Sensitivities and uncertainties of modeled ground temperatures in mountain environments

摘要: Abstract. Model evaluation is often performed at few locations due to the lack of spatially distributed data. Since quantification model sensitivities and uncertainties can be independently from ground truth measurements, these analyses are suitable test influence environmental variability on evaluation. In this study, a physically based mountain permafrost quantified within an artificial topography. The setting consists different elevations exposures combined with six types characterized by porosity hydraulic properties. for combination all factors, that allows whole modeling domain. We found vary strongly depending input factors such as topography or soil types. analysis shows single may not representative For example, sensitivity modeled mean annual temperature albedo ranges between 0.5 4 °C elevation, aspect type. South-exposed inclined more sensitive changes in than north-exposed slopes since they receive solar radiation. increases decreasing elevation shorter duration snow cover. properties considerably types: rock clay, instance, properties, while gravel peat, accurate estimates significantly improve temperatures. discretization ground, time have impact (MAGT) cannot neglected (more 1 several parameters). show temporal resolution should least h ensure errors less 0.2 MAGT, uppermost layer most 20 mm thick. Within topographic setting, total parametric output expressed length 95% uncertainty interval Monte Carlo simulations range 1.5 clay silt, around 2.4 sand, rock. These comparable surface temperatures measured 10 m × grids Switzerland. increased peat largely their conductivity.