Abstract
The variability of water chemistry on a daily scale is rarely addressed due to the lack of records. Appropriate tools, such as typologies and dimensionless indicators, which permit comparisons between stations and between river materials, are missing. Such tools are developed here for daily concentrations (C), specific fluxes or yields (Y) and specific river flow (q). The data set includes 128 longโterm daily records, for suspended particulate matter (SPM), total dissolved solids (TDS), dissolved and total nutrients, totalling 1236 years of records. These 86 river basins (10^3^โ10^6^ km^2^) cover a wide range of environmental conditions in semiโarid and temperate regions. The segmentationโtruncation of Cโq rating curves into two parts at median flows (q~50~) generates two exponents (b~50inf~ and b~50sup~) that are different for 66% of the analysed rating curves. After segmentation, the analysis of records results in the definition of nine major Cโq types combining concentrating, diluting or stable patterns, showing inflexions, chevron and U shapes. SPM and TDS are preferentially distributed among a few types, while dissolved and total nutrients are more widely distributed. Four dimensionless indicators of daily variability combine median (C~50~, Y~50~), extreme (C~99~, Y~99~) and flowโweighted (C*, Y*) concentrations and yields (e.g. C~99~/C~50~, Y*/Y~50~). They vary over two to four orders of magnitude in the analysed records, discriminating stations and river material. A second set of four indicators of relative variability [e.g. (Y*/Y~50~)/(q*/q~50~)], takes into account the daily flow variability, as expressed by q*/q~50~ and q~99~/q~50~, which also vary over multiple orders of magnitude. The truncated exponent b~50sup~ is used to describe fluxes at higher flows accounting for 75% (TDS) to 97% (SPM) of interannual fluxes. It ranges from โ 0ยท61 to + 1ยท86 in the database. It can be regarded as the key amplificator (positive b~50sup~) or reductor (negative b~50sup~) of concentrations or yields variability. C~50~, Y~50~, b~50sup~ can also be estimated in discrete surveys, which provides a new perspective for quantifying and mapping water quality variability at daily scale. Copyright ยฉ 2011 John Wiley & Sons, Ltd.