Creating an Ensemble Streamflow Forecast Through the Systematic Shifting of QPF

Heavy rainfall events, which can cause flash flooding, are quite common in the U.S. Upper
Midwest. Quantitative precipitation forecasts (QPF) are used in combination with hydrologic
models to help provide an early warning of flash flooding. However, uncertainty in the
forecasted precipitation amount and location remains a challenge. One way to represent the
uncertainty in the forecasted amount is to use an ensemble of QPF to produce an ensemble
streamflow prediction. In this work, we propose an approach to address the uncertainty in QPF,
specifically uncertainty due to spatial displacement errors, by systematically shifting QPF output
to create additional ensemble members for input into hydrologic models. We test ensemble
QPF from the advanced National Oceanic and Atmospheric Administration (NOAA) High
Resolution Rapid Refresh Ensemble (HRRRE) prediction system, which is currently in the
experimental stage. The HRRRE consists of nine members with varying initial conditions and has
3-km horizontal grid spacing. From the nine HRRRE members, 81 QPF ensemble members are
generated and used for input into the National Weather Service (NWS) Hydrology Laboratory-
Research Distributed Hydrologic Model (HL-RDHM) to create an ensemble streamflow forecast.
Nine of the inputs are created with raw QPF from each of the individual HRRRE members. The
remaining 72 members (9 HRRRE members x 8 directional shifts) are created by systematically
shifting each of the nine HRRRE members in the cardinal (N, S, E, and W) and intermediate (NE,
NW, SE, SW) directions. Shifts in the cardinal directions are 0.5⁰ latitude and shifts in the
intermediate directions are 0.5⁰ latitude in both the zonal and meridional directions. An initial
test has been completed on the Pecatonica River and East Branch Pecatonica River watersheds
located in southwest Wisconsin for an event that occurred on July 19-20, 2017. Initial results
show that the systematic shifting of QPF did increase the overall forecast spread (both in
intensity and timing) compared to using the nine raw QPF outputs, allowing for the ensemble
prediction to better capture the event for both watersheds. Further analysis of this technique is
planned for an estimated 20 events covering areas large enough to include multiple watersheds
that experienced flooding adjacent to those that did not experience flooding. Additional shifting
magnitudes (e.g. 0.75⁰ or 1⁰ latitude) may also be investigated with the analysis. Discharge
forecasts (both ensemble mean forecasts and probabilistic forecasts) are compared to the
United States Geological Survey (USGS) streamflow observations to assess the efficacy of this
process.