Brett Candy, Met Office
Land Surface Data Assimilation at the Met Office
Abstract: In recent years a number of advances have been made in Land Surface Data Assimilation at the Met Office. In 2013 the soil moisture nudging scheme was replaced with an Extended Kalman Filter that assimilates screen level observations and ASCAT satellite data to improve the initialization in the Global model. Observations are processed before being used to improve their representativeness. A screen only 3DVAR step is run to obtain a gridded screen analysis, thus preventing the system from creating “island” patterns at the observation locations. Satellite data is bias corrected to match the soil model climatology that has been calculated with a 30 year model run. A set of soil models with perturbed initial conditions is run to calculate the Jacobian between model and observed variables. Comparison against observed data shows that EKF provides a clear improvement over the previous nudging scheme.
Results from recent developments will be also presented. First, Global LST (Land Surface Temperature) is being tested to improve also the soil temperature representation in the model initial state. And second, the EKF algorithm used at the Global model is being tested on the high resolution UKV domain, which covers the British Isles. At the present time the soil moisture in this modelling setup is being initialized from a reconfiguration of Global model values as it currently does not have any Land Surface DA. These two developments potentially represent significant improvements in the initialization of the soil variables.