Evaluating the potential of an open sensor network to support reservoir pre-release decision making

Still in most countries, reservoir flood warnings are threshold-based alerts issued when water levels exceed thresholds. This current practice of releasing water from reservoirs causes flash floods in downstream areas and increases the likelihood of dam failures and public outrage. Pre-release of water from reservoirs is therefore an important strategy for downstream flood mitigation. Hydrological models can simulate river flow with sufficient lead time. Thus, the resulting outputs can be effectively applied to pre-release decision making in the reservoirs. Since the beginning of computer-aided applications, many attempts have been made to establish a decision support system for reservoir flood control. However, this is hampered by manual stations, low quality data, high cost of software and data, unknown parameter values, and lack of expertise, especially in developing countries. Therefore, a total open-source solution combined with low-cost open-source hardware, free and open-source software, and open standards was seen as the only way to overcome reservoir-related flood risk. Moreover, research studies on open-source hardware, software, standards, and data are limited to a few case studies reporting real-time data on certain environmental parameters. Therefore, the application of integrated open-source technologies for reservoir flood control remains an unexplored area. In this background, a hydrological model powered by integrated open-source technology is presented in this research for reservoir pre-release decision making. The model was tested for the Deduru Oya watershed using the SWAT (Soil and Water Assessment Tool) toolkit. The calibration results appear to be satisfactory for both daily and hourly time intervals. Thus, this model helps to simulate the inflow of the reservoir and determine the level of reservoir gate opening.