Towards soil property retrieval from space: A one-dimensional twin-experiment

Soil moisture is among the key environmental variables controlling evaporation, infiltration and runoff. However, the temporal evolution of soil moisture is not easy to measure or monitor at large scales due to its spatial variability, which is largely driven by local variation in soil properties and vegetation cover. Consequently, soil moisture estimates using land surface models are typically made using effective physical parameterisations based on low-resolution and/or erroneous soil property information. Thus, land surface models have an urgent need for more accurate and detailed soil parameter data sets than are currently available, in order to undertake regional or global simulation studies at high spatial resolution and with the required accuracy. To overcome this limitation, the possibility of estimating the soil hydraulic properties through model calibration to remotely sensed near-surface soil moisture observation is explored. The study presents a methodology that demonstrates this potential using a synthetic twin experiment framework, thus avoiding the need to deal with possible model-observation biases. Moreover, it explores a range of scenarios, with the objective to determine the best meteorologic conditions for soil property retrieval and hence the most efficient use of computational resources when applying the methodology at large scales. These scenarios include: (a) short dry-down period, (b) short dry period, (c) short wet-up period, (d) short wet period and (e) full 12-months with multiple wetting and drying periods. The methodology was also tested for four different soil types including a homogeneous column of sand, a homogeneous column of clay, a duplex column of clay over sand, and a duplex column of silty sand over clay. The study showed that soil hydraulic parameters were best retrieved when using the full 12-month period, with the sequential retrieval of three parameters at a time being the most suitable approach when retrieving the six parameters, with the most sensitive parameters retrieved first. 2013 Elsevier B.V. All rights reserved.