Accounting for errors in SOC estimates introduced by proximal sensing methods by Bertin Takoutsing, 2019 CLIFF-GRADS Recipient
Bertin Takoutsing from Cameroon completed his CLIFF-GRADS research stay at ISRIC – World Soil Information, supervised by Dr. Gerard Heuvelink. He is in the final stages of his PhD at Wageningen University in The Netherlands.
Quantification of the uncertainty in DSM products is very important for policy decision makers and land users, as decisions based on inaccurate soil information can ultimately have extensive and profound impacts, and impair end-users’ decisions.
Land health projects are consistently included among the priorities in the assessment of carbon sequestration potentials and soil organic carbon (SOC) stocks. Most of these projects use proximal soil sensing (PSS) methods such as Mid-Infrared and Near-Infrared (MIR/NIR) spectroscopy to generate soil data used in digital soil mapping (DSM) processes. Though these PSS methods are cost-effective, and time-saving as compared to traditional wet chemistry techniques, errors and uncertainties that are propagated through PSS methods are often ignored or neglected. This may lead to inaccurate DSM model outputs and poor decisions by the end-users. There is need for further exploration of the DSM approaches that account for uncertainties in soil measurements to improve the reported accuracies of the final SOC estimates.
As a CLIFF-GRADS recipient and during my PhD stay at ISRIC – World Soil Information, Wageningen, The Netherlands, I was involved in the research project that focused on incorporating measurement errors in soil observations in the state-of-the-art DSM approaches used to map carbon potentials and soil organic carbon (SOC) stocks. This was also a capacity building opportunity to improve my scientific knowledge and skills on geostatistics, spatial analysis and digital soil mapping that enable me to enhance my PhD research outputs both in content and quality.
As main outcomes, we were able to quantify the measurement errors in soil observations generated using conventional laboratory methods and PSS, and analyze how these propagate though the covariance structure of the spatial model to affect the reported accuracies of SOC stocks estimates.
I’m really grateful for the CLIFF-GRADS Programme that enabled me to get an insight of soil organic carbon stocks estimation through digital soil mapping and develop methods to analyze the propagation of measurement errors to the final estimates. On a more personal note, it was a fabulous opportunity to interact with other scientists, exchange experiences, develop new skills and expand the professional network which I am now using to foster my scientific career.