Improved farm soil mapping using near infrared reflection spectroscopy

Information on soil texture, soil organic matter content (SOM), nutrient status and pH is fundamental for efficient crop production and for minimising negative effects on the environment. Farmers obtain this information, on which decisions on fertiliser and lime requirements are based, through farm soil mapping. Although there is a general awareness that within-field and within-farm variations might not be adequately captured using conventional sample point density, simply increasing the number of sample pointes would increase the cost to unacceptable levels. In this thesis, near infrared reflection (NIR) spectroscopy was used to obtain more accurate information on within-field or within-farm variations in a number of soil properties. One central objective was to estimate the within-field variation in N mineralisation, to allow for improved N fertilisation strategies. Another was the development of economically feasible strategies for increasing sample point density in conventional farm soil mapping for improved decision support in precision agriculture. The results presented here show that NIR spectroscopy can be used to estimate N mineralisation (measured as plant N uptake) in fields with large variations in SOM, and that the additional predictive capacity of NIR compared with SOM is related to variations in soil texture. The results also demonstrate that it is possible to make small farm-scale calibrations with a very limited number of calibration samples for clay and SOM content, producing information at a considerably higher density than conventional farm soil mapping. Within-field calibrations for pH and easily available P, K and Mg-AL also proved possible, but more calibration samples were needed. Predictions for silt failed regardless of the number of calibration samples