Due to their persistent severe winds, derechos threaten human security and properties, and they are as hazardous and fatal as many tornadoes and hurricanes. However, the automatic detection of derechos is challenging due to the complex criteria used to define the phenomenon and the need for spatiotemporally continuous observations. This study proposes a compromised definition of derechos, which not only contains the key features of derechos described in the literature but also allows their automatic identification using either observations or model simulations. The automatic detection is composed of three algorithms: the Flexible Object Tracker algorithm to track mesoscale convective systems (MCSs), a semantic segmentation convolutional neural network to identify bow echoes, and a new algorithm to classify MCSs as derechos or non-derecho events. Using the automatic detection approach, we develop a novel high-resolution (4 km and hourly) observational dataset of derechos over the United States east of the Rocky Mountains from 2004 to 2021. We investigate the derecho climatology in the United States based on the dataset. Many more derechos are identified in the dataset (~31 events per year) than in previous estimations (~6–21 events per year); however, the spatial distribution and seasonal variation patterns resemble earlier studies with a peak occurrence in the Great Plains and Midwest during the warm season. In addition, derechos produced around 20 % of damaging gust (≥ 25.93 m s^-1) reports during the study period in the central and eastern United States.