Core Project 3. Uncertainty analyses and model diagnostics

Establishment of assessing methods for hydrologic model performance through a development of uncertainty evaluation indices

PI: A. Kawamura at Tokyo Metropolitan University, kawamura-akira@metro-u.ac.jp
S. Kazama at Tohoku Iniv
,
kazama@kaigan.civil.tohoku.ac.jp

Objectives

Objective 1: Development of new uncertainty evaluation methods and indices
Objective 2: Establishment of a guide line for selecting hydrologic models using new uncertainty evaluation methods and indices

Participants

Detailed description of the range of approaches to be evaluated within the working group

In Japan, many physically based distributed hydrological models have been developed and applied to many watershed basins, while the storage function model, a simple lumped conceptual model are still widely used for actual watershed planning. This is because 1) the indices to show the model performance and/or uncertainty are not well defined, 2) the methods to assess model performances and/or uncertainty are not clear, thus 3) the performances of newly developed models are not well assessed. In association with watershed properties, data availability and the purpose of model use, a suitable hydrologic model will be different in model structure, distributed or lumped, and so on.

In this study, uncertainty evaluation methods and indices to show the model performance and uncertainty are newly devised to assess the uncertainty derived from forcing data, model structure and model parameter identifications. Through the newly developed indices, existing lumped and distributed models are assessed in their forecasting performance and uncertainty, and in what environment what kind of model shows higher performance than other models is clarified. Finally, a guide line to recommend which model should be applied in ungauged basins will be established.