Monthly and Annual Water Budgets of Interstate Watersheds in the Delmarva Peninsula

Abstract

With climate change expected to change the hydrosphere globally, it is becoming more important for communities to manage water supplies. Water budgets can help these communities by keeping track of the inputs and outputs of water in a watershed. This study uses two datasets that combine models and observations to develop monthly and annual water budgets for watersheds of interstate Chesapeake Bay tributaries in the Delmarva Peninsula. CONUS404 is a novel dataset that reports hydroclimatic variables on a fine spatiotemporal resolution, while TerraClimate is a similar modeled dataset but is less comprehensive in its modeling and less refined in its temporal resolution. Analysis of these water budgets and their derived datasets leads to several findings. The modeled datasets are consistent with the general trends in stream gage discharge data but do not replicate the observed data recorded at the gages. The monthly and annual water budgets show that the water budget is balanced using TerraClimate data, but there is a nearly systematic bias for water budgets using CONUS404. Mann-Kendall testing of annual precipitation time series lends credence to an increasing trend in TerraClimate precipitation across all watersheds, indicating the presence of nonstationarities. However, testing of CONUS404 precipitation is less conclusive. Water years were split into terciles—dry years, average years, and wet years—based on precipitation, and analysis of water budget components based on these terciles reveals an inverse relationship between the distribution of water leaving the watershed as evapotranspiration and runoff. In dry years, a larger fraction of the incoming precipitation goes towards evapotranspiration as outflow, but in wet years, the fraction going to evapotranspiration decreases while the fraction going to runoff increases. Water storage is depleted in dry years, while wet years tend to experience an increase in storage. Further, Markov chain analysis was conducted on the annual precipitation terciles, revealing a slight tendency in TerraClimate terciles to be followed by their respective types, though results are less certain for CONUS404. Water years were also split into terciles based on the change in water storage, ΔS. Markov chain analysis of these terciles shows that years experiencing a high depletion of water storage are generally followed by a year with high replenishment of water storage and vice versa. Both reanalysis datasets may not be suitable for water budget studies in the Delmarva Peninsula due to the region’s unique geology. The relatively simple subsurface models used in both TerraClimate and CONUS404 likely fail to replicate the peninsula’s hydrogeology.