Algorithmic Characterization of Lake Stratification and Deep Chlorophyll Layers From Depth Profiling Water Quality Data
Algorithmic Characterization of Lake Stratification and Deep Chlorophyll Layers From Depth Profiling Water Quality Data
复制标题
DOI：
10.1029/2018wr023975
复制
发表时间：
2019-05
影响因子：
5.4
通讯作者：
Wenzhao Xu;P. Collingsworth;B. Minsker
中科院分区：
文献类型：
--
作者：
Wenzhao Xu;P. Collingsworth;B. Minsker
我们开发和测试算法，用于快速和一致地分析水质剖面数据，如温度和荧光，用于识别湖泊热分层和深叶绿素层（DCL）。目前，剖面数据的处理和关键特征的识别是人工和主观的，因此，不同采样事件的结果不具有可比性。在这项研究中，我们开发了一种方法，使用分段线性表示算法来近似线性段的垂直温度分布，从中可以提取分层模式。我们还提出了一种自动峰值检测算法来识别DCL的位置和大小。这些算法被应用于美国环境保护署大湖国家项目办公室收集的水质剖面数据，该办公室每年在大湖的固定地点使用电导率、温度和深度剖面仪进行深度剖面分析。算法产生的结果与人类判断相似，但有一些异常值表明专家错误、算法限制和定义层的模糊性。我们还展示了算法如何分析温度和荧光轮廓的形状以检测不寻常的模式。以苏必利尔湖为例，揭示了春夏季温跃层、DCL和储热变化的时空变化趋势。结果表明，湖的东部盆地储存了更多的热量。本文提出的方法将有助于充分利用历史深度剖面数据，并通过提供一致的方法使未来的采样过程受益。
We develop and test algorithms for rapidly and consistently analyzing water quality profile data such as temperature and fluorescence that are used to identify lake thermostratification and deep chlorophyll layers (DCL). Currently, the processing of profile data and identification of key features are manual and subjective, and thus, the results are not comparable from one sampling event to another. In this study, we develop a method to approximate vertical temperature profiles with linear segments using a piecewise linear representation algorithm, from which stratification patterns can be extracted. We also propose an automated peak detection algorithm to identify the location and magnitude of DCL. The algorithms are applied to water quality profile data collected by the United States Environmental Protection Agency Great Lakes National Program Office, which conducts annual depth profiling using conductivity, temperature, depth profilers at fixed locations in the Great Lakes. The algorithms generate similar results to human judgments, with some outliers that show expert errors, algorithm limitations, and ambiguities in defining layers. We also show how the algorithms can analyze the shape of temperature and fluorescence profiles to detect unusual patterns. Lake Superior is used as a case study to reveal spatial and temporal trends of the thermocline, DCL, and the heat storage change from spring to summer. The results reveal that more heat was stored in the eastern basin of the lake. The methods proposed here will help take full advantage of historical depth profiling data and benefit future sampling processes by providing a consistent method.