从稀疏温盐测量中观测有限深度双扩散交换的可观测性
Observability of Finite-Depth Double-Diffusive Exchange from Sparse Temperature-Salinity Measurements
浏览论文内容
中文总结 AI 辅助
研究有限深度双扩散交换在测量数据简化后的可观测性，通过对受控交换历史用多种观测格式采样并与实际产品比较，发现孤立剖面识别路径能力弱，剖面束信息量更大，粗垂直采样和断面式采样各有特点，有用观测单元是集合或断面。
中文摘要 AI 辅助
双扩散界面可支持空间有序但观测稀疏的交换历史。解析三维计算包含标量梯度展宽、保持紧凑、与有限深度层的远程部分连接或水平组织的路径，而实地测量通常提供剖面、重复投放、自主浮标记录或水文断面。本文研究测量数据简化后有限深度双扩散交换的哪些与路径相关的特征仍可观测。将四个受控有限深度交换历史视为已知真值场，用垂直剖面、剖面束、垂直粗化、Argo 式和断面式观测格式进行采样。将所得可观测值与选定的冰系剖面仪、Argo 和 CCHDO/GO-SHIP 产品进行比较，以将合成测量置于实际观测环境中。孤立剖面是弱路径标识符，而剖面束信息量更大。粗垂直采样可在扭曲局部界面宽度估计的同时保留广泛的路径分离，断面式采样仅在站间距解析相关模式时才增加水平尺度信息。因此，隐藏的有限深度双扩散交换的有用观测单元是集合或断面，而非孤立投放。
英文摘要
Double-diffusive interfaces can support exchange histories that are spatially organized but only sparsely observed. A resolved three-dimensional calculation contains the route by which scalar gradients broaden, remain compact, connect with remote parts of a finite-depth layer, or organize horizontally, whereas field products usually provide profiles, repeated casts, autonomous-float records, or hydrographic sections. We ask which route-relevant features of finite-depth double-diffusive exchange remain observable after that measurement reduction. Four controlled finite-depth exchange histories are treated as known truth fields and sampled with vertical-profile, profile-bundle, vertical-coarsening, Argo-style, and section-like observing formats. The resulting observables are compared with selected Ice-Tethered Profiler, Argo, and CCHDO/GO-SHIP products to place the synthetic measurements in realistic observing contexts. Isolated profiles are weak route identifiers: at the final comparison time, strict case accuracy remains below 0.48, route-family accuracy is about 0.53--0.56 in leave-one-out profile classification, and 22 of 36 profiles are closer to a different-route profile than to a same-route profile under the current feature set. Profile bundles are substantially more informative. In the bundle-enumeration framework, the one-profile late-time route-family baseline is 0.694, while two-, three-, and four-profile bundles reach route-family accuracies of 0.917, 0.961, and 0.994. Coarse vertical sampling can preserve broad route separation while distorting local interface-width estimates, and section-like sampling adds horizontal-scale information only when station spacing resolves the relevant mode. The useful observing unit for hidden finite-depth double-diffusive exchange is therefore an ensemble or section, not an isolated cast.
