Are precipitation and antecedent wetness enough for accurate flooding prediction?
Date:
Flooding is a major hazard worldwide and is often initiated by, or amplified by, the degree of water stored in a watershed, particularly soil moisture and groundwater. Precipitation and/or antecedent wetness are commonly used to represent the degree of watershed storage in flooding prediction. There has been no consensus on a metric for antecedent wetness as it could be accumulated precipitation/streamflow or current soil moisture/water table depth which are often hard to estimate and observe. Here, we show modulations on the magnitude of streamflow by snow and evapotranspiration (ET), which are not commonly considered in flooding prediction. We conducted integrated hydrologic modeling in the US Mid-Atlantic basin driven by the meteorologic forcing of water year 2021. Interestingly, hurricane Ida in summer with the heaviest precipitation didn’t induce the annual maximum of streamflow. The generally higher magnitudes of streamflow occurred in winter with more-intermittent, less intense precipitation. In this case, the accumulated precipitation is a less valid choice to represent antecedent condition for flooding prediction. A shallower water table is much more representative of the wetter antecedent conditions during winter for flooding generation. A water balance analysis showed the deeper water table in summer was caused by stronger ET. Yet a precipitation event in winter with weak ET also didn’t induce a significant local maximum of streamflow as the precipitation was snow dominated. Additionally, we conducted global warming scenarios by increasing the air temperature 1, 2, and 4 ℃. Peak streamflow was largely dampened with warming due to the increased ET, yet the combined effects of increased snowmelt and ET resulted in an increased baseflow in winter (December to February). This increase is highly nonlinear with warming. In the three scenarios, 4℃ warming showed a minimum increase of streamflow during early winter due to the least snow and a maximum increase during late winter due to the fastest snow melting. Our results highlight the importance of including of ET and snow processes in flooding/streamflow prediction, even in more humid regions.