Accurate projections of flooding are critical to climate change adaptation, as floods are a
devastating and costly natural disaster sensitive to climate. Floods occur mainly in
response to extreme precipitation, and the northeastern United States has experienced a
dramatic increase in extreme precipitation over the past twenty-five years. However, the
drivers of floods in the Northeast, as well as the effects of climate change on those
drivers, are poorly understood. Here, we use a random forest model of streamflow forced
with a regional climate model to examine the historical and future streamflow dynamics
of four watersheds across a range of latitudes in the Northeast. We find that streamflow
in the cold season (November-May) is primarily driven by 3-day rainfall in three
watersheds, and 30-day rainfall in the fourth. In the warm season (June-October),
streamflow is primarily driven by antecedent soil moisture in all watersheds, and 3-day
rainfall in three watersheds. By the end of the century (2070-2099), cold season
streamflow increases across all four rivers (31.4% – 54.9%) in response to enhanced total
and extreme precipitation. Warm season streamflow also increases in two watersheds
(56.3% and 184.5%) due to increased precipitation and antecedent soil moisture but
decreases in the other two (-5.7% and -27.4%) because of reduced precipitation. This
analysis contributes to the understanding of both projected changes in flooding as well as
the drivers of those changes across a range of watersheds characteristics and flow
regimes in the Northeast, which is essential for reducing future flood risk.