Towards development of on-line soil moisture content sensor using a fibre-type NIR spectrophotometer

• Published in: Soil & tillage research 2005, Vol.80 (1), p.171-183
• Main Authors: Mouazen, Abdul Mounem, De Baerdemaeker, Josse, Ramon, Herman
• Format: Article
• Language: English
• Subjects: Agronomy. Soil science and plant productions; Biological and medical sciences; fiber-type near infrared spectrophotometers; Fundamental and applied biological sciences. Psychology; measurement; Moisture content; On-line measurement; online measurement; Physical properties; Physics, chemistry, biochemistry and biology of agricultural and forest soils; Sensor; sensors; Soil science; soil water content; Spectrophotometer; spectrophotometers; Water and solute dynamics
• ISSN: 0167-1987; 1879-3444
• DOI: 10.1016/j.still.2004.03.022

As the soil moisture content ( w) is a deterministic factor for site-specific irrigation, seeding, transplanting and compaction detection, an on-line measurement system will bring these applications into practice. A fibre-type visible (VIS) and near-infrared (NIR) spectrophotometer, with a light reflectance measurement range of 306.5–1710.9 nm was used to measure w during field operation. The spectrophotometer optical unit was attached to the subsoiler chisel backside to perform the light reflectance measurement from the soil surface on the bottom of the trench opened by the proceeded chisel. The spectrophotometer–optical unit system was calibrated for w under stationary laboratory conditions on samples collected from an Arenic Cambisol field with different soil textures. A partial least square analysis was carried out in order to establish a statistical model relating soil light spectra with the gravimetric w of the 0.005–0.26 kg kg −1 range. This model was validated with the full cross validation method resulting in a small root mean square error of cross validation (RMSECV) of 0.0175 kg kg −1 and a high validation correlation of 0.978. Further validation of the model developed in the laboratory under stationary state showed also a small root mean square error of prediction (RMSEP) of 0.0165 kg kg −1 and a prediction correlation of 0.982. When the NIR sensor-model system was used to determine w, based on on-line field measurement, a relatively larger RMSEP of 0.025 kg kg −1 and lower prediction correlation of 0.75 were found. However, a reasonably similar spatial distribution of w was found between the on-line NIR measurement and oven drying methods. Therefore, the on-line NIR w sensor developed is recommended to provide valuable information towards the site-specific applications in soils.