D3.4 REPORT ON INVERSIONS FOR WIND FIELD DERIVATION/ IMPROVEMENT

Infrasound recordings in the 0.1 to 4 Hz band as input of inversion procedures to delineate the vertical structure of the wind in a range of altitude where ground based or satellite measurements are rare and where fine-scale atmospheric structures are not resolved by the current atmospheric specifications. As infrasound is measured worldwide, this allows for a remote sensing technique that can be applied globally. This report provides an overview of studies that have focused on the sensitivity of infrasound to the upper atmosphere. Long-term measurements of infrasound and associated wave parameters have been compared to state-of-the-art atmospheric specifications. Combined with propagation modeling, such information is useful in reducing uncertainty in knowledge of wind speed models in the upper atmosphere. Furthermore, we provide a summary of infrasound inversion techniques that have been developed recently. These techniques have been applied to real data sets in order to derive wind field updates. We also discuss work in which the inversion results are compared with measurements from independent techniques, such as Lidar and wind radiometer. In this work, it is found that the inversion results are in line with comparisons between Lidar and micro-wave radiometer observations and model output from the European Center for Medium-Range Weather Forecast (ECMWF).
Complementary data coupled with weather prediction – chemical transport models allow the derivation of wind fields from tracer observations in the altitude range of 10-30 km. As direct wind observations are sparse, this indirect approach can help to improve the data base from the medium to global scale considerably. For the ARISE project, an inversion scheme has been developed that can be implemented within a chemical assimilation system to provide daily initial wind fields. A three-step exercise: A coupled isentropic wind-advection model has been set up to demonstrate the feasibility of the approach, i.e., the inversion of the advection scheme only, which is much more efficient c.t. inversion of the NWP model. It was proven that the model is sufficiently accurate while computationally efficient. Second, the potential for chemical inversion was demonstrated by assessing the global potential to restore distorted initial wind fields in terms of forecast skill. Finally, the adjoint of the SACADA advection scheme was derived and implemented into the GME-SACADA data assimilation system to allow the processing of MIPAS trace gas observations and to test results with the GME weather forecast model. For the sample period of February 2003, we show the merits and limitations of our approach. Results are promising but more work is needed to verify results.