An efficient way to obtain soil texture data

Posted on April 27, 2023 by lembrechtsjonas

When we designed our large-scale citizen science project ‘CurieuzeNeuzen in de Tuin’, we soon realized we had a problem on our hands: if we wanted to get an accurate idea of soil temperature and especially soil moisture from our 5000 measurement locations, we needed to have data on local soil conditions, and at least the soil texture.

That was not entirely impossible, we thought, as we had a laser diffraction device in the lab to assess soil fractions that could be used. We just had to ask all citizens to take a soil sample and send it to us. The problem came to the surface when we calculated our time investments: most laser diffraction analyses – at least those we knew of – could handle four samples per hour. And that was at their best behaviour only, excluding errors, cleaning of lenses, replacing of components etc.

Sampling the soil around our ‘garden dagger’

Now, four samples an hour for 5000 samples gives a total of 1250 hours of soil texture analyses, which is 156 days of relentless work, or 8 months of continuous labwork for a lab technician; if not beaten to death halfway through by the extreme dullness of the task at hand.

A delivery to our university of 1276 boxes with soil samples, frankly hammering home the message that this was quite the amount of samples to deal with

It was obvious that we needed a better solution. That solution was found with our colleagues at the Earth and Life Institute of UCLouvain. They had a different way of measuring soil texture, using visible near-infrared spectroscopy. This technique was much faster – in a blink of an eye a soil sample could be scanned.

Laser diffraction involves boiling soil in acid, which is pretty cool at first, but can be rather time-consuming, especially for soils rich in organic material

Spectroscopy works fast, but the results need to be calibrated. For a subset of the data – say 10% – the traditional laser diffraction method still needs to be used, which can then provide you with a calibration curve to identify the fractions of clay, silt, and sand in your sample.

The spectroscopical analysis is as simple as putting a scanner on top of such a dried soil sample and processing the result

In a new paper together with the team at UCLouvain, we now provide a better calibration formula for this spectroscopical analysis, which takes into account one mathematical issue with texture data: the sum of clay, silt and sand is always 100%, so you need to model them together to avoid impossible soils with more (or less) than 100% of volume.

That issue has now been solved and published in the journal ‘Soil and tillage research’. Thanks to this fabulous method, our citizen participants also had to wait only four months to get information on their garden soil texture on their dashboard, a most remarkable achievement they are probably not even aware of.