Development of the Upgraded Tangent Linear and Adjoint of the Weather Research and Forecasting (WRF) Model

The authors propose a new technique for parallelizations of tangent linear and adjoint codes, which were applied in the redevelopment for the Weather Research and Forecasting (WRF) model with its Advanced Research WRF dynamic core using the automatic differentiation engine. The tangent linear and ad...

Full description

Saved in:
Bibliographic Details
Published inJournal of atmospheric and oceanic technology Vol. 30; no. 6; pp. 1180 - 1188
Main Authors Zhang, Xin, Huang, Xiang-Yu, Pan, Ning
Format Journal Article
LanguageEnglish
Published Boston American Meteorological Society 01.06.2013
Subjects
Online AccessGet full text

Cover

Loading…
More Information
Summary:The authors propose a new technique for parallelizations of tangent linear and adjoint codes, which were applied in the redevelopment for the Weather Research and Forecasting (WRF) model with its Advanced Research WRF dynamic core using the automatic differentiation engine. The tangent linear and adjoint codes of the WRF model (WRFPLUS) now have the following improvements: A complete check interface ensures that developers write accurate tangent linear and adjoint codes with ease and efficiency. A new technique based on the nature of duality that existed among message passing interface communication routines was adopted to parallelize the WRFPLUS model. The registry in the WRF model was extended to automatically generate the tangent linear and adjoint codes of the required communication operations. This approach dramatically speeds up the software development cycle of the parallel tangent linear and adjoint codes and leads to improved parallel efficiency. Module interfaces were constructed for coupling tangent linear and adjoint codes of the WRF model with applications such as four-dimensional variational data assimilation, forecast sensitivity to observation, and others.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
ISSN:0739-0572
1520-0426
DOI:10.1175/JTECH-D-12-00213.1