On the effect of scaling conceptual model complexity on stochastic response for water quality modeling.

This paper extends previous work comparing the response of water quality models under uncertainty. A new model, the River Water Quality Model no. 1 (RWQM1), is compared to the previous work of two commonly used water quality models. Additionally, the effect of conceptual model scaling within a single modelling framework, as allowed by RWQM1, is explored under uncertainty. Model predictions are examined using against real-world data for the Potomac River with a Generalized Likelihood Uncertainty Estimation used to assess model response surfaces to uncertainty. Generally, it was found that there are tangible model characteristics that are closely tied to model complexity and thresholds for these characteristics were discussed. The novel work has yielded an illustrative example but also a conceptually scaleable water quality modelling tool, alongside defined metrics to assess when scaling is required under uncertainty. The resulting framework holds substantial, unique, promise for a new generation of modelling tools that are capable of addressing classically intractable problems.

QualViz: a tool for visual representation of water quality models.

As water quality models and their implementation have become increasingly diverse, complex and proprietary, a need for more thorough understanding of the differences between each alternative arises. The work presented here proposes a novel visualization paradigm for water quality applications which can be used to understand difference between implementations of identical and different conceptual models. A proof-of-concept visualization tool was developed and tested again three scenarios for four different conceptual models of biochemical kinetics. Results show representative figures illustrating how the approach can communicate differences in model complexity and dynamic behaviour. The proposed tool should help ensure more suitable application of water quality models in varied contexts. A discussion of quantifying model complexity in a single metric is also presented, and recommendations are made on the selection of various representational forms for communicating and exploring specific model characteristics.

An objective test of stochastic behavior in riverine water quality models.

Predictions of river water quality models are subject to substantial uncertainties, which depend not only on parameterization and calibration strategies but also on the structure of the conceptual model itself. To evaluate the importance of this effect and associated implications for stochastic models, investigations were conducted based on a segment of the Potomac River in the eastern USA. Two commonly used conceptual representations of real-world processes were used, and their simulation of DO, BOD and NH(4) components scrutinized. A GLUE approach to the inverse problem was then used to examine how uncertainty changed along the river network for each conceptual model. Differences were observed not only between deterministic instances of each conceptual model, but also between their response surfaces as a whole. Uncertainties within the river network are substantially influenced by the selection of calibration data used, as well as the primary source of the constituent examined. The suggested methodology can be used to test conceptual model validity for specific applications. The results of this study will help users select and assess models for varied problems, and refine appropriate data collection and monitoring schemes.