Accounting for climate uncertainty in hydrologic model calibration and ensemble Kalman filter based on an existing historical climate ensemble dataset
HAPpy Hour Seminar
Aug. 24, 2018
3:30 pm MDT
NCAR Foothills Lab, FL2-3107
Our research demonstrates using the Gridded Ensemble Precipitation and Temperature Estimates dataset (Newman et al., 2015), covering the contiguous United States, northern Mexico, and southern Canada, to represent the precipitation and temperature uncertainty in model calibration and the ensemble Kalman filter. First, an ensemble climate based model calibration framework is proposed to explicitly account for climate data uncertainty in model calibration. The framework performance is verified with 30 synthetic experiments and 20 Québec catchments real case studies. Results show that the framework effectively reduces the inaccurate flow predictions caused by poor quality climate measurements and improves the overall performance of ensemble flow predictions. Second, we propose the direct use of the Newman et al. (2015) dataset in the ensemble Kalman filter application. Our research for the first time compares the Newman et al. (2015) dataset based climate ensemble with the carefully tuned hyper-parameters based climate ensemble in real ensemble flow forecasting. The forecast performance comparison of 20 Québec catchments shows that the Newman et al. (2015) ensemble yields improved or similar deterministic and probabilistic flow forecasts relative to the tuned hyper-parameters based climate ensemble, especially for short lead times (i.e., 1-3 days) when the influence of data assimilation dominates. However, the analysis and experiment time required to use the Newman et al. (2015) dataset is much less compared to the fastidious hyper-parameter tuning.
Hongli Liu, PhD student at the University of Waterloo, Ontario