By Thomas H. Yorke2, Jérôme Le Coz3 and Tony Bostin4
The availability of adequate fresh water for human consumption, agriculture, industries, cooling of industrial and electric generating facilities, and aquatic and riparian resources is critical, or will become so, in all countries. Accurate and verifiable streamflow data are essential for making confident estimates of available fresh water, for designing roads, bridges and other infrastructure, for calibrating and validating hydrological models, and for assessing the potential impact of floods. The quality of that data is ever more important as resources become more limited and water allocation decisions more impactful on local, regional and national socio-economic development.
“Communicating Hydrometric Data Quality: What, How, and Why”4 provides a good synopsis of the current state of how streamflow data are collected, processed, reviewed and categorized. It describes the old paradigm of national organizations that had standard operating procedures for collecting and validating streamflow data as accurate and verifiable, then looks at the new paradigm of many disparate providers sharing data using the WaterML 2.0 standard5, which offers no complementary standards for characterizing the quality and reliability of the data. WaterML 2.0 makes the need for transparency in hydrometric procedures – external audits by independent organizations, in addition to internal procedures – more acute than ever, but this emerging issue is not addressed in the ebook.
The author discusses the codes used for communicating data quality by five organizations: WMO, U.S. Geological Survey (USGS), Water Survey of Canada, National Environmental Monitoring Standards (NEMS) of New Zealand, and the Open Geospatial Consortium (OGC). This includes a rudimentary comparison of the data quality categories of good, fair, poor, estimated, unchecked and missing. Obviously, it would be relevant to extend the exercise to more organizations worldwide.
The ebook section “Standards for Characterizing Data Quality” is very informative and includes numerous links to other publications, including those of the International Standards Organization, USGS, and WMO. It contains information on various aspects of stream gauging that affects data quality, including site selection; water-level measurement methods and equipment; discharge measurement methods, technologies and equipment deployment; and computational procedures using the stage-discharge ratings, velocity-index ratings, and modeling and other estimation techniques.
“Categorization of Hydrometric Data Quality” is the core of the ebook. It identifies NEMS as the only agency reviewed that has a standard with testable criteria for data quality categorization. The section has two examples, including flow charts, showing the performance objectives for water level data and streamflow data. The flow charts include decision nodes for defining six data quality zones, which provides a good example of the need for more standardized practices in the production and delivery of hydrometric data.
A missing component of the ebook is information on etermining the quantitative uncertainty of streamflow data. NEMS data quality zones are based on decision nodes of calculated uncertainty, however, the book stops short of providing guidance on calculating uncertainty. Perhaps calculating uncertainty is beyond its scope, but it would have been useful for the author to point to procedures and examples for calculating the uncertainty of streamflow data available in ISO 748 and ISO 5168. The ebook properly makes reference to the GUM (JCGM, 2008) as the general uncertainty analysis framework.
Uncertainty analysis of hydrometric data is complex and difficult to achieve routinely and in a reproducible, auditable way. However, it is necessary to make progress on the quantification of uncertainty, according to an explicit level of confidence (usually 95%), because the qualitative grading of data through a quality code is unsufficient to assess whether data are fit for a given use. Unlike quantified uncertainties, quality codes cannot be propagated to hydrological statistics or other hydrological products.