Is There Value in Very High Resolution Weather Forecasts? Experiences from The Jefferson Project at Lake George

The Jefferson Project at Lake George, NY is a collaborative project between IBM Research, Rensselaer Polytechnic Institute and The FUND for Lake George. A core component of the project is to develop a system capable of real-time observations and interactive modeling of the atmosphere, watershed hydrology and lake circulation. Here, we discuss the challenges of implementing a robust, very high resolution (dx = 0.33 km) operational weather forecast system using WRF-ARW in a region of complex topography. The system includes data assimilation (3DVAR), 3D visualization and an automated verification framework. Of particular interest is the ability for the weather model to predict i) wind stress at the lake surface for input into the lake circulation model; and ii) precipitation within the watershed for input into the stream network model. Results from select case studies and nine months of operational forecasts are used to demonstrate how model resolution effects the simulation of these three fields. We find the surface wind patterns (viz. wind stress) become significantly different as model resolution increases, although only in specific areas of the lake and during certain weather conditions (e.g., strong E-W winds). The average lake-wide wind stress, however, is roughly equal for the inner three domains (dx = 3.0, 1.0, 0.33 km). This suggests a coarser resolution simulation is a computationally reasonable approach for longer-term coupled atmosphere-lake applications at Lake George. For precipitation, comparison to our local observational network and radar-derived products suggests there is no statistically significant difference in the accuracy of accumulated precipitation across the inner three domains. However, the orographic signature in precipitation becomes more prominent at higher resolution impacting how precipitation is divided among the many sub-watersheds of Lake George.