Water-quality data collected at 20 stations in Michigan were used to develop regression equations relating loads of dissolved and suspended substances to discharge. These equations and mean daily discharge were used to estimate long-term loads, which then were converted to estimates of drainage basin yields. These yields were compared to measured yields and to previous estimates.
Equations were developed for each station that express the relation between load and discharge (L = aQb) for 19 substances and properties. Fifty percent of the equations had standard errors of 22 percent or less; 90 percent of the equations had standard errors of 80 percent or less. Regression exponents indicate that the load increases as discharge increases in all cases, but for about two-thirds of the substances or properties, the increase is less rapid than that of discharge. Seventy-eight percent of the nitrogen, phosphorus, and sediment loads, taken as a group, increase more rapidly than discharge. Concentrations of about 63 percent of the substances or properties decrease as discharge increases, however.
Although comparative data for most drainage basins are scant, yields estimated by use of regression equations did not differ appreciably from measured values at locations where comparison was possible.
