Estimation of subsurface porosities and thermal conductivities of polygonal tundra by coupled inversion of electrical resistivity, temperature, and moisture content data

• Published in: The Cryosphere
• Main Authors: E. E. Jafarov, D. R. Harp, E. T. Coon, B. Dafflon, A. P. Tran, A. L. Atchley, Y. Lin, C. J. Wilson
• Format: Article in Journal/Newspaper
• Language: English
• Published: Copernicus Publications 2020
• Subjects: Environmental sciences; GE1-350; Geology; QE1-996.5; Arctic; permafrost; The Cryosphere; Tundra
• Online Access: https://doi.org/10.5194/tc-14-77-2020; https://doaj.org/article/42fb65b5f30346ea8705f49588a0ba7a

Studies indicate greenhouse gas emissions following permafrost thaw will amplify current rates of atmospheric warming, a process referred to as the permafrost carbon feedback. However, large uncertainties exist regarding the timing and magnitude of the permafrost carbon feedback, in part due to uncertainties associated with subsurface permafrost parameterization and structure. Development of robust parameter estimation methods for permafrost-rich soils is becoming urgent under accelerated warming of the Arctic. Improved parameterization of the subsurface properties in land system models would lead to improved predictions and a reduction of modeling uncertainty. In this work we set the groundwork for future parameter estimation (PE) studies by developing and evaluating a joint PE algorithm that estimates soil porosities and thermal conductivities from time series of soil temperature and moisture measurements and discrete in-time electrical resistivity measurements. The algorithm utilizes the Model-Independent Parameter Estimation and Uncertainty Analysis toolbox and coupled hydrological–thermal–geophysical modeling. We test the PE algorithm against synthetic data, providing a proof of concept for the approach. We use specified subsurface porosities and thermal conductivities and coupled models to set up a synthetic state, perturb the parameters, and then verify that our PE method is able to recover the parameters and synthetic state. To evaluate the accuracy and robustness of the approach we perform multiple tests for a perturbed set of initial starting parameter combinations. In addition, we varied types and quantities of data to better understand the optimal dataset needed to improve the PE method. The results of the PE tests suggest that using multiple types of data improve the overall robustness of the method. Our numerical experiments indicate that special care needs to be taken during the field experiment setup so that (1) the vertical distance between adjacent measurement sensors allows the signal ...