Modeling 63 years of salinity condition in Galveston Bay, Texas: From historic to recent drought events

Background/Question/Methods

Texas is prone to multi-year droughts interrupted by watershed-drenching episodes of rainfall.  Texas estuaries, lying at the receiving end of these catchments, are affected by subsequent highs and lows of surface inflows draining from inland watersheds which directly impact salinity condition, nutrient loading, and a variety of estuarine attributes.  Historic records provide some measured insights into salinity condition and species responses at particular locations and times, but broader insights into whole-bay patterns of salinity over extended periods do not exist ... until now.  That is, until we developed a 63-year simulation of salinity condition in Galveston Bay using the Texas Water Development Board’s TxBLEND hydrodynamic and salinity transport model.  For this, we compiled input for river hydrology, tidal elevation, precipitation, evaporation, and salinity boundary condition and developed model grids representative of changing bathymetric conditions over a 63-year period from 1950 - 2012.  This exploratory effort resulted in a modeling tool for examining the impact of long-term patterns of freshwater inflow, as a result of drought and water-planning activities, on salinity condition in Galveston Bay.  The simulation period allowed for examination of salinity patterns from Texas’ historic drought of record in the 1950s to the recent, ongoing drought of today.

Results/Conclusions

TxBLEND simulates salinity condition, water level, and circulation in estuaries and has been calibrated and validated for velocity, water level, and salinity over the period 1987 – 2005.  However, historic salinity records were critical for evaluating model performance for 1950 – 1986 and 2006 – 2012.  Fortunately, nearly 6,000 independent measurements of salinity, in addition to long-term monitoring records from 13 stations, were available for model validation.  Model performance was evaluated by r², root mean square error, Nash-Sutcliffe efficiency criterion, and percent bias.  Overall, TxBLEND captured well the long-term rise and fall of salinity in response to periods of drought and high inflows, but tended towards overprediction during the earlier period and in upper estuary locations.  The model also failed to accurately capture short-term, high frequency variability in salinity.  Intended to be a proof of concept, results from this multi-decadal simulation of salinity condition will be presented and discussed with respect to future applications which also must consider model performance and the need to improve model validation through further refinement of TxBLEND and acquisition of additional salinity data, particularly during the drought of the 1950’s.  The report (Guthrie et al. 2014) is available here:  www.twdb.texas.gov/surfacewater/bays/major_estuaries/trinity_san_jacinto/index.asp.