Since the late 1930's, spillway requirements for most reservoirs in the United States have been based on the probable maximum precipitation (PMP) developed by the Hydrometeorologic Section, Hydrologic Services Division of NOAA. Probable maximum precipitation is used to calculate the probable maximum flood (PMF). Probable maximum precipitation, found by maximizing storm characteristics like wind velocities and dew point, is usually much larger than the maximum observed historic storm rainfall in a watershed.
The regional NOAA Hydrometeorologic Reports specify storm characteristics, but they do not specify hydrologic processes within watersheds, the watershed state as the storm begins, or the PMP storm track and speed as it moves over a watershed. These factors can have a significant effect on the calculated PMF.
Hydrologic simulation, calibrated using the historic records in a watershed, represents the response of the land surface and soils, and the response of the channel network and floodplains to rainfall and flood flows. Simulation does not automatically 'solve' for the appropriate watershed state or storm track characteristics. It does, however, provide information that can be used to make the necessary choices. For example, PMP events can be simulated over a range of historical watershed states (initial conditions), and the sensitivity of the PMF to watershed state can be explored. Some of these sensitivities are not obvious; e.g. a very high initial snowpack as a PMP event begins can reduce the PMF by storing water within the snowpack. Alternate storm tracks can change PMF peak flows by 25 to 50 percent in larger watersheds.
PMF analysis requires choices about the simultaneous occurrence of independent rare events. The final choices that are made for a PMF require judgments: It is reasonable, given the PMP/PMF philosophy, to use the historical maximum watershed soil moisture at the time of year of the PMP. It is also reasonable to use an initial snowpack for that time of year that will maximize the PMF. It is unreasonable to assume a storm track and speed that matches and adds to the flood wave as it moves downstream in a large watershed, if this storm track and speed is unlike anything observed historically.
Hydrologic simulation for Probable Maximum conditions, or simulation of
Probable Maximum Floods, has been done for many reservoirs in the