A special study of the Sharav Cyclones indicates that they are the result of large-scale weak baroclinicity, enhanced by vigorous boundary-layer baroclinicity between the North African coast and the Mediterranean. It is illustrated how the jet stream plays a major role in the vertical circulation in producing a complex cyclonic circulation dominated by at least three mechanisms: large-scale interior baroclinicity, boundary-layer baroclinicity, and jet stream related circulations. The main characteristics of the Sharav Cyclone (also called the Saharan Depression of Kamsin Depression) in the Mediterranean are reviewed. Unlike the cold winter cyclone, the Sharav Cyclone is a spring cyclone. Its tracks lie mainly along the North African coast and turn to the north near the southeastern Mediterranean. Its warm front is active and is sometimes associated with extremely high surface temperatures. Its cold front is shallow. The Sharav Cyclone moves eastward relatively fast, typically faster than 10 m s-1, and with as mall speed variability.

In general, there is an upper level trough to the west of the surface low and the surface horizontal scale is of the order of 500--1000 km. Finally, it is frequently associated with heavy dust/sand storms and low visibilities. Some of these features are illustrated in a case study of the April 28--30, 1986, cyclone. Vertical cross sections indicate a deep circulation associated with the exit region of an upper level jet. In addition to presenting evidence that the Sharav Cyclone is a deep tropospheric circulation, it is shown that the transverse indirect circulation at the exit region of the jet is a major component of its circulation. The classic two-level baroclinic model (Phillips, 1954) is applied. The effects of the major diabatic heating due to the sensible heat flux above the North African desert and the large north to south temperature gradients are incorporated through the thermal wind of the basic state. The model predicts the fast eastward motion, the relatively smaller horizontal scale and the fast growth rate. Furthermore, the model predicts an annual maximum growth rate in April and a secondary peak in October, which agrees with the frequency of occurrences of the Sharav Cyclone. ¿ American Geophysical Union 1989