1. Introduction
The Arctic regions are experiencing unprecedented changes in climate (IPCC, 2014). Mountainous areas of the Arctic are the remotest and least studied but provide the largest share of Arctic runoff (Hinzman et al., 2005; Viviroli et al., 2011). Recent studies have shown that due to climate warming, Arctic runoff is changing too (Makarieva et al. 2019a; Rawlins et al., 2010; Stuefer et al., 2017; Tananaev et al., 2016) but the mechanisms behind these observed changes are not fully understood due to a lack of data.
This lack of data can partly be compensated by hydrological modelling. Very few process-based hydrological models have been applied in continuous permafrost mountainous river basins at daily time scale (Walvoord & Kurylyk, 2016). Among them, the Cold Regions Hydrological Modelling platform (CRHM) was applied to a basin located in the Canadian Rocky Mountains (Pomeroy et al., 2007; Fang et al., 2013; Marsh et al., 2020; Fang & Pomeroy, 2020). The VIC model, a mesoscale process-based distributed hydrological model, that represents vegetation cover, soil layers, variable infiltration, and non-linear baseflow (Endalamaw et al., 2017), was tested on the Putuligayuk (471 km2) and Kuparuk (8000 km2) watersheds in Alaska and has shown satisfactory results in terms of calculation of characteristics of snow cover, depth of thawing and runoff hydrographs (Bowling & Lettenmaier, 2010). The TopoFlow model, which is a deterministic distributed hydrological model, was also used in the simulation of hydrological processes on a small permafrost catchment of Imnavait Creek (2.2 km2) in Alaska (Schramm et al. 2007). The results of modelling of short-term reproduction of storage-related processes, such as the beaded stream system, the spatial variability of the active layer depth, and the complex soil moisture distribution, showed that the model performs adequately.
Several Russian models were applied to the mountainous Kontaktovy Creek and its tributaries at the Kolyma Water Balance Station (KWBS) located in North-Eastern Siberia (Makarieva et al., 2018). Kuchment et al. (2000) developed a physically based distributed model for several basins in the Upper Kolyma region and highlighted the importance of ground thawing depth for water losses and infiltration. Gusev & Nasonova (2008) estimated the soil and vegetation parameters for the SWAP (Soil Water–Atmosphere–Plants) land surface model, validated the model against variable states and applied it for runoff simulations on a daily time scale. The Hydrograph model was applied to simulate ground thaw/freeze processes (Lebedeva et al., 2015) and the consequent contribution to the hydrologic response at the KWBS (Semenova et al., 2013). In addition, Semenova et al. (2015 a,b) simulated the effect of pyrogenic transformation of vegetation cover on runoff formation processes and demonstrated the important role of landscapes on flow in the permafrost zone.
The physically based SWAT model was applied for runoff modeling in the Heihe River watershed, a river with a peak elevation of 5584 m in northwest China with permafrost (Li et al., 2009). The authors concluded that the consideration of the impacts of snow melt and frozen soils on the hydrological process is key to improving performance of hydrological models in mountainous areas (Zhang et al., 2016).
The study of water and heat dynamics (including the interaction between soil temperature and moisture under freeze-thaw cycles) was conducted at a monitoring site in the Tanggula Mountains, located in the permafrost region of the Qinghai-Xizang (Tibet) Plateau (QXP) in China. The results obtained using the COUPMODEL model were compared with observed ground temperature and moisture data from different depths within the active layer (Hu et al., 2015).
Models are also used for projections of future states of the hydrological system in the Arctic (Krogh & Pomeroy, 2019; Pohl et al., 2006; Rasouli et al., 2014). The problematic issue of future projections is that large-scale, relatively simple, conceptual hydrological models are calibrated against streamflow series in gauging stations of large rivers (Hudson & Thompson, 2019; Nijssen et al., 2001), while more process-based models require detailed observational information for their parametrization and are usually applied in well-studied, small research basins (Changing Cold Regions Network project, Improving Processes & Parameterization for Prediction in Cold Regions Hydrology project; Marsh et al., 2020; Zhang et al., 2008). We know that the issue of non-stationarity in hydrological response means that calibrated models often struggle to reproduce hydrological response under a significantly wetter or drier climate (Vaze et al., 2010), yet only infrequently are hydrological models tested to see if they can reproduce currently observed changes before they are applied to produce future projections. These problems undermine the ability of the hydrological community to deal with impacts of a warming Arctic on the hydrological cycle.
The aim of the research is to investigate runoff formation processes and the factors driving recent changes in hydrological response in the Suntar-Khayata ridge, the mountainous, permafrost, hard-to-reach region of Eastern Siberia. An additional research question is to determine if a hydrological model parameterized based on short-term (1957-1959) historical observations of a range of hydrological, climatological and landscape measurements can effectively reproduce long-term (1956-2012) streamflow series and recent changes.
In this paper, we will:
• Parameterize a hydrological model using compiled data and the results of previous regionalization of the model parameters.
• Verify the model based on available observations of variable states of snow and frozen ground.
• Determine if the model can reproduce observed streamflow over the period 1956-2012.
• Assess the ability of the model to reproduce recent changes of streamflow.
• Investigate the factors driving streamflow changes using the hydrological model.
The novelty of the study is the approach that allows for continuous long-term simulations of streamflow and active layer dynamics in a remote basin with complicated mountainous permafrost environment based on process understanding and scarce data of short-term observations conducted more than 60 years ago.