Monitoring for St. Louis Bay Estuary and Watershed Model Calibration Project Number: MX974727-03
The development of a mathematical model of the physical, chemical and biological processes at the watershed scale requires integration of large quantities of geophysical data into a model to describe the particular study area. The model development for St. Louis Bay watershed has been revisited and refined based on the previous modeling efforts, literature review, and consultation with soil scientists and agronomists of Extension Service of MSU. The primary aspects revisited and refined included fertilization-related input parameters, plant uptake-related input parameters, nutrient input methods, non-crop land simulation using PQUAL module, and recalibration of hydrology in the Jourdan River. Some of the refined model inputs have been substantiated by the St. Louis Bay watershed soil sample data and extensive edge-of-field collected from related studies. The computed time series of water quality constituent were then set up as the boundary conditions of the bay water quality model (WASP). The primary results from this modeling research can be summarized as follows:
- The model inputs of nutrient distribution in the soil has been proved to be valid based on extensive soil sampling research in both the St. Louis Bay watershed and other watershed with similar characteristics.
- The phosphorus input form for the St. Louis Bay watershed model, was substantiated by the results from edge-of-field experiment. In addition, the validity of phosphorus mass balance applied in the model was also proved by the edge-of-field experiment.
- Whether the modeler should use AGCHEM modules to simulate the nutrient transportation in the cropland depends on the modeling purpose, data availability and watershed characteristics.
- The overall model performance is responsive to the manner in which loads are applied. Hence, more attention should be focused on the correct estimation of boundary loading forcing functions instead of iterative calibration of input parameters. The modeling performance depends on the correct characterization of types, locations, and magnitudes of the pollutants of concern. For the long period modeling, a better choice is to develop the nutrient inputs from the fertilization based on the average loadings for the simulation period instead of the most recent recommended fertilization rates.
- It is very important to make sure that what parameters can be calibrated and what parameters can not.
- For the long period modeling, if there are indications of obvious land use changes, separate modeling should be considered. It is recommended that the impacts of land use change on water quality simulation be studied.
- Contribution analysis of pollutant sources is a very effective method to help calibrate the developed model. During the calibration of the St. Louis Bay watershed model, the contributions of pollutant from background, point, and non-point sources were stepwise-added to examine the modeling performance. This will allow the modelers to compare the contributions from different sources and provide the basis for TMDL determination.
- The spatial analysis of observed water quality data can give an insight of how much sub-watersheds should be delineated.
- Four modeling scenarios were devised based on the results of sensitivity analysis to improve water quality simulation. Compared with base scenario, each of the 4 modeling scenarios show significant improvement. More extensive measured data are needed to confirm that these scenarios reflect the annual and seasonal variation of the water quality.