Optimal temporal resolution for hydrological modeling studies

Abstract

The impact of climate change on water resources has been an important field of research over the recent decades. Better water management is one way to adapt to a changing climate. To achieve this, having better rainfall-runoff models would be beneficial. Rainfall-runoff models are essential for improving our understanding and ability to forecast floods and droughts. Most hydrological modeling studies have been conducted at the daily-scale since most meteorological data was only available at this time step. However, higher temporal resolution data has become increasingly available at the sub-daily time scales. Previous work has shown that using data with a higher temporal resolution often results in an improvement in modelling accuracy. However, using a finer time scale increases the computational requirements of running the hydrological model. Finding the best time step is therefore an exercise in maximizing simulation accuracy while keeping the computational burden as low as possible. Smaller catchments may benefit more from a smaller time step due to the faster reaction time, whereas larger catchments may be relatively insensitive to the computation time-step. This highlights the importance of taking catchment size and shape into account when trying to determine the best temporal resolution. Even though there have been numerous studies on runoff models with sub-daily and even sub-hourly time steps, a systematic knowledge of how catchment size affects the time resolution option and how to determine the most efficient time step for the model has not yet been achieved. This work therefore investigated how the temporal resolution of a lumped hydrological model impacts simulation results, and if the impact is related to catchment size. Three-hundred and thirty-nine (339) catchments with quality-controlled hourly precipitation were selected covering most of the contiguous UnitedStates. Hourly meteorological data was aggregated at 2, 3, 4, 6, 12 and 24-hour time steps and one hydrological model was calibrated on all catchment and for each of the 8 time-steps. To study the impact of catchment size, the catchments were regrouped into 6 different size classes, from smaller than 500 km2 to larger than 4500 km2 , each group containing approximately the same number of catchments. Results showed that using as small a time-step as possible was beneficial to all catchment size classes, as it systematically improved the simulated flow bias as well as the magnitude and timing of peaks flows. The smallest size classes benefited the most from a reduced time step whereas for larger catchments, going to time step smaller than 8-hour only provided marginal improvements.

Date	26 Sept 2022
Original language	American English
Awarding Institution	École de technologie supérieure
Supervisor	François Brissette (Supervisor)