Uncertainty of water-quality predictions in ungauged basins (PUBS)

Abstract The quality of freshwater at any point on the landscape reflects the combined effects of natural and anthropogenic processes along hydrological pathways, for which the process relations may be cyclical (inter-dependent and/or periodic) and cascading. Local, regional, and global differences in human activities (e.g. mining, industry, agriculture, waste treatment and disposal), climate, and streamflow, are considerable, which, in turn, cause varying effects on water quality and quantity. Natural characteristics also greatly control human activities, which then affect the natural composition of water. These cause-and-effect relations produce uncertainty in water-quality estimates, because they are difficult to quantify. For example, although rock type is a primary control on water quality, annual dissolved major-ion yield estimates from empirical relations using rock type, precipitation quantity, population density, and temperature for large rivers in the US, are precise only to one order of magnitude, which in part is due to nonlinear responses of large non-homogeneous areas. Human activities, as reflected by land-use, can likewise be used to qualitatively indicate differences in constituent concentra-tions, particularly for constituents that are primarily controlled by non-point sources, e.g. nutrients and pesticides. However, the large concentration varia-tions prevent an accurate estimate. For many naturally occurring substances, uncertainty of estimated concentration or yield is larger for constituents primarily derived from point sources than from non-point sources, until the major point sources and related contributions are known. Also, uncertainty of the estimates of average constituent concentration or yield decreases with increasing averaging time, e.g. a predicted annual flux is more precise than an instantaneous or daily flu