A Practical, Objective, and Robust Technique to Directly Estimate Catchment Response Time

Methodologies to estimate the response time of a catchment to new rainfall inputs based on rainfall and streamflow observations require the analyst to make a number of uncertain and subjective steps. Moreover, these methods make the assumption that the water producing the discharge peak fell in the last rainfall event, which does not necessary apply to all the environments and conditions. Hence, here we present a practical, objective, and robust method to estimate catchment response time (Tr) based on hourly rainfall and streamflow time series only, which removes most of the sources of uncertainties arising from current methodologies by restating the conceptual hypothesis and minimizing the user’s choices. The proposed method, used originally in the field of economics to assess the temporal correlation between two variables, has been adapted to be used for the first time in the field of hydrology. The method does not make any assumption about the rainfall‐runoff transformation (no hydrograph separation needed), does not require event selection or parameter estimation, and it is easily reproducible. The above features make the proposed method a useful tool even under different hypothesis regarding the hydrograph water age. The method agrees well with the traditionally used method to estimate Tr from observed hyetographs and hydrographs (Spearman rank correlation r = 0.82). The proposed method gives robust results for relatively short records, and works in presence of noise and bias in the time series. Plain Language Summary Methodologies to estimate the time delay of the transformation of rainfall into river discharge based on rainfall and discharge records require a number of highly subjective and uncertain steps. Moreover, the assumptions behind these methods have been proven incorrect, at least in some environments. For this reason, we present a different method which removes those incorrect assumptions and most of the sources of uncertainties arising from the other methodologies. Unlike existing methods, the proposed methodology does not make any assumption about the processes that transform rainfall into river discharge, does not require event identification or parameter estimation and is easily reproducible. We demonstrate that the new approach compares well with the traditionally used method and also works for short and noisy records. Key Points The median catchment response time (Tr) computed with the proposed method matches the median Tr computed w