This brings together in one place all of the theory on which PEST and its suite of utility software is based. It also covers important issues like how models should be used in the decision-making context, uncertainty analysis, model-based hypothesis testing, and the effects of model defects on calibration, prediction and uncertainty analysis. This book makes an essential reference text for those engaged in environmental modelling. Find out more about the book.
A new version of PEST named PEST_HP is now available from this site.
“HP” stands for “highly parallelized” or “high performance” (you can choose). This version of PEST is optimised for use in contexts where a modeller has many computing nodes at his/her disposal – either though an office network, a cluster, or (as is increasingly common) through using the cloud. Evidence to date is that it out-performs BEOPEST significantly – both in speed and in tolerance of model numerical performance difficulties. Its superior performance is based on making maximum use of the computing resources at its disposal in more aggressive searches for parameter upgrades, and through other improvements to its inversion algorithm. It also includes new functionality that assists calibration of difficult, nonlinear, models with long run times.
If you use PEST.cloud, then you use PEST_HP automatically. Alternatively, it can be purchased as a stand-alone item for your own cloud, network or cluster use.
PEST.cloud is a PEST cloud enabler. It gives a modeller ridiculously easy access to the cloud computing environment for model calibration using PEST_HP, without having to know anything but the barest essentials about cloud computing. Uploading/downloading of PEST and model files, distribution of files to computing nodes, and shutting down of those nodes when the inversion process is finished, is all taken care of. Meanwhile the modeller gets to use PEST_HP for efficient inversion, and has access to run-time statistics generated by PEST_HP while calibration is in progress. On completion of the inversion process, a brand new “PEST whisperer” can inspect PEST output files and offer suggestions that may help a modeller diagnose problems with PEST setup, and/or get the most out of using PEST to calibrate his/her model.
Courses are planned for the US, UK and Italy. See the training page.
At last, documentation for PEST and its Utilities has been completely re-written!
The new PEST manual is detailed and comprehensive. It now comes in two parts.
Part I describes PEST itself (including Parallel PEST and BEOPEST), as well as the two PEST-compatible global optimisers CMAES_P and SCEUA_P. Part II documents over 100 utility support programs whose tasks include the following:
Here are some downloads that can help you quickly learn more about PEST, its philosophical underpinnings, and implications of PEST technology for model-based decision support:-
See the downloads page for much more.
These pages were last updated on 24th November, 2017. (Version 14.2 of PEST; notification of PEST course in Washington DC area, Sep 2018.)
C4SF Login |
Web Site Managed via C4Site Factory,™ a trademark of
Echo Valley Graphics, Inc.
©2018 Copyright Echo Valley Graphics, Inc. All Rights Reserved.