Improved Grunsky method for streamflow prediction in ungauged catchments

This study proposes the R-Grunsky, a generalized method for streamflow prediction in ungauged catchments. Using data from 716 Brazilian catchments, we generalized the rainfall-streamflow coefficient (α) of the original framework based on mean annual temperature and precipitation. We established a relationship between α and catchments’ attributes, then fitted a multiple linear regression model to derive α values. We noted that the R-Grunsky approach provides suitable streamflow estimations, exhibiting a Kling-Gupta efficiency (KGE) of 0.90 on a mean annual scale, and KGE > 0.50 for nearly 13% catchments on a mean monthly scale. Our results are similar to those with the original Grunsky method, which fits an α value for each catchment. However, this new approach brings the advantage of needing only precipitation and mean temperature data to predict streamflow for any catchment. Hence, this methodology holds promise for diverse water resources projects, facilitating informed decision making regarding water security.