89th American Meteorological Society Annual Meeting

Tuesday, 13 January 2009: 1:30 PM
Results of phase 2 of the Distributed Model Intercomparison Project
Room 127BC (Phoenix Convention Center)
Michael B. Smith, NOAA/NWS, Silver Spring, MD; and B. Cosgrove, S. Reed, Z. Zhang, F. Moreda, Z. Cui, N. Mizukami, and S. Sheldon
The National Weather Service (NWS) continues the development and deployment of distributed hydrologic models for operational river, flash flood, and water resources forecasting. To expedite the research to operations pathway, the NWS has organized phases 1 and 2 of the Distributed Model Intercomparison Project (DMIP). DMIP 1 and 2 were designed as venues for the scientific community to test their models in experiments related to NWS forecasting mission goals. Phase 1 of DMIP focused on lumped and distributed modeling of simple basins in the southern Great Plains. Following the successful completion of DMIP 1, the NWS deployed a gridded distributed model to field offices in early 2007. We envision that enhanced versions of the distributed model will be developed and deployed as a result of DMIP 2 activities.

DMIP 2 is currently underway. Experiments are designed to revisit issues not fully addressed in the DMIP 1 Oklahoma basins and also to examine new distributed modeling issues in the mountainous western United States. Tests were designed for the Oklahoma basins to investigate the use of distributed models for prediction of soil moisture as well as streamflow. In addition, more observations from stream gauge points were gathered to help assess the performance of distributed models in ungauged areas. Fourteen groups submitted simulations for the Oklahoma experiments. Another major thrust of DMIP 2 is to examine modeling issues in mountainous areas with complexities such as sparse data collection networks, orographically enhanced precipitation, snow, complex terrain features and others. Six groups submitted simulations for the Sierra Nevada basins.

In this presentation, we will discuss the results of the Oklahoma and Sierra Nevada experiments. We will present goodness-of-fit statistics for lumped and distributed streamflow comparisons. Results of using distributed models to compute estimates of soil moisture and snow water equivalent will also be discussed.

Supplementary URL: