Data Assimilation

D. Nicolsky, V. Romanovsky, G. Panteleev

A variational data assimilation algorithm is developed to reconstruct thermal properties, porosity, and parametrization of the unfrozen water content for fully saturated soils. The algorithm is tested with simulated synthetic temperatures. The simulations are performed to determine the robustness and sensitivity of algorithm to estimate soil properties from in-situ high resolution-in-time temperature records in the active layer, and once-a-year measurements in a relatively deep borehole. The algorithm is applied to estimate soil properties at several sites along the Dalton Highway. The presented approach is quite general and can be applied to many problems requiring finding an optimal set of soil properties, and uncertainties in found values.