Evaluation of non-intrusive sensors for measuring soil physical properties


Cereals & Oilseeds
Project code:
01 October 1999 - 30 September 2001
AHDB Cereals & Oilseeds.
AHDB sector cost:
£104,337 from HGCA (project No. 2243).
Project leader:
J A KING1 , P M R DAMPNEY1 , R M LARK2 , H C WHEELER2 , R I BRADLEY3 , T MAYR3 , and N RUSSILL4 1 ADAS Ltd., Boxworth Research Centre, Boxworth, Cambridgeshire, CB3 8NN 2 Silsoe Research Institute, Wrest Park, Silsoe, Bedfordshire, MK45 4HS 3 National Soil Resources Institute, Cranfield University, Silsoe, Bedfordshire, MK45 4DT 4 TERRADAT Ltd., Ocean House, Hunter Street, Cardiff Bay, CF10 5FR



About this project


Knowledge of soil physical properties has always been important for decisions concerning cropping and crop management inputs, especially the use of fertilisers and lime. Information on the geographical distribution of soils is important if precision farming methods are to be used. HGCA project 2243 has investigated the practical usefulness and applicability of three soil sensor technologies for measuring and mapping soil types within fields, namely Electro-Magnetic Induction (EMI) which measures the apparent electrical conductivity of the soil (ECa), Ground Penetrating Radar (GPR) and Spectral Reflectance. The work was carried out in 1999-2002 on 4 experimental sites in England on contrasting soil landscapes (limestone, glacio-fluvial outwash, chalk, river terrace). Emphasis was given to studies on EMI.

The EMI sensor was housed in a metal-free cart drawn by an ATV at 10-15kph, with a Geographical Positioning System (GPS). It was simple and easy to use, producing a single data value at each measurement point which was able to identify soil texture variations especially where there was interaction between texture and soil wetness. Although ECa was overwhelmingly influenced by soil moisture, the data distinguished heavier less permeable soils from those that were more permeable and free draining. Regression analysis showed that subsoil clay and organic matter contents, and topsoil sand and organic matter contents were the main factors influencing the ECa; topsoil bulk density was also important. Since soil moisture had such a strong impact on the ECa, direct predictive relationships between ECa and soil properties could not be derived. This means that some in-field examination of soils will always be needed following an EMI survey to confirm the nature of the soils present in different zones. EMI data was more closely related to topsoil properties than output data from the cluster analysis of sequences of yield maps (HGCA project 2116). This suggests that the EMI sensor is reacting to the properties of the upper layers of the soil whereas the cluster analysis approach will be reacting to the soil conditions experienced by the whole crop root system, commonly well over 1m deep. The pattern of ECa variation was remarkably stable irrespective of whether the soil was wet or dry at the time of sensing. Using geo-statistical analysis, a between-pass spacing of c.20 m usually achieved an error of less than 25%; this is considered acceptable as a cost-effective and practical approach.

GPR proved to be a slower technique producing data that was difficult to interpret. Readings could not be obtained on clay soils but information on the depth to bedrock or free-water interfaces was obtained on sandy material. Spectral reflectance measurements of the bare soil surface did not have any clear or reliable relationships with topsoil properties. Neither of these techniques are considered to have any short-term potential as a practical or cost-effective method for agricultural soil sensing.

The project has shown that EMI is a reliable method for obtaining information on within-field soil patterns. Future work should develop an integrated use of EMI with other precision farming techniques for gathering information to allow improved crop management decisions both within and between fields.