Statistical approaches to assimilate ASCAT soil moisture information—I. Methodologies and first assessment

Land surfaces are characterised by strong heterogeneities of soil texture, orography, land cover, moisture, snow, and other variables. The complexity the surface properties is very challenging to represent accurately in radiative transfer models, which have a limited reliability over land, especially for observations such as given. This has resulted difficulties assimilating land-surface related satellite numerical weather prediction models. Simple statistical relationships between variables therefore been considered last 20 years. In this study, we propose compare two approaches: cumulative distribution function (CDF)-matching (used normalisation an inversion technique) neural network (NN) methods. CDF-matching finds simple monovariate relationship at local scale dependent on model (LSM) it calibrated. NNs global multivariate models able exploit auxiliary information synergy multiple instruments, but solution no characteristics constrain solution. One these methods will be better suited, depending application—in particular simplicity satellite/variable relationship. We illustrate concepts here using Advanced SCATerometer (ASCAT) moisture (SM) retrieval assimilation context. approaches compared combined. also more traditional scheme forward modelling, could attractive purposes. show that, context ASCAT, approach seems suited than modelling. that possible combine obtained NN localised LSM offered CDF-matching.