Going Digital: Air Photo Interpretation for Geotechnical Investigations

Air Photo Interpretation (API) is a critical component of understanding sites yet it is my observation that this is seldom done in many of the residential geotechnical reports I get to see in my role within the Tasmanian Government.

The availability of aerial photographs has never been easier as Land Tasmania (and presumably other jurisdictions) has recently completed a major project by scanning their entire collection dating back to the 1940s. These photos can be searched online (from theLIST website) and able to be delivered in digital form to one's desktop. Until recently this used to incur a fee of around $50 per image but recently (2022) these photos are now free. There are probably several reasons why API is not being done (I will avoid the politics) but I am going to provide a low cost and efficient solution for working with digital images in stereovision that may represent one of the greatest obstacle for practitioners.

The traditional method of viewing aerial photographs was to obtain photographic prints and to use a stereoscope. So having ordered some high quality photographs (in digital form) from the supplier you will discover that most office printers/print bureaus are unable able to produce prints of acceptable quality and at an affordable price. Even if you did obtain quality prints it is near impossible to source acceptable tracing paper to draw your interpretation on! We can thank the digital revolution for this situation! Given we are being forced into the digital world whether we like it or not, it has led me to explore computer-based solutions.

One of the most useful resources available is Google Earth, but it is limited in that is only dates back to c. 2003 (missing 60 years of history) and does not provide a stereoscopic 3D capability with which to view subtle morphological features. After some searching and experimentation, I have discovered a freeware application called StereoPhoto Maker Pro (SPM) that, in my opinion, is the best solution available at this moment in time.

SPM will run on any modern computer and can load and automatically (or manually) align stereo pairs within seconds to allow you to pan and zoom over the area of overlap. I tend to view these images in anaglyph mode using inexpensive glasses (e.g. red/green) but side-by-side is also available. The quality of the stereo 3D experience is as good if not better than the traditional method. However the one major limitation of SPM is that it does not allow freehand drawing of features of interest as it has not been developed for this purpose. My workaround is to georeference one of the photos into a GIS environment (if needed) and, with a dual monitor setup, manually transfer the features observed in stereo directly into the GIS. With a bit of practice and without any need for proprietary software this produces an acceptable result at the level of accuracy one would normal require for desktop geotechnical purposes (e.g. decimetre to metre is easily achievable).

SPM comes with sufficient documentation and various YouTube resources to quickly get you going. However, there are a couple of issues you may encounter in the workflow.

1. SPM has an image size limitation and will provide an error message when attempting to load photos that are too big *(modern colour photos are huge, compared with scanned old B&W prints). I am not sure what the size limit is at present, but I resample them with an image editor, such as Irfanview (my favourite). I tend to reduce the images to 25%, the quality remains acceptable and SPM is happy! You are welcome to experiment.  Irfanview has a batch processing capability that will allow efficient downsampling of many photographs if necessary. [*Note: the latest version SPM  as at Jan 2022 is 64 bit and you may not need to resample as much. Best to try loading first and if it exceeds the size limit, then resample]
2. I use the Georeference Plugin in QGIS to bring the photos into a GIS environment. It took me a little time to practice working with this tool but I am getting better and for small areas in a photo, a surprisingly good result can be obtained without the need for expensive photogrammetry software.

 I have provided some screen grabs below to help you along the way:

Irfanview settings for Batch conversion can be accessed from the File menu. The Set for all images  can be accessed from the Advanced option on the Batch conversion menu. I have set 25% for the Resize value but you are welcome to experiment. The Batch conversion panel allows one to import multiple images and the program remembers the settings for next time you need to run this utility.

Here is an example image below of the SPM programme in greyscale anaglyph mode. After loading these photos using the second button on the left (open Left Right Images) I used the Auto button to align them. Most of the buttons are relatively intuitive but try viewing some YouTube videos to get an idea of additional functionality.  The photos I used here (from LINZ), while big, did not need resampling. The image was able to be zoomed up to 200x that was sufficient in this case to clearly discern individual houses.

So this is my solution to the API problem. Please let me know if you have a better solution, have questions or need some tuition.