Evaluation of Advantages of the Continuous SAC-SMA Model over an Event API Model

One of the key elements of the Advanced Hydrologic Prediction System (AHPS) proposed by the National Weather Service (NWS) is the Sacramento Soil Moisture Accounting (SAC-SMA) model. Through AHPS the NWS River Forecast Centers (RFCs) will be able to calibrate and implement the SAC-SMA model and abandon the old event based Antecedent Precipitation Index (API) models. The SAC-SMA model will improve the accuracy of the RFC forecasts because the API models are based on regional empirical relationships between runoff and rainfall characteristics while the SAC-SMA model is calibrated to reflect the hydrologic characteristics of individual basins.

This paper presents the results quantifying the improved skill of the event SAC-SMA model over the event API models. The models were run for forty to fifty year periods without any adjustment of model parameters or states. The models were not adjusted in order to evaluate the model performance in the absence of forecaster skill. A simple base flow operation was added to the API model to generate a base flow during low flow periods. The first tests were performed for two headwater basins of the Des Moines River basin, Iowa. Two measures of error have been used to quantify the SAC-SMA improvements over the API model: hydrograph shape error was computed as a root mean square error of simulated hydrographs during flood events, and flood peak error was estimated as an difference between observed and predicted flood peaks. The comparison over more than 200 flood events found hydrograph shape accuracy improvements of 20% for low flood events and of 40% for high flood events, and peak error improvements of 23%. These tests will be extended over different river regimes and climate regions.