One-Level Diagnostic Modeling of Mesoscale Surface Winds in Complex Terrain. Part I: Comparison with Three-Dimensional Modeling in Israel

Abstract

A one-level sigma-coordinate model originally developed by Danard and modified by Mass and Dempsey and Alpert et al., is applied to the study of surface flow over an averaged summer diurnal cycle in Israel. The detailed flow features are compared to three-dimensional modeling studies and to dense surface wind observations. The winds al a height of 10 m from the one-level model were found comparable to those obtained by three-dimensional simulations, and in some cases the one-level model predicted observed surface flow features that were not simulated by the three-dimensional simulations, probably because of the finer horizontal grid resolution in the one-level model. The two models had similar deficiencies in diagnosing observed flow features in many cases. A severe drawback of the one-level model is the inability to advance the sea-breeze front (SBF) over a ridge crest correctly. Based upon an earlier vertical cross-sectional study by Alpert et al., an explanation for this discrepancy is suggested. In a detailed, high-resolution analysis of the summer mesoscale flow, the surface horizontal winds from the one-level high-resolution model and the three-dimensional model are compared to the relatively dense network of wind observations in Israel every three hours, for an averaged diurnal cycle. Several features of the surface flow are revealed and illustrated, including (i) Flow convergence in the evening near the coast; (ii) A convergence line south of Lake Kinneret (Sea of Galilee) that forms in the morning, moves southward along the Jordan Rift Valley, and finally merges with the SBF in the afternoon; (iii) Relatively strong nocturnal flow at the southern Mediterranean coast of Israel, possibly due to the concave shoreline there. These, as well as many other observed flow features, are simulated by both the one-level model and the three-dimensional model, though the one-level model required only modest computer resources. Hence, this study illustrates that the trade-off between horizontal resolution and explicit vertical resolution may be most beneficial, at least when topographic and surface forcing dominate.