Download Postscript File

Automatic Differentiation as a Tool for Sensitivity Analysis of a Convective Storm in a 3-D Cloud Model

Seon Ki Park - Kelvin K. Droegemeier - Christian H. Bischof

This paper appears in the following reference:

Park, S.K., K.K. Droegemeier, and C.H. Bischof, 1996: Automatic differentiation as a tool for sensitivity analysis of a convective storm in a 3-D cloud model. Computational Differentiation: Techniques, Applications, and Tools, M. Berz, C. Bischof, G. Corliss, and A. Griewank, Eds., SIAM, Philadelphia, PA, 205-214.

Abstract:

The ADIFOR automatic differentiation tool is applied to a 3-D storm-scale meteorological model to generate a sensitivity-enhanced code capable of providing derivatives of all model output variables and related diagnostic (derived) parameters as a function of specified control parameters. The tangent linear approximation, applied to a deep convective storm by the first of its kind using a full-physics compressible model, is valid up to 50 min for a 1 % water vapor perturbations. The result is very encouraging considering the highly nonlinear and discontinuous properties of solutions. The ADIFOR-generated code has provided valuable sensitivity information on storm dynamics. Especially, it is very efficient and useful for investigating how a perturbation inserted at earlier time propagates through the model variables at later times. However, it is computationally very expensive to be applied to the variational data assimilation, especially for 3-D meteorological models, which potentially have a large number of input variables.

 

Keywords: tangent linear approximation, forward sensitivity, adjoint sensitivity, vartiational data assimilation, convective storm, 3-D cloud model, moist convection, supercell storm

1. Introduction
2. Automatic Differentiation - ADIFOR
3. Methodology - Sensitivity to Perturbations
4. 3-D Storm Model and Control Run
5. TLM Validation
6. Sensitivity Results
7. Discussion
8. References

About this document ...