ASSESSMENT OF PCB CONTAMINATION IN THE DELAWARE RIVER ESTUARY DECADAL SCALE CONSISTENCY CHECK

The decadal scale consistency check presented in this paper was performed by HydroQual at the request of the Delaware River Basin Commission (DRBC) based upon the recommendation of DRBC's Expert Panel convened to assist DRBC evaluate the technical issues associated with the PCB TMDL. Funding for HydroQual's decadal scale consistency check was provided by the Delaware Estuary TMDL Coalition. The modeling framework used is a modified version of WASP5/TOXI5 applied to PCBs in the Delaware River Estuary to support development of a Stage 1 TMDL. The model for the TMDL required an integrated mass balance approach for hydrodynamics, sorbent dynamics and PCB transport and fate. The Stage 1 TMDL modeling work was completed by DRBC and Limno-Tech, Inc. (LTI) and is presented in Bierman et al., 2004, and in the other papers presented in the WEFTEC04 Delaware River PCB TMDL session.Historical hindcast simulations (1930-2002) were performed to check the long-term (decadal scale) behavior of the model. There are large uncertainties in the PCB forcing functions (current loads, historical loading trend) and ambient concentration data. Also, the present analysis neglects episodic events (e.g., hurricanes, 50-year flood) and long-term changes in non-PCB forcing functions (e.g., POC loads from municipal wastewater treatment plants (WWTP), nonpoint sources (NPS)) that could be important in the fate and transport of PCBs on the decadal scale. Therefore, a meaningful quantitative statistical model-data comparison can not be performed and only a qualitative appraisal is made.Based on our review of the hindcast simulation results with the current model:The model is in reasonable agreement with the historical water column concentrations, both observed and deduced from the dated core for the period following the 1980s;The model is in reasonable agreement with the contemporary sediment data in the upper estuary (Zones 2-3);The model appears to be inconsistent with the historical sediment data. The model predicts a relatively fast rate of decrease in sediment concentrations which is not seen in the data, although that comparison is limited by the high variability in the sediment data;The model predicted time course of water column and sediment bed concentrations also appear to be inconsistent with the fish tissue concentrations. The PCB concentrations in the fish have remained relatively constant over the past 10 years and increased over the past 2 years, which is not reproduced by the model. Thus, there appears to be an important inconsistency between the historical sediment and fish tissue data and the model predictions.At present it is not clear what the source(s) of the problem is but possible causes include error(s) in forcing functions (current and/or historical), the model (e.g., mixed layer depth) and/or the data or how they are interpreted. Recommendations are presented for model improvements in three areas: PCB forcing functions; the effect of episodic events and long-term changes in non- PCB forcing functions; and sediments, bioaccumulation and fish tissue concentrations. It is important to resolve these discrepancies for the next phase of the TMDL process.