An adaptive sampling strategy developed for studies of diffuse volcanic soil gas emissions

Abstract

The spatial distribution of diffuse soil degassing from a volcanic source and quantifying of the amount of gas release are key aspects for understanding the gas release pattern. The measurement of the entirety of a particular area of interest cannot be possible due to obvious logistical and economic restrictions. To reduce the logistic and economic impacts, spatial interpolation techniques are generally used with limited observation, which is always associated with a high level of uncertainty in the quantification of diffuse soil degassing. Here we are developing an adaptive sampling strategy and robust analytical tool that can be used to guide measurement strategies by optimizing the sampling locations and reducing the uncertainty for gas emission estimation. Our adaptive sampling strategy hybridized two well-known algorithms: kriging interpolation and genetic algorithm optimizer, to optimize the locations of future sampling. We have validated our tool using a synthetic data set, and then applied it to two volcanic soil degassing data sets (Pacaya Volcano and Yellowstone Volcano). Both the synthetic data results and case study applications demonstrated that the developed adaptive sampling tool significantly outperforms random and pseudo-random approaches to reduce the uncertainty of gas quantification.