Simultaneous Estimation of Soil Hydraulic and Thermal Properties Based on Multiobjective Optimization Algorithms

Simultaneous simulations of liquid water, water vapor, and heat transport are essential for modeling unsaturated hydrological processes, especially in semi-arid and arid regions. Modeling such coupled hydrothermal processes greatly depends on accurate estimations of soil hydraulic and thermal properties. However, many contributions for estimating these parameters using inversion methods use a single observation as the objective variable, e.g., soil water content is the most common. This study employ multiobjective algorithms to evaluate the worth of different observation types in simultaneous estimations of the soil hydraulic and thermal properties in Inner Mongolia, China. The coupled hydrothermal processes are quantified by HYDRUS-1D model, within which a multialgorithm, genetically adaptive multiobjective (AMALGAM) algorithm is employed to investigate four types of observations that may be available including soil water content, soil temperature, matrix potential, and heat flux in soil profiles. Different combinations of the four measurement types are considered as objectives, resulting single-, dual-, triple-, and quadruple-objective optimization schemes. The results demonstrate that incorporating additional observation types, such as soil water content and matrix potential, significantly improves the overall simulation accuracy of the coupled model. Particularly, the soil water movement is closely linked to the observation of water content, which plays a crucial role in the inversion process. While adding temperature or heat flux to the multi-objective optimization further refines the accuracy of inversion. Considering the cost-benefit ratio of different observation types, simultaneous measurement of water content and temperature is the most practical approach for the inversion since these two variables can be observed simultaneously by the same set of probes such as with a TDR.