Portal de Periódicos da CAPES

Approaches in highly parameterized inversion—PEST++ Version 3, a Parameter ESTimation and uncertainty analysis software suite optimized for large environmental models

2015; United States Geological Survey; Linguagem: Inglês

10.3133/tm7c12

2328-7047

David E. Welter, Jeremy T. White, Randall J. Hunt, J. E. Doherty,

Atmospheric and Environmental Gas Dynamics

First posted September 18, 2015 For additional information, contact: Wisconsin Water Science Center U.S. Geological Survey 8505 Research Way Middleton, Wisconsin 53562-3586 http://wi.water.usgs.gov/ The PEST++ Version 1 object-oriented parameter estimation code is here extended to Version 3 to incorporate additional algorithms and tools to further improve support for large and complex environmental modeling problems. PEST++ Version 3 includes the Gauss-Marquardt-Levenberg (GML) algorithm for nonlinear parameter estimation, Tikhonov regularization, integrated linear-based uncertainty quantification, options of integrated TCP/IP based parallel run management or external independent run management by use of a Version 2 update of the GENIE Version 1 software code, and utilities for global sensitivity analyses. The Version 3 code design is consistent with PEST++ Version 1 and continues to be designed to lower the barriers of entry for users as well as developers while providing efficient and optimized algorithms capable of accommodating large, highly parameterized inverse problems. As such, this effort continues the original focus of (1) implementing the most popular and powerful features of the PEST software suite in a fashion that is easy for novice or experienced modelers to use and (2) developing a software framework that is easy to extend. The PEST++ Version 3 software suite can be compiled for Microsoft Windows®4 and Linux®5 operating systems; the source code is available in a Microsoft Visual Studio®6 2013 solution; Linux Makefiles are also provided. PEST++ Version 3 continues to build a foundation for an open-source framework capable of producing robust and efficient parameter estimation tools for large environmental models.