Debris flow disasters have increased in Taiwan due to various environmental factors. These disasters often bring a lot of rock and mud, causing a threat to the lives and properties of residents in the affected areas. The weather is changeable due to more and more extreme rainfall events. A monitoring system is needed to provide early-warning of debris flow disasters to reduce the loss of life and property. The number of installed precipitation stations is not adequate for the current early-warning system. Rainfall patterns are greatly affected by the variation in the topography. Therefore, the current system cannot fully integrate basin-wide rainfall data and lacks information on spatial dependency between rainfall stations. This paper proposes a watershed-based debris flow early-warning system that applies the OGC SWE standards to design its architecture. The standardized data exchange mechanism is used to integrate and share heterogeneous monitoring resources. A hierarchical architecture is proposed to build a wide range of precipitation stations. The system presents high density debris-flow-prone area monitoring. We propose dependency aggregation and SWE integration schemes that enable the system to collect data from upriver under dependency relationship of debris-flow-prone streams and achieve automated early-warning of debris flows. We use the SWE open source provided by 52North to implement the proposed watershed-based debris flow early-warning system. We develop a simulator using real rainfall data in Taiwan to compare to the current system. The experimental results demonstrate that our system can improve the sensing data problem and efficiently advance the warnings issue time.