Geoscientific Model Development (May 2020)

Evaluating integrated surface/subsurface permafrost thermal hydrology models in ATS (v0.88) against observations from a polygonal tundra site

  • A. Jan,
  • E. T. Coon,
  • S. L. Painter

DOI
https://doi.org/10.5194/gmd-13-2259-2020
Journal volume & issue
Vol. 13
pp. 2259 – 2276

Abstract

Read online

Numerical simulations are essential tools for understanding the complex hydrologic response of Arctic regions to a warming climate. However, strong coupling among thermal and hydrological processes on the surface and in the subsurface and the significant role that subtle variations in surface topography have in regulating flow direction and surface storage lead to significant uncertainties. Careful model evaluation against field observations is thus important to build confidence. We evaluate the integrated surface/subsurface permafrost thermal hydrology models in the Advanced Terrestrial Simulator (ATS) against field observations from polygonal tundra at the Barrow Environmental Observatory. ATS couples a multiphase, 3D representation of subsurface thermal hydrology with representations of overland nonisothermal flows, snow processes, and surface energy balance. We simulated thermal hydrology of a 3D ice-wedge polygon with geometry that is abstracted but broadly consistent with the surface microtopography at our study site. The simulations were forced by meteorological data and observed water table elevations in ice-wedge polygon troughs. With limited calibration of parameters appearing in the soil evaporation model, the 3-year simulations agreed reasonably well with snow depth, summer water table elevations in the polygon center, and high-frequency soil temperature measurements at several depths in the trough, rim, and center of the polygon. Upscaled evaporation is in good agreement with flux tower observations. The simulations were found to be sensitive to parameters in the bare soil evaporation model, snowpack, and the lateral saturated hydraulic conductivity. Timing of fall freeze-up was found to be sensitive to initial snow density, illustrating the importance of including snow aging effects. The study provides new support for an emerging class of integrated surface/subsurface permafrost simulators.