| Abstract |
In this presentation, various activities to apply automatic differentiation method are presented: from carbon sequestration program
to climate modeling systems. With regard to the carbon sequestration program, the AD method was applied to accomplish a robust derivative
computational scheme for the Jacobian matrix with the TOUGH2 code developed by LBNL. The TOUGH2 code has been world-widely used for a
variety of subsurface fluid flow systems including geological carbon control, geothermal and nuclear waste storage systems, etc.
Significantly important role of AD method to enhance the capability of the inverse modeling code of the TOUGH2 model are expected to develop the
AD-based optimal fluid flow modeling schemes. Also, the work of the AD-based sensitivity analysis and parameter tuning schemes for a climate
modeling system is introduced especially with the sea-ice component. The result shows that the AD method provides accurate sensitivity data for
various sea-ice modeling codes. This work was extended to develop efficient parameter tuning algorithm to provide better matching between
simulation and observational data.
|
