Projecting Hydrologic Modeling Uncertainty across Multiple Basin Scales into Future Climate Periods.
Abstract
To assess the uncertainty associated with hydrologic modeling, the four major sources of uncertainty must be accounted for: input, structural, parametric, and measurement uncertainty in the target observed data. To consider each source of uncertainty, the underlying likelihood distributions can be sampled. When considering continental scale hydrologic models, the process of sufficiently sampling these sources of uncertainty, particularly the structural and parametric uncertainties, may require unattainably high computational resources. We present a methodology and results for a stochastic framework towards estimation of stream flow uncertainty bounds under historic and future climate. Structural and parametric relative partitions of uncertainty are transferred to the Hudson Bay Drainage Basin (HBDB) from a detailed uncertainty analysis case study of the Lower Nelson River Basin (LNRB). Quantile regression of simulated streamflow from the LNRB and the corresponding upstream antecedent precipitation is used to generate an estimate of the relative contribution of structural and parametric uncertainty to total uncertainty. Relatively partitioned structural and parametric uncertainty increased as antecedent precipitation increased, was generally highest in spring, and lowest in winter.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2019
- Bibcode:
- 2019AGUFMNG13A..07S
- Keywords:
-
- 3315 Data assimilation;
- ATMOSPHERIC PROCESSES;
- 0555 Neural networks;
- fuzzy logic;
- machine learning;
- COMPUTATIONAL GEOPHYSICS;
- 1640 Remote sensing;
- GLOBAL CHANGE;
- 3275 Uncertainty quantification;
- MATHEMATICAL GEOPHYSICS