Soil moisture product merged from satellite and modelled data for Great Britain, April 2015-December 2017

A merged dataset blending two remote sensing (ASCAT and SMAP) and one modelled (JULES-CHESS land surface model) soil moisture datasets was produced under the NERC funded HydroJULES programme. The merging was done using Triple Collocation Analysis which minimises the errors from a statistical point of view, producing an improved product, continuous in time and space.

Evaluation using in-situ soil moisture measurements from the COSMOS-UK network shows that the merged soil moisture integrates the characteristics of model simulation and satellite observations and particularly improves the limited temporal variability of the JULES-CHESS simulation.

Soil moisture is of fundamental importance to many hydrological, biological biochemical and meteorological processes. Accurate information about soil moisture is therefore valuable to a wide range of stakeholders.

Despite this, widespread, continuous and reliable measurements of soil moisture are currently not widely available. Remote sensing satellite data provides regional soil moisture data, but the accuracy is variable and missing data frequent. Soil moisture models, on the other hand, provide continuous data both in time and space, but are indirect estimates based on numerical simulations rather than observations and do not always capture the temporal dynamic of drying and wetting accurately.