Hydrologic-land surface models (H-LSMs) offer a physically-based framework for representing and predicting the present and future states of the extensive high-latitude permafrost areas worldwide. Their primary challenge, however, is that soil temperature data are severely limited, and traditional model validation, based only on streamflow, can show the right fit to these data for the wrong reasons. Here, we address this challenge by (1) collecting existing data in various forms including in-situ borehole data and different large-scale permafrost maps in addition to streamflow data, (2) comprehensively evaluating the performance of an H-LSM with a wide range of possible process parametrizations and initializations, and (3) assessing possible trade-offs in model performance in concurrently representing hydrologic and permafrost dynamics, thereby pointing to the possible model deficiencies that require improvement. As a case study, we focus on the sub-arctic Liard River Basin in Canada, which typifies vast northern sporadic and discontinuous permafrost regions. Our findings reveal that different process parameterizations tend to align with different data sources or variables, which largely exhibit inconsistencies among themselves. We further observe that a model may fail to represent permafrost occurrence yet seemingly fit streamflows adequately. Nonetheless, we demonstrate that accurately representing essential permafrost dynamics, including the active soil layer and insulation effects from snow cover and soil organic matter, is crucial for developing high-fidelity models in these regions. Given the complexity of processes and the incompatibility among different data sources/variables, we conclude that employing an ensemble of carefully designed model parameterizations is essential to provide a reliable picture of the current conditions and future spatio-temporal co-evolution of hydrology and permafrost.

Permafrost thaw/degradation in the Northern Hemisphere due to global warming is projected to accelerate in coming decades. Assessment of this trend requires improved understanding of the evolution and dynamics of permafrost areas. Land surface models (LSMs) are well-suited for this due to their physical basis and large-scale applicability. However, LSM application is challenging because (a) LSMs demand extensive and accurate meteorological forcing data, which are not readily available for historic conditions and only available with significant biases for future climate, (b) LSMs possess a large number of model parameters, and (c) observations of thermal/hydraulic regimes to constrain those parameters are severely limited. This study addresses these challenges by applying the MESH-CLASS modeling framework (Modelisation Environmenntale communautaire-Surface et Hydrology embedding the Canadian Land Surface Scheme) to three regions within the Mackenzie River Basin, Canada, under various meteorological forcing data sets, using the variogram analysis of response surfaces framework for sensitivity analysis and threshold-based identifiability analysis. The study shows that the modeler may face complex trade-offs when choosing a forcing data set; for current and future scenarios, forcing data require multi-variate bias correction, and some data sets enable the representation of some aspects of permafrost dynamics, but are inadequate for others. The results identify the most influential model parameters and show that permafrost simulation is most sensitive to parameters controlling surface insulation and runoff generation. But the identifiability analysis reveals that many of the most influential parameters are unidentifiable. These conclusions can inform future efforts for data collection and model parameterization.

Permafrost thaw has been observed in recent decades in the Northern Hemisphere and is expected to accelerate with continued global warming. Predicting the future of permafrost requires proper representation of the interrelated surface/subsurface thermal and hydrologic regimes. Land surface models (LSMs) are well suited for such predictions, as they couple heat and water interactions across soil-vegetation-atmosphere interfaces and can be applied over large scales. LSMs, however, are challenged by the long-term thermal and hydraulic memories of permafrost and the paucity of historical records to represent permafrost dynamics under transient climate conditions. In this study, we aim to understand better how LSMs function under different spin-up states, which facilitates addressing the challenge of model initialization by characterizing the impact of initial climate conditions and initial soil frozen and liquid water contents on the simulation length required to reach equilibrium. Further, we quantify how the uncertainty in model initialization propagates to simulated permafrost dynamics. Modelling experiments are conducted with the Modelisation Environmentale Communautaire-Surface and Hydrology (MESH) framework and its embedded Canadian land surface scheme (CLASS). The study area is in the Liard River basin in the Northwest Territories of Canada with sporadic and discontinuous regions. Results show that uncertainty in model initialization controls various attributes of simulated permafrost, especially the active layer thickness, which could change by 0.5-1.5 m depending on the initial condition chosen. The least number of spin-up cycles is achieved with near field capacity condition, but the number of cycles varies depending on the spin-up year climate. We advise an extended spin-up of 200-1000 cycles to ensure proper model initialization under different climatic conditions and initial soil moisture contents.

Soil moisture (SM) is a key factor in the exchanging process of the hydrological cycle, which is rather difficult to be directly observed. In situ measurements on SM, however, are subjected to the point scales. Use of land surface models has been a promising way to explore SM variations, especially for poorly gauged high mountain regions such as the Yarlung Zangbo River (YZR) basin located in Southeast Tibetan Plateau. This study made an attempt to investigate the spatiotemporal variations of SM and discuss hydro-meteorological factors impacting SM evolution based on the Global Land Data Assimilation Systems (GLDAS) outputs during the period 1970-2009. Results show that (a) GLDAS data sets have high agreement and low bias with in situ measurements and consistent spatial distribution with ERA reanalysis data sets; (b) an abrupt change of SM is detected in 1992 and a significantly decreasing trend happens during 1970-2009 and 1993-2009; (c) precipitation is the dominant climatic factor controlling SM during the period 1970-2009, whereas surface air temperature is the critical factor for the significant change of SM. Owing to the significant increasing of surface air temperature since 1992, its impact on SM increased by similar to 91% than that before 1992. Evapotranspiration (ET) and snow water equivalent (SWE) are also taken into consideration, showing relatively weak influence on SM, which may be due to the dynamic process of coupled SM-ET or low snow cover area fraction across the YZR basin. Findings in this study have important implications for SM variations in poorly gauged high mountain regions which may largely influences downstream water availability.

Permafrost soils contain enormous amounts of organic carbon, which could act as a positive feedback to global climate change due to enhanced respiration rates with warming. We have used a terrestrial ecosystem model that includes permafrost carbon dynamics, inhibition of respiration in frozen soil layers, vertical mixing of soil carbon from surface to permafrost layers, and CH4 emissions from flooded areas, and which better matches new circumpolar inventories of soil carbon stocks, to explore the potential for carbon-climate feedbacks at high latitudes. Contrary to model results for the Intergovernmental Panel on Climate Change Fourth Assessment Report (IPCC AR4), when permafrost processes are included, terrestrial ecosystems north of 60 degrees N could shift from being a sink to a source of CO2 by the end of the 21st century when forced by a Special Report on Emissions Scenarios ( SRES) A2 climate change scenario. Between 1860 and 2100, the model response to combined CO2 fertilization and climate change changes from a sink of 68 Pg to a 27 + -7 Pg sink to 4 + -18 Pg source, depending on the processes and parameter values used. The integrated change in carbon due to climate change shifts from near zero, which is within the range of previous model estimates, to a climate-induced loss of carbon by ecosystems in the range of 25 + -3 to 85 + -16 Pg C, depending on processes included in the model, with a best estimate of a 62 + -7 Pg C loss. Methane emissions from high-latitude regions are calculated to increase from 34 Tg CH4/y to 41-70 TgCH(4)/y, with increases due to CO2 fertilization, permafrost thaw, and warming-induced increased CH4 flux densities partially offset by a reduction in wetland extent.