Text Box:

General overview of the Scheme

The EO Land Data Assimilation Scheme under development in the project contains three main components: The Process Model; the Observation Operator; and the Assimilation Method.

 

 

 

 

 

 

 

 

 

The Process Model describes the expected evolution of the state vector in time.  The complexity and physical basis of the forward model may vary considerably.  In NWP, this is complex, fully physical 3-D model of the atmosphere that provides a strong constraint in the assimilation system.  For land processes, the available models do not support a strong constraint on observations. In the case of LDAS  the scheme offers a simple LAI empirical model and a zero order model.

The Observation Operator predicts a set of observations given the complete set of state vectors at all points in the model space.  It consists of:

· The surface radiate simulator to predict the directional and spectral distribution of light scattered from the land surface. In EO-LDAS the 1-D semi discrete model is used in EO-LDAS.

· Radiative Transfer Functions to propagate the reflected signal to the direction satellite, and account for other atmospheric scattering

· Geometric and interpolation functions to map the simulated observations onto the EO observations.

The Assimilation Method.  In most cases this is based on techniques to minimize a suitable cost function.  EO-LDAS uses a constrained cost function minimization approach based on  4D-VAR techniques. 

 

For more details of the implementation of the scheme, please see the EO-LDAS Prototype section.