Precision agriculture is often based on management strategies that require soil property information at high spatial and temporal resolution. The resulting demand for soil property data can be met by proximal soil sensing (PSS) techniques that are increasingly becoming affordable and being used to minimize the laborious and cost-intensive soil sampling and conventional laboratory analysis. Worldwide, this service is often provided by small-scale consultancy companies, which together with the lack of information at user-scale about production cost of soil maps and the uncertainty of soil properties estimations make farmers less aware about such services. In this review, we aim at providing a complete and up-to-date evaluation in costs and accuracy of using electromagnetic PSS techniques (electromagnetic induction (EMI), ground penetrating radar (GPR), time- and frequency-domain reflectometry (TDR, FDR), visible near and mid infrared spectroscopy (vis-NIR, MIR), x-ray fluorescence (XRF) and gamma-ray spectroscopy (γ-ray)) for predicting and mapping soil properties that are related to soil health and fertility indicators. The soil properties are grouped as hydrological, texture, acidity, carbon, nitrogen, nutrients, depth, bulk density, mineralogy and other property classes. From an extensive literature search, 209 research papers to evaluate the accuracies of soil property prediction across different landscapes were reviewed and of these, 115 research articles were selected for the purposes of the review. In each case, the normalized root mean square error (nRMSE) and standard deviation were computed to yield comparable results reflecting techniques accuracies and precision, thereby suggesting their suitability for mapping a specific soil property. The lack of information on costs in the literature motivated us to perform a market analysis through questionnaires directed to service companies, a unique aspect of this study. The preliminary results suggest that estimating soil properties with high accuracy can be obtained using point-based measurement methods such as XRF and vis-NIR-MIR, both in laboratory and on the field. The use of soil spectral libraries (SSL) to such aim (a method with low associated costs) gives also accurate estimations. However, they require scanning of soil samples with close contact to the sensor. Among on-the-go methods, EMI and γ-ray are popular and often used in combination for both the estimation of individual properties and delineation of management zones. The use of GPR for mapping soil indicators related to soil fertility is still sparse, however, some recent literature shows potential for rapid estimation of root zone depth, soil moisture and electrical conductivity. Around 35 PSS companies from 11 countries have been contacted to answer a questionnaire with questions about 1) PSS techniques they commonly use, 2) soil properties they estimate, 3) expected accuracies, and 4) costs associated with fieldwork, data analysis and reporting. Overall, maps of soil texture and organic matter were the most requested services with costs ranging from 100 to 1000 Euros per hectare, depending on the technique and country. Overall, the total cost of the service is distributed 40-45% to data analysis and reporting, 40-45% to fieldwork, and the rest 15-20% is related to the specific service.
