Submarine Ice Thickness Data Analysis for Climate Change and Model Improvement

通过分析观测资料，特别是阐明年际变化，并通过对观测到的和模拟的冰厚之间的差异进行严格评估，提供资金以巩固关于北极海冰厚度的知识。到2005年春天，从1969年到1999年的30年间进行的大约35次航行的海底冰吃水数据应该可以公开。提出了两个任务来分析这些数据。首先，使用多元回归分析将数据拟合为统计表示。这意味着将年际变化从空间和季节变化中分离出来。其次，将产生厚度数量的最佳插值场，主要是平均值，但也有模式，厚度的概率密度函数和开放水分数。厚度数据将分为冬季和夏季两组，并对其进行分析，以生成整个31年数据集的最佳场，其中包括厚度最大的年份和厚度最小的年份。分析数据的两种表示形式——多元回归和最佳插值字段——将在项目结束时被放入国家档案馆，供社区使用。一个内部的、最先进的北冰洋冰-海耦合模式，处理海冰的动量、热量和质量平衡，包括海冰厚度与积雪的详细处理，已经被证明可以成功地模拟北冰洋观测到的平均冰厚和从20世纪80年代末到现在冰厚的普遍下降。然而，该模式在再现厚度的空间场，即整个北冰洋的变化方面并不令人满意。通过集中注意模式与观测值之间差异的空间结构，应该可以对模式进行实质性改进。建议分析这些差异是如何在空间、时间和厚度本身上构成的，并调查冰-海洋模式及其强迫数据的哪些方面最有可能引起这些差异。强迫数据，如云量、辐射和降雪，以及模式成分，如漂流和隆起，都是详细调查的候选数据。改进的海冰模型对于了解北极与全球气候系统的相互作用是必要的。如果能可靠地预测异常薄的海冰出现的时间和地点，它将对土著猎人的安全以及通过北极的航运路线等各种话题产生影响。

Funds are provided to solidify knowledge of arctic sea ice thickness through analyses of observations, which will, in particular, elucidate interannual variations, and through a critical assessment of discrepancies between observed and simulated ice thickness. By the spring of 2005, submarine ice draft data should be publicly available from some 35 cruises that took place during three decades - 1969 to 1999. Two tasks are proposed to analyze these data. First, the data will be fit to a statistical representation using multiple regression analysis. This is meant to isolate interannual variability from spatial and seasonal variability. Second, optimally interpolated fields of thickness quantities will be produced, primarily the mean, but also the mode, the probability density function of thickness, and the open water fraction. Thickness data will be separated into winter and summer groups and analyzed to produce optimal fields for the entire 31-year data set, for several years at a thickness maximum, and for several years at a thickness minimum. Both representations of the analyzed data- the multiple regressions and the optimally interpolated fields- will be placed in a national archive for community use by the end of the project. An in-house, state-of-the-art, coupled ice-ocean model of the Arctic Ocean, treating the combined momentum, heat, and mass balances of sea ice, including a detailed treatment of sea ice thickness with snow cover, has been shown to simulate successfully observed mean ice thickness averaged over the Arctic Ocean and the general decline in ice thickness from the late 1980s to the present. However, the model is unsatisfactory at reproducing the spatial field of thickness, i.e. the variations across the Arctic Ocean. Substantial improvement in the model should be possible through focused attention on the spatial structure of the discrepancies between model and observations. It is proposed to analyze how these discrepancies are structured in space, time, and with thickness itself and to investigate what aspects of the ice-ocean model and its forcing data are most likely causing these discrepancies. Forcing data, such as cloudiness, radiation, and snowfall, and model components, such as rafting and ridging, are candidates for detailed investigation.Improved sea ice models are necessary to understand the interactions of the Arctic with the global climate system. If the timing and location of abnormally thin sea ice were reliably predictable, it would have implications for topics as varied as the safety of indigenous hunters and the routing of shipping through the Arctic.