Transient calibration of a groundwater-flow model of Chimacum Creek Basin and vicinity, Jefferson County, Washington: a supplement to Scientific Investigations Report 2013-5160
A steady-state groundwater-flow model described in Scientific Investigations Report 2013-5160, ”Numerical Simulation of the Groundwater-Flow System in Chimacum Creek Basin and Vicinity, Jefferson County, Washington” was developed to evaluate potential future impacts of growth and of water-management strategies on water resources in the Chimacum Creek Basin. This supplement to that report describes the unsuccessful attempt to perform a calibration to transient conditions on the model. The modeled area is about 64 square miles on the Olympic Peninsula in northeastern Jefferson County, Washington. The geologic setting for the model area is that of unconsolidated deposits of glacial and interglacial origin typical of the Puget Sound Lowlands. The hydrogeologic units representing aquifers are Upper Aquifer (UA, roughly corresponding to recessional outwash) and Lower Aquifer (LA, roughly corresponding to advance outwash). Recharge from precipitation is the dominant source of water to the aquifer system; discharge is primarily to marine waters below sea level and to Chimacum Creek and its tributaries. The model is comprised of a grid of 245 columns and 313 rows; cells are a uniform 200 feet per side. There are six model layers, each representing one hydrogeologic unit: (1) Upper Confining unit (UC); (2) Upper Aquifer unit (UA); (3) Middle Confining unit (MC); (4) Lower Aquifer unit (LA); (5) Lower Confining unit (LC); and (6) Bedrock unit (OE). The transient simulation period (October 1994–September 2009) was divided into 180 monthly stress periods to represent temporal variations in recharge, discharge, and storage. An attempt to calibrate the model to transient conditions was unsuccessful due to instabilities stemming from oscillations in groundwater discharge to and recharge from streamflow in Chimacum Creek. The model as calibrated to transient conditions has mean residuals and standard errors of 0.06 ft ±0.45 feet for groundwater levels and 0.48 ± 0.06 cubic feet per second for flows. Although the expected seasonal trends were observed in model results, the typical observed annual variation of groundwater levels of about 2 feet was not. Streamflow at the most downstream observation point was about three times larger than simulated streamflow. Because the transient version of the model proved inherently unstable, it was not used to simulate forecast conditions for alternate hydrologic or anthropogenic changes. Adaptation of alternate stream simulation packages, such as RIV, or newer versions of MODFLOW, such as MODFLOW-NWT, could possibly assist with achieving calibration to transient conditions.