Graduation Semester and Year




Document Type


Degree Name

Master of Science in Civil Engineering


Civil Engineering

First Advisor

Dong-Jun Seo


Compared to forecasts of short-term precipitation accumulations (daily or shorter) at lead times larger than a few days, those of longer-term accumulations (3-daily or longer) are significantly more skillful owing to the larger temporal scale of aggregation. If one can utilize this skill present in medium-range precipitation forecast in hydrologic prediction, it is very likely that the lead time of hydrologic forecasts, in particular, of streamflow and soil moisture may be extended. Though forecasts of longer-term accumulations of precipitation are more skillful than those of shorter-term accumulations, precipitation forecasts in general are too uncertain to be used as deterministic, or single-valued, input. The main goal of this study is to increase forecast lead time of streamflow forecasts by using medium range ensemble precipitation forecasts. A premise for this study is that, in the ensemble paradigm, forecasting of precipitation and streamflow provides extending forecast lead time with improved forecast skill. To utilize forecast skill in medium range precipitation forecasts in the ensemble paradigm, this study uses Hydrologic Ensemble Forecast Service (HEFS). In the HEFS, the Meteorological Ensemble Forecast Processor (MEFP) was used to generate ensemble precipitation hindcasts using the Global Ensemble Forecast System (GEFS) reforecast data. Raw streamflow hindcasts were generated via the Community Hydrologic Prediction System (CHPS) using the Sacramento Soil Moisture Accounting model (SAC-SMA) and unit hydrograph. To reduce biases and uncertainties in the hydrologic model results, raw streamflow ensembles were post-processed by the Ensemble Postprocessor (EnsPost). The precipitation, raw and post-processed streamflow ensembles were verified using the Ensemble Verification System (EVS) to assess the quality of hindcasts. Ensemble hindcasts of precipitation and streamflow were generated using the HEFS for a 26-year period between 1986 and 2011. The study area consisted of five headwater basins located upstream of the Dallas-Fort Worth (DFW) metropolitan area in the Upper Trinity River Basin in Texas. The main findings of this study include: (1) adjusting modulation canonical events is a very effective way to improve predictive skill in ensemble forecasts of precipitation, raw, and post-processed streamflow forecasts: (2) GEFS-forced medium-range precipitation hindcasts for the study area have valuable skill in 1-, 3-, 5-daily, weekly, and biweekly-aggregated hindcasts; (3) in the ensemble paradigm, forecast skill in medium-range precipitation forecasts can be effectively utilized to improve the quality of streamflow forecasts in extended forecast lead time via HEFS. This study used the HEFS successfully, demonstrating the HEFS’s portability in the Unix/Linux environment outside of National Weather Service (NWS). This study also showed that the HEFS is an effective tool for generating skillful forecasts of precipitation and streamflow ensembles. This study would provide water resources managers with improved streamflow forecasts for the extended forecast lead time to effectively manage water resources and to mitigate water-related hazards.


Ensemble, Forecast, Streamflow, Precipitation, HEFS


Civil and Environmental Engineering | Civil Engineering | Engineering


Degree granted by The University of Texas at Arlington