A general framework for an unmixed layer PBL model

Mixed-layer models are computationally efficient, but they do not realistically represent the structure of the boundary layer under many conditions. Many of the deficiencies of the mixed-layer model can be attributed to the assumed flat profiles. A new method is proposed that, by relaxing the assumption of well-mixed profiles, makes possible an integral PBL parameterization that is computationally efficient, yet accurately describes the mean structure of the boundary layer. The vertical structure of the mean variables in the PBL is represented by a truncated series of Legendre polynomials. The first Legendre mode, the layer average, is identically a mixed-layer model. Additional modes add structure to the vertical profiles and represent corrections to the mixed-layer model. Only a few modes are necessary to produce vertical profiles comparable to the predictions of high-resolution models. Results of the model are shown for a variety of PBL stability regimes.