2/16/2012 SEWER COMMISSION Minutes

Home Browse Search

Massachusetts Estuary Project CMEP) Linked Watershed Embayment Model Peer Review The Panel notes that the SMAST RMA4 model assumes a linear relationship between reduction in nitrogen load from the watershed and net benthic nitrogen load from the sediments. This is a simplification that ignores potential"memory effects"in the sediments and it adds additional uncertainty to model results for load reduction scenarios. Key Issue 2—RMA4 Model Calibration and Validation Model calibration is the process of adjusting model parameters within physically defensible ranges until the resulting predictions give the best possible fit to the observed data. The traditional paradigm in enviromnental modeling has been calibration to one set of data and validation of the calibrated model to a second,independent set of data that was not used in the calibration. U.S.EPA(2009)recommends best practices for evaluation of environmental models to help determine when a model, despite its uncertainties, can be appropriately used to inform a decision. The proposed"tools"or practices emphasized by EPA include model corroboration, and sensitivity and uncertainty analysis. Model corroboration is the use of quantitative and qualitative methods to evaluate the degree to which a model corresponds to reality. In practical terms,it is the process of"confronting models with data." In some disciplines,this process has been referred to as validation.`The EPA guidance states that in general,the term"corroboration"is preferred because it implies a claim of usefulness and not truth. Wells (2005)has actually argued that validation is a fallacy and that it is all calibration. The characterizations of model calibration,validation and verification of the RMA4 water quality model appear to differ among the SMAST modeling applications. In Howes et al. (2001),it is stated that calibration of dispersion coefficients in the RMA4 model is typically conducted using salinity, and then verification(validation)is typically conducted using nitrogen concentrations. In the Pleasant Bay(Howes et al. 2006)and Bournes Pond(Howes et al.2005)applications,the RMA4 model is calibrated to nitrogen concentrations and.then verified(validated)to salinity. The Panel recommends that the SMAST Team adopt terminology that is consistent with the U.S.EPA (2009)guidance and avoid characterizing their models as"validated." A claim of model validation tends to confer a model with legitimacy even though the use of models to develop TMDLs involves conducting . forecast simulations for nutrient loadings and/or enviromnental conditions outside the range of those in the model calibration datasets. Environmental processes are extremely complex and inherently uncertain, and even the best and most sophisticated models are only simplistic representations. If the SMAST Team is compelled to state whether their models have been validated,the Panel recommends that they characterize validation as a process,not an end result,'and point out that model validation cannot ensure acceptable predictions(Hassan,2004). The Panel further reconunends that the SMAST Team clarify their use of salinity and nitrogen concentrations to determine dispersion coefficients in the RMA4 model. Salinity can be used to calibrate the dispersion coefficients in a water quality transport model because it is a conserved constituent. Once physical processes are calibrated,they should not be modified during the water quality calibration (Chapra 2003). However,the use of salinity to determine dispersion coefficients can become problematic in cases where high rates of tidal exchange and small freshwater inflows tend to minimize salinity gradients,especially in the open bays. One possible alternative is to use the dispersion of momentum coefficients(eddy viscosities)that can be computed in RMA2 to specify the dispersion of mass coefficients in RMA4.The Schmidt Number is defined as the ratio of the eddy viscosity coefficient to the mass dispersion coefficient. Thus,if an estimate of the Schmidt Number can be made,the mass dispersion coefficients in RMA4 can be deteriuned from the computed eddy viscosities in RMA2.As discussed by Duan(2004),various studies have been conducted to determine the Schmidt Number as it relates to vertical coefficients for eddy viscosity and mass dispersion(Rodi, 1984,Demuren and Rodi, 1986). These authors recommend a value of 0.5 in a fully 3D model. However,Ye and McCorquodale(1997)found the Sclunidt Number should December 30,2011