Absolute and convective instabilities and their roles in the forecasting of large frontal meanderings. Issue 10 (22nd October 2013)

• Record Type: Journal Article
• Authors: Liang, X. San; Robinson, Allan R.
• Is Part Of: Journal of geophysical research. Volume 118:Issue 10(2013:Oct.)
• Page Start: 5686
• Page End: 5702
• Publication Date: 2013-10-22
• Subjects: Oceanography -- Periodicals; 551.4605
• DOI: 10.1002/jgrc.20406
• Languages: English
• ISSNs: 2169-9275

Abstract

<abstract abstract-type="main"> <title> <x xml:space="preserve">Abstract</x> </title> <p>[1] Frontal meanderings are generally difficult to predict. In this study, we demonstrate through an exercise with the Iceland‐Faeroe Front (IFF) that satisfactory predictions may be achieved with the aid of hydrodynamic instability analysis. As discovered earlier on, underlying the IFF meandering is a convective instability in the western boundary region followed by an absolute instability in the interior; correspondingly the disturbance growth reveals a switch of pattern from spatial amplification to temporal amplification. To successfully forecast the meandering, the two instability processes must be faithfully reproduced. This sets stringent constraints for the tunable model parameters, e.g., boundary relaxation, temporal relaxation, eddy diffusivity, etc. By analyzing the instability dispersion properties, these parameters can be rather accurately set and their respective ranges of sensitivity estimated. It is shown that too much relaxation inhibits the front from varying; on the other hand, too little relaxation may have the model completely skip the spatial growth phase, leading to a meandering way more upstream along the front. Generally speaking, dissipation/diffusion tends to stabilize the simulation, but unrealistically large dissipation/diffusion could trigger a spurious absolute instability, and hence a premature meandering intrusion. The belief that taking in more data will improve the forecast does not need to be true; it depends on whether the model setup admits the two instabilities. This study may help relieve modelers from the laborious and tedious work of parameter tuning; it also provides us criteria to distinguish a physically relevant forecast from numerical artifacts.</p> </abstract>