4. Discussion
The approach developed here has been shown to provide the capability to carry out rapid assessment of a range of soil physical and chemical parameters based solely on RGB or CIELab colour, either of which can be readily derived from Munsell observations (or colour-corrected digital photographs). This could be used to provide information on soil properties while in the field, using relatively simple observations and standard
computer equipment. The ability to carry out cheap and rapid soil assessments would be improved if more accurate evaluations of some parameters (e.g. P and K) were possible, which may be achievable with a larger training dataset. However, the parameters that can be accurately assessed (C, N and texture) would provide important information
regarding organic matter content and composition, and therefore allow characterisation of soil ecosystem service provision capability (e.g. carbon sequestration, water holding capacity, productivity). CIELab colour was found to be a slightly better overall predictor of soil characteristics than RGB, but this is balanced against the ease of use of RGB when dealing with digital image colour and visualisation using computer equipment. The benefits of CIELab over RGB are not therefore considered strong enough to argue for its use in deriving soil chemical composition instead of RGB.