Eastern Region technical attachment (National Weather Service (U.S.)) ; no. 2013-01

"Hydrologic simulation comparisons were made between the Sacramento Soil Moisture Accounting (SAC - SMA) model and the Continuous Antecedent Precipitation Index (API - CONT) model for six Susquehanna River Basin (SRB) headwater basins. Using a 6 - hour time step, the difference in cumulative simulation error (stage) between observations and model simulations was calculated over a 19-month period (August 2010 - February 2012) to visualize event, monthly, and seasonal model trends. Next, simulated crests (stage) were compared to observed crests for five rain-driven events. Finally, Ensemble Streamflow Prediction (ESP) was used on a daily time scale to provide a long-term (1950-1998) perspective of model performance compared to historical high flow events. Results show that the SAC-SMA model consistently produced more accurate streamflow simulations in terms of the cumulative error in stage over time. However, we could not establish a distinct seasonality (i.e. wet or dry periods, snowmelt) pattern in which one model was consistently more accurate across all basins. The API-CONT model more accurately predicted large event (flood) crests based on the Probability of Detection (POD), False Alarm Rate (FAR), Critical Success Index (CSI), and bias verification statistics. Results suggest that use of the lumped SAC-SMA could improve upon the lumped API-CONT for low and medium flow forecasts, which may be particularly useful for longer lead time applications such as water supply management. However, no improvements would be expected for flood forecasts"--Abstract.

National Weather Service (NWS)
Office of Oceanic and Atmospheric Research (OAR)

Susquehanna River Watershed