Historical satellite, aircraft, and in situ data have shown that two anticyclonic gyres (the western and eastern Alboran gyres) are major ocean features of the Alboran Sea. An examination of several years of satellite imagery indicates that large variations in the surface expression of these two gyres occur and that on occasion one or the other gyre disappears (the disappearance of both gyres at the same time was not seen). The initial disappearance of either gyre occurs on a time scale of a week to 2 weeks, whereas the return may take from 3 weeks to 2 months. Various forcing mechanisms, i.e., winds, mass flux inflow through the Straits of Gibraltar and Sicily, and/or density, have been used in numerical ocean circulation models to study the dynamics of the western Mediterranean Sea. Various model results show relationships similar to those shown by the satellite imagery. However, no single forcing mechanism has been positively identified as the source of the disappearances, and the events may be a result of a combination of forcing mechanisms.

No abstract

No abstract

Sinking rates of particles were superimposed on the x-z-t current field observed at 15S off the coast of San Juan, Peru in March-May, 1977 to calculate particle trajectories in the upwelling circulation. Vertical velocities were calculated by a modified variational objective analysis technique using the measured onshore and longshore currents in conjunction with the physical constraint of mass continuity. The calculated vertical flow showed considerable temporal and spatial variability, with the mean vertical transport varying by two orders of magnitude over the 16 km wide continental shelf. Changes in direction occurred rapidly (within 24 h) as has been observed for horizontal circulation in this region. The vertical velocity of water was much greater than the sinking rates of particles during the 52 day period, so that the net vertical transport of particles was controlled by the vertical velocity of the water. Reseeding of sinking particles from the surface offshore-flowing layer into the deeper onshore flow could not be demonstrated for this period, which may explain why the measured biomass and primary productivity were anomalously low in 1977. Vertical mixing greatly increased the possibility of reseeding by transporting material downward into the onshore flow. We suggest that recirculation of particles may normally occur in the Peruvian upwelling system, but that the continuity probably involves movement in the longshore as well as the cross-shelf direction.

Abstract. Nested boundary techniques are developed based on the results of Perkins [1993] and Blake [1991]. We focus on the numerical and physical consistency needs across the nested boundary. These techniques replace the transition zone used by other researchers with a numerically and a physically based correction step. We demonstrate our method using a nested, reduced gravity version of the Naval Research Laboratory (NRL) primitive equation ocean model and a two-layer hydrodynamic finite-depth version of the same model. Numerical experiments are performed using integrations of both an idealized double gyre and a realistic Greenland Iceland Norwegian (GIN) Sea configuration. To illustrate the need for improved boundary treatment, we present a numerical study of boundary errors. The study illustrates the fragile nature of nested boundary conditions. With even small errors, a dramatic impact is observed on the formation (or lack thereof) of the Atlantic-Norwegian Current, which is responsible for transporting North Atlantic water to the Arctic Ocean, in the GIN Sea configuration. In this paper we propose an alternative to transition zones based on a numerically and a physically based correction that utilizes the two-sided information available along the internal boundary, providing some support from the entire domain of dependence. Perkins [1993] stressed the two-sided nature of nested domains by viewing the nested model as one overall or aggregate solution. She showed that the "usual" one-sided nesting methodology did not lead, in general, to an aggregate solution that was continuous across the internal boundary. A boundary method that results in values at coincident coarsefine mesh locations being continuously "locked" together as we integrate in time seems attainable, but it cannot be achieved unless we utilize values on both sides of the internal boundary.Ookochi [1972] tried to use information from both sides of the internal boundary to find a nesting scheme that was both continuous (or single valued) and conservative. He concluded that he had to either ignore continuity or conservation; he sacrificed conservation. Among other methods that have utilized two-sided information, Kurihara et al.'s [1979] study enforced energy conservation across the internal boundaries, but they conclude that conservation did not work well for them. Supporting their conclusions, the work of Peggion [1994] indicates that conservation of energy is not appropriate on a nested mesh unless the numerical artifacts introduced where the grid is refined are taken into account. The application of her results would more appropriately be termed energy management, rather than conservation.These numerical artifacts do not effect conservation of mass, however.In this paper we demonstrate numerically some of the errors that can occur when continuity is not maintained. In order to 3483