Investigating satellite radiance Data Assimilation at different scales in an idealised convective model framework


Numerical Weather Prediction (NWP) models are based on the numerical integration of a system of differential equations which needs an initial condition to be solved. The atmosphere is a chaotic system characterised by limited predictability and high sensitivity to such initial conditions[1], meaning that the ability to predict its evolution in the future is crucially dependent on the accuracy with which the initial state is generated.
Data Assimilation (DA) schemes have been developed and applied in weather forecasting in order to produce the best possible initial conditions to run NWP models. These schemes combine model data with recent observations in order to reduce the gap between the predicted state of the atmosphere and the observed one[2], [3]. However, since DA schemes used for operational purposes are very computationally expensive, the use of idealised or simplified models is often considered as an alternative. This is also the case in my PhD.
In a previous study conducted by T. Kent at the University of Leeds[4], [5], a toy model based on a 1-D modified shallow water system of equations has been proved to be a suitable tool for DA research (we have recently submitted a paper about this, so watch this space![6]). The model is able to simulate convection and precipitation in a variety of weather conditions across several spatial scales. Once tuned, the DA algorithm (a Deterministic Ensemble Kalman Filter) led to features and behaviours in line with those usually seen in real NWP models, especially in terms of error growth rates and impact of the observations on the final analysis.

Purposes and aims

Satellite observations are a significant component of operational data assimilation schemes. Their efficient use to improve the performance of NWP models is still subject of research, with clouds being a key area of interest. In this project, we plan to further investigate and extend T. Kent’s DA scheme[4], [5] in order to test new configurations and support the satellite DA research at the Met Office, which is co-sponsoring the project. In particular, we will focus on how the assimilation of simulated radiance observations impact on NWP models at different spatial and temporal scales, since they are known to be affected by different error growths. The idea of a satellite observing periodically the earth will be replicated by means of a moving observational window that cycles in time across a periodic domain. Moving towards an observation operator able to assimilate simulated satellite radiance will be also a fundamental step in reducing the gap between our idealised configuration and real-word DA schemes.