Modeling challenges for predicting hydrologic response to degrading permafrost

Abstract

The fate of the approximately 1,700 billion metric tons of carbon (Tarnocai et al. 2009) currently frozen in permafrost affected regions of the Arctic and subarctic is highly uncertain (IPCC 2007), primarily because of the potential for topographic evolution and resulting drainage network reorganization as permafrost degrades and massive ground ice contained in ice-rich permafrost soils melts. Computer modeling is a key tool in untangling these complex feedbacks to understand the evolution of the Arctic and subarctic landscapes and the potential feedbacks with the global climate system. Some of the challenges associated with modeling the hydrologic system in and around degrading permafrost are discussed in this essay. Modeling requirements depend very strongly on the spatial resolution of the model. Two different classes can be identified, depending on whether microtopography is explicitly resolved or incorporated into the model through a subgrid parameterization. The focus here is on the computational challenges associated with microtopography-resolving models using hydrologic response of polygon mires as an example. In such microtopography-resolving models, horizontal grid spacing on the order of 0.25mwould typically be required. Although highand low-centered ice wedge polygons have been identified as important controls on Arctic surface hydrology (e.g. Liljedahl et al. 2012) and evolution from lowto highcentered polygon landscapes is expected as Arctic temperatures increase (Jorgenson et al. 2006), as far as we are aware, there is no existing computer code that represents the full range of processes required to model the co-evolution of surface topography, active layer, and permafrost at the microtopography-resolving scale. For microtopography-resolving models of hydrology in permafrost landscapes, it is convenient to partition the large number of coupled processes into four critical sets: subsurface thermal/hydrology, surface thermal processes, mechanical deformation, and overland flow processes. However, it is important to recognize that the partitioning is somewhat arbitrary and that multiple tightly coupled processes exist within each set.