Increasing Skill for Hydrologic Monitoring and Forecasting
Speaker: Joshua Roundy, Graduate Student
Series: EEWR Brown Bag Seminars
Location: Engineering Quad E219
Date/Time: Friday, April 22, 2011, 12:00 p.m. - 1:00 p.m.
Extreme hydrologic events in the form of droughts or floods are a significant source of social and economic damage in many parts of the world. Having sufficient warning of extreme events allows managers to prepare for and reduce the severity of their impacts. A hydrologic forecast system can give predictions that range from time scales of hours (short range) to monthly (seasonal) that can be used by mangers to make better decisions; however there is still much uncertainty associated with such systems. Therefore, it is important to understand the forecast skill of the system before transitioning to operational usage. Using our national hydrologic monitoring and forecast system, seasonal forecasts from the NCEP Climate Forecast System (CFS) are downscaled and used to drive the Variable Infiltration Capacity (VIC) land surface hydrologic model to give hydrologic variables with lead times of up to six months. We focus on the Apalachicola-Chatahoochee-Flint (ACF) River Basin in the South Eastern United States, which has experienced a number of severe droughts in recent years. The performance of the VIC model is evaluated using observational forcing and compared to observed streamflow. The effectiveness of the forecast system to predict streamflow is evaluated from 30 years of CFS hindcasts from 1981-2010 and compared to modeled streamflow driven by observed atmospheric forcing. The forecast skill from the CFS is compared with forecasts based on the Ensemble Streamflow Prediction (ESP) method, which uses initial conditions and historical forcings to generate seasonal forecasts. The skill of the system to predict extreme hydrologic events is assessed.
It is shown that there is skill in predicting drought continuation at short lead times and that the use of a seasonal forecast model has a small advantage over the traditional ESP method. However, the seasonal forecasts show no skill for predicting seasonal flooding. In order to correctly predict flooding more accurate precipitation estimates are needed. In addition flooding usually occurs at times much smaller than the seasonal scale, therefore in order to make meaningful flood forecasts it is necessary to incorporate medium range weather forecasts. The transition from seasonal (monthly) to medium range (daily) hydrologic forecasts requires a better representation of the hydraulics and hydrology. Improved hydrologic model skill for estimating streamflow is achieved through the implementation of a new routing model and a process for estimating sub-grid heterogeneity in baseflow. Increasing the skill of hydrologic predictions is only the beginning step to implementing a medium range global scale hydrologic forecast model.