Description:

Accurate representation of land surface states is essential to obtaining useful weather, hydrological, and climate model forecasts and generating good reanalyses. However, land surface models are subject to many errors, including from model parameters, mis-specified or missing physical processes and meteorological forcings or other boundary conditions. These shortcomings are addressed by land data assimilation systems (LDASs), which merge in-situ or satellite-based observations with estimates from land surface models to produce a statistically optimal analysis. 

Given on-going land surface model evolution and an ever increasing availability of high-quality observations, LDASs have the potential to support Earth System predictions over a wide range of timescales, directly using data to inform decision makers. However, there are still only a limited number of examples of land DA clearly improving forecast skills. This indicates that the community still needs to make considerable progress with coupled and off-line systems.

This session will highlight recent advances in land DA systems including: (i) the use of novel observations; (ii) algorithm development; (iii) methodological developments in bias correction and forward operators; (iv) incorporation into coupled land-atmosphere systems. In addition, results from the application of LDASs and their impacts on numerical weather prediction, water resources, food security, and ecosystem management, are also strongly encouraged, as well as illustrations of the work involved in transitioning LDASs from research into operations.