Optimizating SWAT performance using a Python tool

Carla Camargos*, Tobias Houska, Lutz BreuerCamargos.S.Carla@umwelt.uni-giessen.de

Justus-Liebig University Giessen, Germany

This project has received funding from the European Union’s Seventh Framework Programme for research, technologicaldevelopment and demonstration under grant agreement no 607000.

Motivation

9/19/2018 2018 SWAT International Conference: Optimization of SWAT performance using a Python tool 2

Sensitivity Analysis

Calibration

Uncertainty Analysis

Validation

Artificial Bee Colony (ABC)

Differential evolution Markov chain (DE-MC_Z)

DiffeRential Evolution Adaptive Metropolis (DREAM)

Fourier Amplitude Sensitivity Test (FAST)

Fitness Scaled Chaotic Artificial Bee Colony (FSCABC)

Latin hypercube sampling (LHS)

Monte Carlo (MC) Markov Chain Monte Carlo (MCMC)

Maximum Likelihood Estimation (MLE)

Robust Parameter Estimation (ROPE)

Simulated Annealing (SA)

Shuffled Complex Evolution Algortithm (SCE-UA)

9/19/2018 2018 SWAT International Conference: Optimization of SWAT performance using a Python tool 3

http://fb09-pasig.umwelt.uni-giessen.de/spotpy/

• Free

• Open source

• Parallel processing

SPOTPY + SWAT

9/19/2018 2018 SWAT International Conference: Optimization of SWAT performance using a Python tool 4

SPOTPY + SWAT

9/19/2018 2018 SWAT International Conference: Optimization of SWAT performance using a Python tool 5

SPOTPY + SWAT

9/19/2018 2018 SWAT International Conference: Optimization of SWAT performance using a Python tool 6

SPOTPY + SWAT

9/19/2018 2018 SWAT International Conference: Optimization of SWAT performance using a Python tool 7

SPOTPY + SWAT

9/19/2018 2018 SWAT International Conference: Optimization of SWAT performance using a Python tool 8

http://fb09-pasig.umwelt.uni-giessen.de/spotpy/Tutorial/8-Algorithm_guide/

Comparing methodologies

9/19/2018 2018 SWAT International Conference: Optimization of SWAT performance using a Python tool 9

9/19/2018 2018 SWAT International Conference: Optimization of SWAT performance using a Python tool 10

MCMCMCLHS

Dens

ityDe

nsity

Dens

ity

Comparing methodologies

9/19/2018 11

LHSMCMCMC

9/19/2018

Flow duration curve

Thank you!

More information: http://fb09-pasig.umwelt.uni-giessen.de/spotpy/

Camargos.S.Carla@umwelt.uni-giessen.de

9/19/2018 2018 SWAT International Conference: Optimization of SWAT performance using a Python tool 13

9/19/2018 2018 SWAT International Conference: Optimization of SWAT performance using a Python tool 14