Comparison of Observed Mixed-Layer Depths to Model Estimates Using Observed

Publisher Website
Google Scholar
Abstract
Two mixed-layer models were tested diagnostically with the Wangara and O'Neill data, and one year of rawinsonde and surface data from four southeastern U. S. stations. They were tested as prediction models with MOS and LFM model forecasts. One model determines convective mixed-layer depths by intersecting a morning sounding with a dry adiabat. The other is a zero-order entrainment model with a separate expression for the free convection regime and a simple cloud parameterization. Both models performed best with the Wangara and O'Neill data, with root-mean-square errors (RMSE) that were ∼25% of the average observed mixed-layer depth. The RMSE rose to ∼50% with the LFM and MOS forecasts. The zero-order entrainment model was consistently the more accurate model, although not by wide margins. Winter month performance of the zero-order entrainment model improved by 12% when a seasonally varying entrainment coefficient was introduced. Abstract Two mixed-layer models were tested diagnostically with the Wangara and O'Neill data, and one year of rawinsonde and surface data from four southeastern U. S. stations. They were tested as prediction models with MOS and LFM model forecasts. One model determines convective mixed-layer depths by intersecting a morning sounding with a dry adiabat. The other is a zero-order entrainment model with a separate expression for the free convection regime and a simple cloud parameterization. Both models performed best with the Wangara and O'Neill data, with root-mean-square errors (RMSE) that were ∼25% of the average observed mixed-layer depth. The RMSE rose to ∼50% with the LFM and MOS forecasts. The zero-order entrainment model was consistently the more accurate model, although not by wide margins. Winter month performance of the zero-order entrainment model improved by 12% when a seasonally varying entrainment coefficient was introduced.

This publication has 0 references indexed in Scilit: