Final published version
Research output: Contribution to journal › Journal article › peer-review
Concepts of information content and likelihood in parameter calibration for hydrological simulation models. / Beven, Keith John; Smith, Paul James.
In: Journal of Hydrologic Engineering, Vol. 20, No. 1, A4014010, 26.02.2014.Research output: Contribution to journal › Journal article › peer-review
}
TY - JOUR
T1 - Concepts of information content and likelihood in parameter calibration for hydrological simulation models
AU - Beven, Keith John
AU - Smith, Paul James
PY - 2014/2/26
Y1 - 2014/2/26
N2 - There remains a great deal of uncertainty about uncertainty estimation in hydrological modeling. Given that hydrology is still a subject limited by the available measurement techniques, it does not appear that the issue of epistemic error in hydrological data will go away for the foreseeable future, and it may be necessary to find a way to allow for robust model conditioning and more subjective treatments of potential epistemic errors in prediction. In this paper an attempt is made to analyze how this is the result of the epistemic uncertainties inherent in the hydrological modeling process and their impact on model conditioning and hypothesis testing. Some ideas are proposed about how to deal with assessing the information in hydrological data and how it might influence model conditioning based on hydrological reasoning, with an application to rainfall-runoff modeling of a catchment in northern England, where inconsistent data for some events can introduce disinformation into the model conditioning process. A methodology is presented to make an assessment of the relative information content of calibration data before running a model that can then inform the evaluation of model runs and resulting prediction uncertainties.
AB - There remains a great deal of uncertainty about uncertainty estimation in hydrological modeling. Given that hydrology is still a subject limited by the available measurement techniques, it does not appear that the issue of epistemic error in hydrological data will go away for the foreseeable future, and it may be necessary to find a way to allow for robust model conditioning and more subjective treatments of potential epistemic errors in prediction. In this paper an attempt is made to analyze how this is the result of the epistemic uncertainties inherent in the hydrological modeling process and their impact on model conditioning and hypothesis testing. Some ideas are proposed about how to deal with assessing the information in hydrological data and how it might influence model conditioning based on hydrological reasoning, with an application to rainfall-runoff modeling of a catchment in northern England, where inconsistent data for some events can introduce disinformation into the model conditioning process. A methodology is presented to make an assessment of the relative information content of calibration data before running a model that can then inform the evaluation of model runs and resulting prediction uncertainties.
KW - Rainfall-runoff modelling
KW - Uncertainty estimation
KW - Epistemic error
KW - Disinformation
KW - GLUE
KW - Event clustering
U2 - 10.1061/(ASCE)HE.1943-5584.0000991
DO - 10.1061/(ASCE)HE.1943-5584.0000991
M3 - Journal article
VL - 20
JO - Journal of Hydrologic Engineering
JF - Journal of Hydrologic Engineering
SN - 1084-0699
IS - 1
M1 - A4014010
ER -