Modelling the ocean circulation with random numbers

海洋包含一个强有力的地转中尺度涡旋场，在中纬度的空间尺度约为100公里，随着时间尺度从几周到几个月的演变。这些涡旋对于建立海洋环流和海洋内部的示踪剂分布非常重要。在墨西哥湾流和黑潮延伸区附近以及南极绕极流（ACC）中，涡旋是热量向极地输送的主要方式。然而，为了充分模拟数值模式中的中尺度涡动场和相关的示踪剂湍流级联，水平分辨率至少为10公里是必要的。因此，气候模型不可能系统地解决海洋涡旋问题，其影响需要参数化。在中尺度涡旋的确定性参数化的实现方面已经取得了很大的进展。尽管有这样的改进，模式误差的主要来源仍然是这些次网格尺度processes.In目前的建议，我们的目标是提高我们的知识如何海洋涡旋变率影响大尺度气候的不完善或缺失的关闭。我们的目标还在于开发一种新的随机闭包作为当前确定性参数化的替代方法。这样的随机封闭提供了一个显式表示的固有的不确定性相关的参数化过程的优点。将在一个最先进的气候模式上测试新的涡旋随机参数化方法，并将研究对平均气候、其变率和可预测性的影响。

The ocean contains a vigorous geostrophic mesoscale eddy field with spatial scales of approximately 100km at mid-latitudes, evolving over time scales from weeks to months. These eddies are important in establishing the ocean's circulation and tracer distributions in the ocean' interior. In the vicinity of the Gulf Stream and the Kuroshio's extensions and in the Antarctic Circumpolar Current (ACC), eddies are the primary means by which heat is transported poleward. However, to adequately simulate the mesoscale eddy field and associated turbulent cascades of tracers in numerical models, horizontal resolution of at least 10 km is necessary. It is therefore unlikely that climate models will systematically resolve ocean eddies and their effect needs to be parametrized. Large improvements have been made in the implementation of deterministic parameterizations of mesoscale eddies. Despite such improvements, the major source of model error remains the imperfect or missing closure of these subgrid scale processes.In the current proposal, we aim to improve our knowledge of how ocean eddy variability influences the large-scale climate. We also aim to develop a novel stochastic closure as an alternative approach to the current deterministic parameterization. Such stochastic closure offers the advantage of an explicit representation of the inherent uncertainty associated the parametrized processes. The new stochastic parameterization of eddies will be tested on a state-of-the-art climate model and the impact on the mean climate, its variability and predictability will be investigated.