North West Shelf Joint Environmental Management Study:InVitro Inputs - Statistical extrapolation of currents

The application of the MSE modelling approach to the North West Shelf required three scenarios, for currents and contaminants, each spanning four decades to be simulated for many model runs. These scenarios were loosely defined as: 'high winds' to simulate the potential effect of stronger than normal winds; 'normal' where the distributional property of the winds were the same as those experienced currently; 'low winds' where winds were weaker than current conditions. Three-dimensional simulations of currents and the distributions of contaminants for such periods, at temporal resolutions of at least one hour, would require many months of computer time and, in addition, would generate large quantities of current data which would be impractical to incorporate into the agent-based simulation system InVitro. An alternate approach was required that was: 1) relatively fast, 2) made efficient use of limited computational resources and, 3) accurately represented the modelled currents throughout the study region. We chose to statistically model the currents by recognizing that near the coast the currents were primarily driven by the tides and elsewhere from a combination of tides, winds and boundary forcing. Tide and wind data were available but unfortunately we did not have ready access to the boundary forcing information - this is required in nesting the higher-resolution North West Shelf model within a coarse global circulation model. This was not considered to be a major deficiency as the boundary forcing (derived from a global General Circulation Model) was of coarse resolution with inadequate local representation of winds, bathymetry and water mass structure (Scott Condie, pers. Comm.). Our analyses were therefore focused on statistically modelling the currents and contaminants forced by local tides and winds.

将MSE建模方法（MSE modelling approach）应用于西北陆架（North West Shelf）时，需针对海流与污染物构建三种情景，每种情景的时长跨度达四十年，且需通过大量模型运行完成模拟。上述三种情景可大致界定为："强风情景"，用于模拟强于常态的风力所产生的潜在影响；"常态情景"，其风力分布特性与当前实际观测一致；"弱风情景"，对应风力弱于当前条件的情况。若针对上述时段开展时间分辨率不低于1小时的海流与污染物分布三维模拟，不仅需要耗费数月的计算机算力，还会产生海量海流数据，难以集成至基于智能体的模拟系统InVitro。因此需要一种替代方案，需满足以下三点要求：1）运算速度相对较快；2）高效利用有限的计算资源；3）能够精准复现研究区域内所有模拟海流的特征。经分析，近岸海流主要由潮汐驱动，而远海区域的海流则由潮汐、风力与边界强迫共同驱动，因此我们选择采用统计建模方法对海流进行模拟。潮汐与风力数据已可获取，但遗憾的是我们无法直接获取边界强迫数据——若要将高分辨率西北陆架模型嵌套至粗分辨率全球环流模型（General Circulation Model）中，边界强迫数据是必需的。由于该边界强迫数据源自全球环流模型，分辨率较为粗糙，无法精准表征局地风力、水深与水团结构（Scott Condie，个人通信（pers. Comm.）），因此这一缺陷并未被视为严重问题。基于此，我们的分析重点聚焦于由局地潮汐与风力驱动的海流与污染物统计建模。