Abstract Satellite remotely sensed (RS) soil moisture (SM) is commonly assimilated to leverage land surface modeling. However, the improvement in deep soil simulation remains challenging as limited by the relative shallow penetration depth (< 5 cm). With advancements in satellite RS technologies like P ‐band sensors, retrieving deeper SM has become increasingly feasible. Here, we demonstrate the potentially added value of enriched deep soil moisture information in land surface data assimilation (DA) by using ground‐based SM observations from a dense monitoring network in the central Tibetan Plateau. Specifically, a cost function‐based multi‐layer SM DA framework is developed to optimize soil texture profiles and organic matter content. A series of DA experiments were conducted to explore the optimal assimilation and optimization depths. The results suggest that assimilation of top 20 cm SM is adequate to reasonably optimize key soil parameters for all soil layers, and thereby improves SM estimates to the depth of 40 cm. This improvement can further propagate into soil temperature profiles and surface flux (e.g., evapotranspiration) estimates. Besides, the performance of deep SM DA is found insensitive to assimilation frequency varying up to 5 days, highlighting the promise and feasibility of regional land DA with 0–20 cm SM products, which are readily accessible from future satellite missions.
Zhou et al. (Wed,) studied this question.