Bio-irrigation in permeable sediments: an assessment of model complexity
Meysman, F.J.R.; Galaktionov, O.S.; Gribsholt, B.; Middelburg, J.J. (2006). Bio-irrigation in permeable sediments: an assessment of model complexity. J. Mar. Res. 64(4): 589-627 In: Journal of Marine Research. Sears Foundation for Marine Research, Yale University: New Haven, Conn.. ISSN 0022-2402; e-ISSN 1543-9542, more |  |
| Authors | | Top | - Meysman, F.J.R., more
- Galaktionov, O.S.
- Gribsholt, B., more
- Middelburg, J.J., more
| | |
| Abstract | Burrowing benthic animals ventilate their burrow networks, and this enhances the transport of solutes in the sediment and exchange with the overlying water column, a process referred to as bio-irrigation. Various models have been proposed to model bio-irrigation, with different levels of sophistication related to model dimensionality and parameter numbers. Here we address the issue of model complexity for bio-irrigation in permeable sediments. To this end, we simulated flowline patterns and tracer signals using (1) a full 3D model that explicitly models the J-shaped geometry of the burrow in a suitable microenvironment surrounding the burrow, (2) a simplified 2D axisymmetric analogue, which neglects the burrow shaft and only models the location of burrow water injection, (3) a highly simplified 1D model obtained by laterally averaging the microenvironment. Simulation of two separate inert tracer experiments shows that the 2D pocket injection model includes essential features (downward advective transport, spatial heterogeneity of pore water velocity, mechanistic specification of the seepage area) that are lost upon averaging to the corresponding 1D model. This loss of model detail must be compensated for by the introduction of additional, non-mechanistic fitting parameters in the 1D description. Similarly, the extension of the 2D model to a full sophisticated 3D description requires a major increase in computational resources, but only leads to a marginal improvement in the data simulation. Accordingly, we conclude that the 2D description provides an optimal balance between model simplicity and predictive capacity. |
|
