Data assimilation helps us make the most of our observations and our models by combining the information each of them carries in optimised ways.

Data assimilation is a key ingredient for environmental model forecasting. It helps us identify how to improve models of the Earth System by identifying mismatches between the data and the model predictions.

At NCEO we are developing cutting-edge techniques in data assimilation to help strengthen scientific capacity in the UK. We want to provide data assimilation systems that can be coupled to any core model to revolutionise the way data assimilation supports the UK geoscience community.

Staff involved in data assimilation work closely with others involved in satellite retrievals and in modelling. This integrated expertise allows NCEO researchers to develop next-generation data assimilation systems that apply to a wide range of data-model problems.

We work closely with researchers at operational centres like the Met Office and the European Centre for Medium-range Weather Forecasts (ECMWF) to improve on the approximations in their data-assimilation systems for future weather and climate modelling.

Our research priorities are:

• How to treat uncertainty in the data and the models.

• Assimilation into models of the entire Earth System.

• Identifying improvements to environment models, relating to the physical processes they represent, the parameterisations they use, and the parameters for initialising and driving the models.

• How to couple together systems with very different scales in time and space.

• Developing fully non-linear assimilation systems.

• Assessing the impact of novel observing systems.

• Setting up computing infrastructure to provide NERC scientists with access to NCEO’s data assimilation tools, as well as efficient access to models, EO data and driving data sets.

Our current work includes:

• Developing a data assimilation software package – called EMPIRE – that easily couples to any geoscience model of choice.

• Applying EMPIRE, coupled to the Met Office’s land-surface model ‘JULES’, to improve estimates of soil moisture in sub-Saharan Africa.

• Solving the ‘curse of dimensionality’ that limits nonlinear data assimilation methods to low-dimensional systems, due to their excessive cost. NCEO researchers have developed a fully nonlinear method that is efficient even in very high dimensional systems, using an Equivalent-Weights-Particle Filter. The method is much more robust than previous methods and paves the way for efficient data assimilation in real geophysical systems.

• Testing of NCEO’s new nonlinear assimilation method on the climate model HadCM3.

• Applying data assimilation methods to ‘Big Data’ datasets from multiple sensors, such as those produced by ESA’s Sentinel programme, to estimate land surface state in a consistent and efficient way.

• Evaluating different data assimilation methods to understand their limitations and provide guidance on improving them.