2.1 Forcing data
Precipitation is the major driver of local or regional flooding, and it is thus central to acquire an accurate and high-resolution dataset. In the U.S., the Multi-Radar Multi-Sensor (MRMS) precipitation product, developed at the National Severe Storms Laboratory (NSSL), provides 2-min and 1-km rainfall field, making it suitable for flash flood forecasting (Yussouf et al., 2016). It integrates ~180 WSR-88D operational radars, creating a seamless radar mosaic across the CONUS and southern Canada. Recent studies (e.g., Chen et al., 2020; Li et al., 2020) verified the efficacy of MRMS data when compared to gauge-based and satellite-based products during the Hurricane Harvey event. The advantage of using radar rainfall is obvious for flood inundation modeling, as conventional rain gauges cannot readily represent the spatially variable rainfall fields. The MRMS data was downloaded at https://mtarchive.geol.iastate.edu/.
Besides precipitation data, potential evapotranspiration (PET) is a major input into a hydrologic system. In this real-case study, we obtain the PET data from the USGS FEWS data port (https://earlywarning.usgs.gov/fews) at daily temporal and 1° spatial resolution (Allen, Pereira, Raes, & Smith, 1998).