be present in both images. This is harder than it sounds as roads, buildings and field boundaries often change over time, and we are often looking at time-difference of 50 years or more; even rivers move a surprising amount in that time! It may take a few attempts to find a matching point that you are confident in.
  • 3) Once you are happy with your point, switch back to the georeferencer and click the “Add Point” button.
  • 4) Click on your feature on your Corona image.
  • 5) Select “From Map Canvas” in the new window and click the feature on the main QGIS window.
  • 6) Click “OK”.
Add GCP screenshot Adding our first GCP.
A red dot should appear on your Corona image and your modern satellite image – this is your first GCP!
First GCP screenshot Our red GCP is present on both of our images.
  • 7) Repeat this to create four GCPs in total – one in each corner of your Corona image.
Four GCPs screenshot We now have four GCPs in each corner of our Corona image.

Adding more GCPs

To make the georeferencing as accurate as possible we will need at least fifty GCPs in total. But don’t worry! The process now becomes easier. QGIS now has enough information to roughly guess where the Corona image should go on the map.
  • 1) Select “View” > “Link Georeferencer to QGIS” and “Link QGIS to Georeferencer”.
Now when you move around on the Corona image, QGIS will follow approximately on the satellite image! The more GCPs you add, the more accurate this will become, which means the process should get easier as you go on.
You need to add fifty good GCPs for a section of Corona this size. They should be well spread out across the entire image, almost like a rough grid.
Fifty GCPs screenshot Around fifty, well-spaced GCPs have been added to the georeferencer.

Improving the georeferencing

You will notice that as well red points, red lines will also start appearing as you add GCPs. This shows the difference between where you have placed the GCP and where QGIS thinks it should be! The length of the lines will vary naturally, but if any are much longer, check that you haven’t made a mistake when placing it. You may have to move or delete and redo any problematic points with a large level of error. You can do this by using the “Move GCP Point” and “Delete Point” buttons.
GCP error screenshot One point is clearly incorrect and should be deleted and redone, you can see the size of the error for each point in the “Residual (pixels)” column of the GCP table.
At the bottom of the georeferencer it tells us the mean error using the current transformation. To locate our Corona image as accurately as we can, this number should be as small as possible. This is achieved by changing the mathematical model QGIS uses to warp the image to fit the GCPs.
  • 1) Make a note of the current mean error.
  • 2) Navigate to “Settings” > “Transformation Settings”.
  • 3) Change “Polynomial 1” to “Polynomial 2”.
  • 4) Check the new mean error – it should have drastically reduced.
By changing the transformation, we are now allowing QGIS to add a curved warp, rather than just stretching and skewing the image.
  • 5) Try changing to “Polynomial 3” and see how this changes the mean error.
This allows two curves to be added which can make the georeferencing more accurate. However, if there is not a big improvement, it is better to stick to “Polynomial 2”. When we use a higher polynomial, we are distorting the image. It is best to distort the image as little as possible because it makes it easier to interpret, so we should always go for the lowest polynomial that has the accuracy we want.
Reduced error screenshot In this example the error has dropped to a quarter of the size using a more complex transformation. Visually it is apparent that the red lines are much shorter!

Running the georeferencer

Once we are happy with our points and our transformation model, we can run the georeferencer!
  • 1) Navigate to “File” > “Start Georeferencing”.
This may take a few minutes to complete; your Corona imagery should be added automatically to your map.
Added Corona image screenshot The georeferenced Corona image has been added to the map.
  • 2) Zoom in and check you are happy with the results
Do be aware that because of how the imagery was taken, the georeferencing will never be perfect. If nowhere on the image is more than 100m from where it should be, this is a great result!
  • 3) Once you are happy you can close the georeferencer.
This will remove the red GCPs from the map.
After georeferencing screenshot The red GCPs disappear after closing the georeferencer.
Now you can start looking for archaeology!
Corona archaeology screenshot A falaj/qanat is visible in this Corona image from northern Oman.
How does the Corona compare to Landsat or Sentinel-2 imagery in your area of interest?
This article is from the free online

Advanced Archaeological Remote Sensing: Site Prospection, Landscape Archaeology and Heritage Protection in the Middle East and North Africa

Created by
FutureLearn - Learning For Life

Reach your personal and professional goals

Unlock access to hundreds of expert online courses and degrees from top universities and educators to gain accredited qualifications and professional CV-building certificates.

Join over 18 million learners to launch, switch or build upon your career, all at your own pace, across a wide range of topic areas.

Start Learning now