On the value of long-term, low-frequency water quality sampling: avoiding throwing the baby out with the bathwater
On the value of long-term, low-frequency water quality sampling: avoiding throwing the baby out with the bathwater
复制标题
论长期、低频水质采样的价值：避免把婴儿和洗澡水一起倒掉
DOI：
10.1002/hyp.7961
复制
发表时间：
2011
影响因子：
3.2
通讯作者：
Burt T
中科院分区：
文献类型：
--
作者：
Burt T
高频率的水质采样是风靡一时的，这是正确的。Kirchner等人（2004）展示了精细数据如何为数据结构开辟新的见解。他们指出，新的水质采样技术已经改变了，并将继续改变，我们的看法集水过程，使我们能够在时间分辨率，比以前更精细的数量级进行观察。化学行为的高频测量肯定会为流域水文学中的许多关键问题提供新的见解，并将推动新的数据分析和建模方法。这将反过来推动新的实地实验方法。而Kirchner et al.（2004年）的评论具有很大的影响力，并加速了人们对新的水化学特征的兴趣，我们担心，一种解释可能是，研究过去的低频数据几乎没有什么收获，我们可能已经偏离了对这些记录的研究。在这里，我们探索低频水质数据的价值，并提出一个修辞性的问题：它们仍然有用吗？我们这样做是因为我们担心许多低频长期数据集会被放弃，以及人们对此类记录价值的普遍态度转变。我们在此的目的只是为了纠正这种平衡，并指出对长期水质时间序列的研究仍然富有成效。此外，由于需要长时间的记录来评估集水区响应的长期变化，因此不注意这些记录似乎是疏忽，特别是因为最近的工作表明，这些数据可能包含罕见事件或系统行为非常快速变化的证据（Burt，1994年; Burt等人，2010 a;豪登等人，2010年），表明长期趋势（豪登和伯特，2008年，2009年），并设置了可以解释较短记录的背景（伯特等人，2008年）。这最后一点是至关重要的，因为短期集水研究（即< 10年）往往受到年际水文气候变化的高度影响，因此长期趋势往往被掩盖。然而，我们在这篇评论中的主要观点是，我们不能等待数年或数十年的时间来积累新类型的高频数据，因为可以很好地科学利用现有来源。在数据记录仪和可现场部署的分析仪器出现之前，高频采样在现场以及在实验室分析大量水样中花费了更多的时间和精力。这并不是说从未获得过高频数据，而是因为它们并不常见，一个例子可以说明这一点，它是世界各地许多此类记录的象征，这是英格兰东部大乌塞河50年（1957-2007）河水硝酸盐浓度和通量的记录。从1933年1月1日起，可获得从15分钟值得出的日流量（NRFA仪表33002;集水面积1460平方公里），更相关的是，从1957年起，同一地点的硝酸盐浓度可获得（第一个样本1957年1月2日，n= 4250）。采样频率平均每年83次采样;然而，记录受益于1980年代中期非常频繁的采样，在记录的其余部分大约每周采样一次。
High-frequency water quality sampling is all the rage, and rightly so. Kirchner et al.(2004) show how fine-scale data open up new insights into data structures. They note that new water quality sampling technologies have transformed, and will continue to transform, our view of catchment processes by allowing us to make observations at temporal resolutions that are orders of magnitude finer than before. High-frequency measurements of chemical behaviour will certainly yield novel insights into many key questions in catchment hydrology and will compel new approaches to data analysis and modelling. This will in turn drive new approaches to field experimentation. While the Kirchner et al.(2004) commentary has been highly influential and accelerated interest in new hydrochemical signatures, we fear that one interpretation may be that there is little to be gained from studying past low-frequency data and that we may have been diverted from examining such records. Here we explore the value of low-frequency water quality data and ask the rhetorical question: are they still useful? We do this because we fear abandonment of many low-frequency long-term datasets and a general attitudinal shift regarding the value of such records. Our purpose here is simply to redress the balance and to point out that examination of long water quality time series remains fruitful. Furthermore, as long records are required to assess long-term changes in catchment response, it would seem negligent to pay no attention to them, especially because recent work has shown that such data may contain evidence of rare events or of very rapid shifts in system behaviour (Burt, 1994; Burt et al., 2010a; Howden et al., 2010), indicate long-term trends (Howden and Burt, 2008, 2009) and set the context within which shorter records can be interpreted (Burt et al., 2008). This last point is crucial because shortterm catchment studies (ie< 10 years) tend to be highly influenced by inter-annual hydroclimatic variability, such that long-term trends are often obscured. However, our main point in this commentary is that we cannot afford to wait years or decades for new types of high-frequency data to accumulate when good scientific use can be made of existing sources. Before the advent of data loggers and field-deployable analytical instrumentation, high-frequency sampling cost much more time and effort in the field, and in laboratory analysis of large numbers of water samples. It is not that high-frequency data were never obtained, but that they were uncommon.One example to make the point, that is emblematic of many such records around the world, is the 50-year (1957–2007) record of river water nitrate concentrations and flux for the Great Ouse river in eastern England. Gauged daily flows derived from 15-min values are available from 1 January 1933 (NRFA gauge 33002; catchment area 1460 km2) and, more relevant here, nitrate concentrations are available for the same site from 1957 (first sample 2 January 1957, n= 4250). Sample frequency averages 83 samples per year; however, the record benefits from very frequent sampling during the mid-1980s, with approximately weekly samples for the remainder of the record.