Optimal sampling and measurement to estimate soil organic carbon stocks

Human societies need to mitigate climate change. Sequestering atmospheric carbon dioxide in soil as soil organic carbon (SOC) may mitigate climate change. Researchers and policy makers interested in studying or incentivizing SOC sequestration must be able to precisely estimate the amount of SOC in a given plot of land. This estimation is typically accomplished by taking soil cores selected at random from the plot under study, mixing (compositing) some of them together, and measuring the composited samples. The costs of this process can be substantial, and there are uncertainties associated with both sampling and measurement. Given a fixed level of tolerable estimation uncertainty, there are optimal numbers of samples to be taken and measurements to be made that minimize the total cost of estimating the SOC stock. Conversely, given a fixed budget, there are optimal numbers of samples and measurements to minimize uncertainty. We formalize the sampling and measurement processes and derive these optima. We demonstrate the utility of this approach using data from soil surveys conducted in California and Colorado. Among our findings, we show that cheaper, less precise measurement methods may yield sharper estimates of SOC stocks under a fixed budget because the ability to make more measurements outweighs the additional error of individual measurements. We give recommendations for practice and provide software to implement our framework.