The scientists studied streamflow in the Sapucaí River watershed in Brazil to improve water management. They used data from 1989 to 2014 to model streamflow patterns and determine reference streamflow. They found that using the drainage area and streamflow equivalent to total rainfall helped predict streamflow accurately. Linear and non-linear regressions were both effective in regionalizing streamflow, but non-linear regressions with specific predictor variables showed a slight advantage in accuracy.