Early Warning Systems-A Better Design Approach

During the late 1980's with the advent of reliable and inexpensive computerized sensing equipment, and the Bureau of Reclamation's realization (based on risk-based analysis) that many hydrologic dam safety problems could best be addressed by installing Early Warning Systems (EVS) to provide for public safety and minimize economic costs; management began to request EVS designs from Bureau engineers. In the Upper Colorado Regional Office during the first EVS design study, it became apparent that available conventional approaches to EVS design were basically inadequate. In designing an Early Warning System, key technical questions include: 1) What kind (raingage, reservoir-level, etc.; and accuracy) of sensors are required? 2) How many sensors are needed? 3) Where should the sensors be located? 4) For a particular EVS design, what are the system's performance characteristics with regard to: - efficiency in detecting critical events? - frequency of false alarms? - minimum warning time? 5) What specific EVS-activation criteria (e.g., rainfall level) should be used? Conventional approaches to EVS design do not rigorously address these design evaluation questions. As a result of the deficiencies in conventional approaches, in order to minimize over-design costs and under-design risks, the Regional Engineer in the Upper Colorado Region authorized development of a new EVS design approach which would provide answers to the above questions. The newly developed Upper Colorado approach can be used to evaluate EVS performance for both: a) dam-overtopping events, and b) hazardous spillway discharge flood events; and will independently define the minimum number of field stations required for detection of all dam-threatening events, or hazardous spillway-discharge events. This approach also independently defines false-alarm frequency and minimum warning time, for dam-overtopping events and hazardous spillway discharge events. Typically more field stations are required for detection of hazardous spillway-discharge events than for dam-overtopping events. Experience thus far has shown that with precipitation monitoring the optimal number of field stations required cannot be based primarily upon watershed area, which conventional approaches assume. Vith some dams it has been found that less than half of the number of stations prescribed by the conventional 'area-power rule', are actually required, with consequent savings in initial EYS installation costs and ongoing operation, maintenance and repair costs. Conversely, with other dams it has been found that more field stations are needed for a fully-effective system, than the number prescribed by the area-power rule. Primary EVS design evaluation computations with the new Upper Colorado approach, are computer programmed in the FORTRAN language and these programs can be run on any IBM-compatible Personal Computer. 18 pp., 4 references.