Satellite data assimilation and vertical localization

The quality of numerical weather prediction (NWP) forecasts strongly depends on the accuracy of initial conditions. Data assimilation systems provide the best estimate of the initial state, which combines the latest short-range forecast with millions of observations. Satellite observations are therefore of high value since they provide globally highly resolved measurements in space and time. Different sensors for wavelengths from infrared to visible light provide insight into the existence of clouds and the distribution of water vapor in the atmosphere and therefore improve our initial conditions leading to better forecast quality.


Vertical localization

Depending on atmospheric correlations, observations can have a large radius of influence. Satellite observations, in contrast to conventional point observations, are vertically integrated measurements. Thus satellite observations can be influenced by all atmospheric layers from the surface to the stratosphere. Therefore, it is crucial to understand correlations from satellite observations to the model field including sampling errors that arise when computing sample correlations. One efficient way to reduce sampling errors is to "localize" correlations. The goal of our research is to improve vertical localization for satellite observations.


Assimilation Experiments

To estimate the potential value of different satellite channels (i.e. wavelengths), we don’t already need an actual satellite in orbit. In an Observation System Simulation Experiment (OSSE), the NWP system as well as a digital twin of the real atmosphere is simulated from which satellite measurements are calculated using radiation transfer models. Hence, we can study the impact of prospective satellite platforms. Currently, we are designing a new OSSE (based on the WRF model and DART). The OSSE will be used to perform fundamental research in the field of satellite data assimilation.

MSG SEVIRI: Clouds over Europe (reflectance; 600nm / visible)