Soil profile image analysis: estimating soil properties from digital photography

WorkshopPedometricsImage1.png

The workshop will bring together researchers with practical experience of soil profile characterisation, and researchers with image analysis experience. The goals of the workshop are (1) to identify what is possible in terms of using automated image analysis to characterise soil profiles, and (2) to demonstrate the potential of this research area in a practical setting. Image analysis approaches will be described and then tested on images of soil profiles, and the practicalities of field-based soil profile image analysis discussed. Future work will also be discussed and linkages formed.

A worked example session will be used to demonstrate image analysis approaches applied to sample soil profile images. Software for analysing the sample images will be developed by Matt Aitkenhead prior to the workshop, and will include a range of image analysis methods/metrics that can be applied in real-time. The worked examples will be informed by discussion during the earlier session, in which the sample images are discussed and visible features identified and described. This will allow us to demonstrate the detection/measurement of profile features from automated image analysis.

PRELIMINARY TIMETABLE

  • Introductory presentation (Matt Aitkenhead)
  • Image analysis metrics
  • What can the eye do – and how could we use this?
  • Session on naked-eye observation of soil profile images (led by Alfred Hartemink)
  • Practicalities and protocols
  • Worked example session, with sample images analysed in real-time (led by Matt Aitkenhead)
  • Potential linkages to soil survey datasets
  • Closing discussion session (Matt Aitkenhead & Alfred Hartemink)

Each presentation will be 20 minutes, with 10 minutes for questions/discussion. The ‘naked eye observation’ and ‘worked examples’ sessions will last for ~2 hours each. 

REQUIREMENTS

Participants should bring their own laptops with R installed.


Lecturers

Alfred_Hartemink.png

Alfred Hartemink is Chair and Professor of Soil Science at the University of Wisconsin – Madison. He teaches Pedology and Introduction Soil Science. His research focusses on digital soil morphometrics, soil mapping, and soil C.

 

 

Dr Matthew Aitkenhead is a soil scientist who currently works for the James Hutton Institute. Much of his work has involved linking remote sensing and field-based information with existing maps, to produce models of soil character in a rapid and cost-effective manner. Recent work has involved linking these models to mobile phone technology, to produce apps that enable researchers and land managers to monitor the environment.  The growth of national and international legislation requiring large-scale and repeatable environmental monitoring is driving the development of rapid assessment tools for soils, and he is working heavily in this area. He is currently an Associate Editor for three journals: Remote Sensing Letters, Advances in Artificial Neural Systems and Soil Use and Management. He also leads the Scottish Regional Group of the British Society of Soil Science.