PRELIMINARY DRAFT - Precipitation Runoff Modeling System (PRMS) Watershed Model for the Pomperaug River, CT by USGS - November, 2005

The following is an EDITED version of the Summary and Conclusions of the PRELIMINARY DRAFT of the Phase 2 USGS PRMS report. Edits made by PRMS are included in italics.

THE PRELIMINARY DRAFT OF THE SUMMARY AND CONCLUSIONS OF THE USGS TECHNICAL REPORT ARE PRESENTED HERE TO DEMONSTRATE THE NATURE AND STATUS OF THE RESEARCH AND THE RESULTS TO DATE. THE FINAL VERSION OF THE USGS REPORT WILL BE POSTED ON THE COALITION'S WEBSITE UPON COMPLETION OF THE REPORT BY THE USGS.

Summary and Conclusions
The PRMS parameterization strategy developed for the Pomperaug watershed employs a minimum of calibration and optimization by establishing basic relations between the parameter value and physical characteristics of the individual HRU (hydrologic response units) that can be obtained from GIS (Geographic Information Systems) and other general databases for Connecticut. This approach is essential to successful application of PRMS in that it would be impossible to objectively optimize and calibrate all of the unknown (and un-measurable) parameters for each HRU. Additionally, it is important to have confidence that the parameterization process can be objectively repeated in other watersheds and would be relevant to past and future scenarios with changed or different physical watershed conditions. This latter condition is the basis for applying the model to simulate and evaluate the possible effects of various land use and water management strategies on streamflow and hydrology in the watershed.

The two subsurface routing parameters that required optimization could be isolated and calibrated because the groundwater and surface runoff parameters were determined independently based on a priori knowledge ands assumptions. Even though the general value of the parameters were not directly determined from physical characteristics of the watershed the parameter values were distributed and assigned to each HRU based on physical setting of each HRU, thus the subsurface parameters are considered to be secondarily determined objectively.

The PRMS simulations, for the 22-year record from 1976 to 1998, and the 5-year record from 1998 to 2003 both exhibit accurate long-term water balance and runoff characteristics. Additionally, the runoff variability both spatially and temporarily is accurately simulated. This indicates that the precipitation parameterization method and successfully distribute measured precipitation from different input data sets across the watershed and that evapotranspiration is also accurately simulated. Although the day-to-day runoff simulations are acceptable from the modeling standpoint, the simulated daily runoff may have considerable error compared to the observed daily runoff. Thus, the model simulations can provide accurate representations of the hydrologic conditions and flow regime, both temporally and spatially distributed, but must be used with caution when considering the simulated runoff for a particular day. Total monthly and annual runoff estimates may also have error of 10 to 20%, but also show very good long-term accuracy within 5%.

Because the long-term water balance and runoff simulations are accurate, the estimates of ground-water recharge made by PRMS are also considered to be accurate in the long term. The dynamics of the ground-water systems can generally be considered to operate on much longer time frames than the surface dynamics, and thus daily error in the simulated recharge values are of less importance when evaluating the effects of watershed changes on the groundwater.

The following summarizes the conclusions of the PRMS modeling for the Pomperaug River watershed:

• Model parameterization is physically based and transferable to other watersheds.
• Simplifying assumptions are applicable to Connecticut at large (with the possible exception of the Connecticut River Valley) - this needs to be evaluated by developing and calibrating models for representative basins throughout the state.
• Model complexity allows for a broad range of physical settings to be assessed within the model framework.
• Climate parameterization developed by USGS works well in Connecticut.
• Model structure and HRU distribution provides good representation of ground water and surface water complexity and interaction.
• Seasonal variability of streamflow is represented well in the model simulations.
• Simulation of extreme low flows should be improved using MODFLOW.
• Objective calibration of land-surface and ground-water process modules meets project needs.
• Can expect long-term discharge estimation to average within 5% of actual values.
• Estimated average recharge to stratified drift aquifers is on the order of 22 inch3es, which compares well with estimates developed by Mazzaferro (1986) and Starn (2002) of 22 to 24 inches.
• Estimated average recharge to till, approximately 11 inches, is somewhat larger than that estimated by Mazzaferro (1986) and Starn (2002), 8 to 9 inches, suggesting that till aquifers may provide more sustained baseflow than currently considered.

Management Scenarios and Simulation Data Sets
The calibrated PRMS model was used to generate simulation data sets of streamflow and component runoff for each HRU, based on a set of land and water use management scenarios outlined by the PRWC. The management scenario simulation runs are intended for use by PRWC and local town governments to evaluate the temporal and spatial hydrologic effects of different water and land use options such that they can make scientifically based watershed management decisions. The management issues include water availability (both ground water and instream). Source water variability and runoff rates (as it pertains to water quality) and effect of decisions on extreme high and low flow conditions.

Three land use scenarios were run including current conditions (2002), pre-development conditions (no human influence), and maximum development based on zoning information for the towns within the watershed. The current condition scenario forms the body of this report. The pre-development condition scenario assumed a uniform forest land cover for the entire watershed and no reservoirs. These data sets are provided electronically as an appendix to this report. Maximum development is called the full build-out scenario. The full build-out scenario was developed for two runoff management conditions. The first assumes the effective impervious surface is represented by the Connecticut data, as shown on Figure 11. This runoff management option assumes generally rural conditions with relatively low connectedness of impervious surfaces and minimal drainage control. The second runoff management option assumes a developed setting with stormwater collection and discharge to streams (represented by the Alley and Veenhuis, 1983 data shown on Figure 11).

The scenarios include a full build-out development plan that is based on the current zoning (NVCG, 2005), a pre-development land and water use option for comparative analysis, reservoir management option, and an extreme climate scenario designed to evaluate land and water use options under severe drought and high rainfall conditions. Table 15 (to be provided by USGS in final draft) shows the discharge statistics for the different management scenario runs. The data sets themselves are available on CD as text files in Appendix B.

The existing condition simulation and the management scenario also generated spatially and temporally distributed data sets of ground-water model of the watershed. These data were used as input to a calibrated MODFLOW ground-water model of the watershed. The MODFLOW model will be used to evaluate ground-water pumping options and ground-water management scenarios. Streamflow data sets were also used as input to MesoHABSIM instream habitat prediction model (Parasiewicz, 2005).

For comparative purposes, Figure 27 (to be provided) shows the flow duration curves for the simulations representing existing conditions (with flow adjustment and routing), pristine conditions and full build-out conditions. The spatial and temporal distribution of mean recharge and surface runoff for the full build out scenario are shown on Figures 28 and 29 (to be provided by USGS in final draft). Figure 30 shows the effect of increased development (based on the full build-out scenario assuming the runoff collection and discharge to streams) on streamflow during a typical period of time, including moderately high to low flow.

References (to be provided by USGS in final draft)
Alley and Veenhuis, 1983
Civco, 2002
Mazzaferro, 1986
NVCG, 2005
Parasiewicz, 2005
Starn, 2002