Overview
A major limitation in hydrodynamic-water quality modeling is the sparse availability of observed data for model inputs. This study used EFDC+ to develop a water quality model for the shallow estuaries of the Mississippi Sound and Mobile Bay, investigating four interpolation methods to augment sparse input data.
Model Setup
Four interpolation methods were evaluated: last observation carried forward (LOCF), linear interpolation (LI), natural cubic spline interpolation (Spline), and linear weighted moving average (WMA). These methods were used to construct daily water quality time series from sparse monthly data at five boundary conditions. Statistical measures of performance were used to compare interpolated inputs across methods, assess model outputs from each method, and compare modeled outputs against observed data.
Key Findings
The LOCF and Spline interpolated inputs did not perform well with increased data gaps and outliers. The LI and WMA methods produced the most similar interpolated inputs and model outputs. Among all methods, linear interpolation was the most preferred due to its low RMSE and better agreement with observed data, while Spline showed the least agreement and the highest RMSE. The results will support calibration of a hydrodynamic-water quality model for simulating water quality scenarios in the study area.