Case Study 4
Uncertainty of Minimum Flow for west-central Florida Rivers
Case Study 3 (above) describes how the Integrated Northern Tampa Bay (INTB) Model was used by the Southwest Florida Water Management District (SWFWMD) to estimate an unimpacted or baseline streamflow condition. Streamflow MFLs are developed by the SWFWMD after determining a percent reduction of the baseline flow conditions based on environmental criteria associated with significant harm at critical habitat thresholds in the channel and floodplain. The baseline flow condition was established using one 11-year historical period. An allowable temporal percent reduction in baseline conditions is a key factor in the determination of significant harm which pertains to the long-term persistence of flow at critical habitat thresholds. The SWFWMD has established that an MFL event is at least six years long.
Tampa Bay Water (TBW) was interested in quantifying how the baseline flow condition and the MFL are affected by streamflow uncertainty which is influenced by the interaction among climate variability, the hydrologic system, and anthropogenic stresses. Specifically, TBW was interested in quantifying input uncertainty for streamflow due to rainfall uncertainty and well pumping. Rainfall uncertainty is characterized by many key statistics representing the magnitude, temporal pattern, and spatial correlation of rainfall and the co-variance among those statistics. The range of probable magnitude for persistence of flow is significantly influenced by and is an outcome of rainfall uncertainty. Therefore, rainfall uncertainty is directly relevant to defining a streamflow MFL because the MFL depends on persistence of flow.
TBW implemented a numerical experiment to define the sample size required to capture the full uncertainty (probable range in magnitude and persistence) of streamflow. The Monte Carlo (MC) numerical experiment included running the INTB model 1000 times each for two well pumping scenarios using unique 20-year rainfall events that were stochastically derived from 60 or more years of historical rainfall data. The two well pumping scenarios matched those used by the SWFWMD to define the baseline flow condition for the MFL process. The 1000 INTB simulations produced 1000 possible baseline flow conditions that are equally probable to occur in the future.
For an MFL duration event, analysis of the MC results indicates that the minimum sample size required to capture the full streamflow uncertainty is 250 of the 1000 simulations. This is the equivalent of 1500 years of observed data.
When flows from the single 11-year historical period were compared to the MC results showing full streamflow uncertainty, flows during the historical period: (a) fell below the MC inter-quartile range for lower flow magnitudes (% exceedance < 20%) and (b) were at or above the MC inter-quartile range for higher flow magnitudes (% exceedance < 5%). These results clearly confirmed that the 11-year historical period represented a period of both wet and dry extremes which was known to the investigators through an external assessment before the experiment was run.
The numerical experiment indicates the range of probable streamflow persistence at this location, a key factor which established the physical and biological conditions at critical habitat thresholds over a long period of time (i.e., century or more), was not represented within the single 11-year historical period. Without accounting for streamflow uncertainty, an unachievable flow MFL could be established or the water resource could be under-utilized.