A STATISTICAL APPROACH TO REGIONALIZING LUMPED PARAMETER TRANSIT TIME MODEL (LPM) PARAMETERS IN THE CHESAPEAKE BAY WATERSHED

Lumped parameter transit time models (LPMs) are powerful tools for simulating the transport of contaminants through mesoscale watersheds. Depending on the modeling purpose and data availability, two approaches may be appropriate for the parameterization of LPMs. "Traditional" LPMs (T-LPMs) require the direct parameterization of the catchment transit time distribution (TDD), which captures the combined influence of watershed structure (i.e., internal variability) and climatic inputs (i.e., external variability) on catchment transport. "Storage selection” LPMs (SAS-LPMs) require the parameterization of the distribution of water ages in catchment storage that get selected into outflow; this distribution captures the influence of watershed structure but is relatively uninfluenced by climatic inputs. In either case, parameterizing LPMs requires environmental tracer data, which can be spatially and temporally sparse. Although groundwater models can be used to regionalize LPM parameters by synthesizing tracer data with other watershed measurements under a theoretical framework, such modeling is often resource prohibitive. The primary objective of this study is to develop an alternative, statistical modeling approach to regionalizing parameters of T-LPMs and SAS-LPMs for groundwater flows using watershed characteristics such as hydraulic conductivity, surficial geology, and drainage density. The statistical modeling approach will be analyzed to quantify its predictive skill and to make inferences about the relationship between statistically significant watershed characteristics and dominant transport processes. A secondary objective is to test the hypothesis that T-LPM parameters will be less efficiently regionalized with statistical appraoches due to the confounding influence of climatic inputs. Preliminary results are presented along four ongoing tracks of work. Although situated in the Chesapeake Bay Watershed, the research aims to advance a broader goal of making low-cost, robust estimates of watershed transit time distributions more available for contaminant transport modeling in data-scarce regions.