Grid-Based Removal of Systematic Model Bias

All mesoscale model forecasts have biases, particularly near the surface. Many of these biases are systematic and repeatable, thus offering the potential for their removal in post-processing. Until now, most bias removal techniques, such as Model Output Statistics (MOS) or perfect-prog approaches, have been applied only at observation locations. However, there is an acute need for model bias removal on the model grid itself. The demand for grid-based bias removal has become particularly important as the National Weather Service, private sector, and other forecast groups move to grid-based forecast dissemination. Furthermore, grid-based bias removal is an essential step prior to the use of ensemble forecasts in preparing probabilistic products. This presentation will describe a new observation-based bias removal approach that for each gridpoint uses nearby observation locations with similar land use, elevation, and parameter values to compute the biases on an hourly basis. This approach will be evaluated for a large number of cases for surface parameters using 36 and 12-km MM5 forecasts over the Pacific Northwest, and will demonstrated substantial reductions in both bias and mean absolute error.