Calibration and uncertainty analysis for complex environmental models / Калибровка и анализ неопределенности для сложных моделей окружающей среды

PEST stands for “parameter estimation”. When it was originally written in 1994, this is all that PEST did. However over the 20 years that have elapsed since then, the capabilities of the PEST suite of software have expanded enormously. The emphasis has shifted from inversion to inversion-constrained model parameter and predictive uncertainty analysis.

The purpose of this document is to present, in one place, the theory that underpins PEST and that underpins the plethora of utility software that supports and complements PEST. In doing this, it serves the same purpose for the next generation of PEST-like software. This includes PEST++ and pyEMU. As such, it is hoped that this book provides a valuable resource for those who wish to understand inversion and inversion-constrained uncertainty analysis as it pertains to environmental models. These types of models include (but are not limited to) petroleum and geothermal reservoir models, groundwater models and surface water hydraulic and hydrologic models.

As well as presenting theory, an equally important role of this book is to draw some important conclusions from the theory. These conclusions pertain to real world model usage. Some of them question modelling practices that are commonplace today. Others suggest roles for models in environmental management beyond those which they presently play.

Although a basic knowledge of matrix algebra, and of statistical and geostatistical concepts, is required in order to follow the discussion provided herein, mathematical prerequisites are not high. Most readers will have acquired the knowledge to understand the following theory at school or though undergraduate university courses. The early part of the book reminds readers of a few basic linear analysis and statistical concepts that they may have forgotten. Refer to the many good books on these topics for further details <...>