Temporal variability of water chemistry in flowing waters of the northeastern United States: does river size matter?

The seasonal variability of discharge, water chemistry, suspended sediment, and suspended algal abundance was examined in temperate rivers of the nort heastern United States to evaluate the accepted (but rarely tested) assumpt ion that streams are more temporally variable than large rivers. Temporal v ariance of discharge and water-column characteristics was determined for 10 8 stations on 98 rivers with data measured between May to October of 1978 b y the US National Stream Quality Accounting Network (NASQAN). Temporal vari ance of water-column characteristics increased in relation to the mean esti mate of the variable (r(2) = 0.69-0.93, p < 0.001) with the exception of pH (p > 0.05). River size, estimated by median discharge, had little effect ( partial r(2) = 0.01-0.02, p < 0.05) or no effect (p > 0.05) on the temporal variance of water-column characteristics after accounting for differences in the mean. In contrast, hydrological variability (measured by the coeffic ient of variation of daily discharge, flood frequency, number of months wit h floods) was negatively related with median discharge (r(2) = 0.25-0.38, p < 0.001). When rivers were separated into regional drainage basins, hydrol ogical variability was strongly related to median discharge in some regions (r(2) = 0.45-0.76, p < 0.001) and not in others (p > 0.05). Temporal varia nce of water-column characteristics was not affected by river size within i ndividual drainage basins. The intensity of upstream impoundment (measured by the number of impoundments and total impoundment storage volume) also ne gatively influenced the seasonal variability of discharge (r(2) = 0.08-0.24 , p < 0.01) but had little effect (partial r(2) = 0.01-0.03, p < 0.05) or n o effect (p > 0.05) on the temporal variance of water-column characteristic s. Power functions are provided as guidelines for sampling design to estima te the number of samples required to obtain a mean with a given level of pr ecision for each of the water-column characteristics. Nutrient status may b e more important to consider than river size when determining sampling freq uency for monitoring programs. On average, water chemistry is as variable i n large rivers as in small ones.