Cross-catchment comparison of dominant hydrometeorological controls over streamflow inferred through information theory

Abstract:

Process-based hydrological studies provide critical insights on water resources management and model development. Comparing dominant hydrological processes for streamflow generation across catchments is especially crucial in model generalization and prediction in ungauged basins (PUB). Here we develop a hydrological ‘fingerprint’ that maps information flows from dominant hydrometeorological controls to streamflow over a range of aggregation and lag timescales. Applied to multiple catchments across the U.S., he hydrological ‘fingerprint’ enables rapid visualization of the dominant processes and timescales controlling the discharge, providing a basis for process-based classification of watersheds. Preliminary results at the Dry Creek watershed, Idaho, revealed that daily streamflow is dominated by daily precipitation, while the influence of aggregated soil moisture and snowmelt persists up to weeks and months, respectively. By applying a similar analysis to 30 intensively monitored watersheds that include CZO and LTER sites, we target a comparative catchment classification and regionalization scheme to help reduce model uncertainty and enable PUB.