Abstract
In situ turbidity meters are being increasingly used to generate continuous records of suspended sediment concentration in rivers. However, the usefulness of the information obtained depends heavily on the existence of a close relationship between fluctuations in suspended sediment concentration and turbidity and the calibration procedure that relates suspended sediment concentration to the turbidity meter's signal. This study assesses the relationship between suspended sediment concentration and turbidity for a small (1·19 km^2^) rural catchment in southern Brazil and evaluates two calibration methods by comparing the estimates of suspended sediment concentration obtained from the calibrated turbidity readings with direct measurements obtained using a USDH 48 suspended sediment sampler. With the first calibration method, the calibration relationship is derived by relating the turbidity readings to simultaneous measurements of concentration obtained from suspended sediment samples collected from the vicinity of the turbidity probe during flood events. With the second method, the calibration is based on the readings obtained from the turbidity meter when the probe immersed in samples of known concentration prepared using soils collected from the catchment. Overall, there was a close link between fluctuations in suspended sediment concentration and turbidity in the stream at the outlet of the catchment, and the estimates of sediment concentration obtained using the first calibration method corresponded closely with the conventionally measured sediment concentrations. However, use of the second calibration method introduced appreciable errors. When the estimated sediment concentrations were compared with the measured values, the mean errors were ± 122 mg l^−1^ and + 601 mg l^−1^ for the first and second calibration procedures respectively. Copyright © 2007 John Wiley & Sons, Ltd.