DEVELOPMENT OF A NEW METHODOLOGY FOR ADAPTIVE OBSERVATIONS

IN THE CONTEXT OF FOUR-DIMENSIONAL VARIATIONAL DATA ASSIMILATION

Collaborative research project between Portland State University & Florida State University

Funded by NASA - Modeling, Analysis and Prediction Program, Grant No. NNG06GC67G


RESEARCH TEAM RESEARCH OBJECTIVES: This project aims to develop a new methodology for adaptive (targeted) observations in the context of atmospheric four-dimensional variational (4D-Var) data assimilation. The research builds on current adjoint based methods such as singular vectors, sensitivity to analysis and to observations, to develop new targeting strategies that take into account the particular details of a 4D-Var data assimilation and forecast system and the interaction between time distributed adaptive observations and the routine observational network. These novel methods are expected to outperform objective targeting strategies with a low additional computational effort. The research is designed in stages of difficulty, starting from simpler to more complex models. The milestones of the project are:

  1. Implementation of targeted observations methods with a 2D global shallow-water model using adjoint sensitivity and singular vectors (first order adjoint-based methods). Assessment of the impact of the model error through the use of various numerical discretization schemes
  2. Development of a second-order adjoint model for the 2D shallow-water model, computation of the leading Hessian singular vectors (HSV), spectral analysis
  3. Implementation of targeted observations methods using HSV and sensitivity to observations in the 4D-Var context (second order adjoint-based methods). Assess the potential benefits of targeting methods using second order adjoint information.
  4. Development of targeted observations techniques with the NASA/GMAO finite-volume general circulation model. Carry out validation experiments using GEOS-5 DAS data.
  5. Investigate the potential use of targeting methods for "a priori" timely selection of which satellite observations are included into the data assimilation and forecast system
  6. Development of a state-of-the-art visualization package to facilitate analysis of 3D time-varying data and sensitivity fields.

PUBLICATIONS